A Review and Prospects of Manufacturing Process Knowledge Acquisition, Representation, and Application

: The evolution of knowledge acquisition and representation in manufacturing technologies is vital for translating complex manufacturing data into actionable insights and advancing a comprehensive knowledge framework. This framework is pivotal in driving innovation and efficiency in intelligent manufacturing. This review aggregates recent research on knowledge acquisition and representation within the manufacturing process, addressing existing challenges and mapping potential future developments. It includes an analysis of 123 papers that focus on harnessing advanced intelligent analytics to extract operationally relevant knowledge from the extensive datasets typical in manufacturing environments. The narrative then examines the methodologies for constructing models of knowledge in manufacturing processes and explores their applications in manufacturing principles, design, management, and decision-making. This paper highlights the limitations of current technologies and projects emerging research avenues in the acquisition and representation of process knowledge within intelligent manufacturing systems, with the objective of informing future technological breakthroughs.


Introduction
In the era of smart manufacturing and automation, the roles of knowledge acquisition, representation, and application in the manufacturing process have become increasingly prominent.Currently, converting data from manufacturing processes into structured, usable knowledge is recognized as a core challenge in the field.With the rise of the Industry 4.0 concept, related research has garnered widespread attention [1,2], impacting not only the enhancement of production efficiency and product quality but also leading to fundamental changes in global manufacturing practices [3].
Since the late 20th century, continuous breakthroughs in computer technology and data acquisition methods have significantly advanced the ways in which knowledge is acquired and represented in manufacturing [4].These transformations not only involve technological innovations from simple data collection to complex knowledge representation [5] but also include rapid advancements in machine learning and artificial intelligence, which are providing new perspectives for the extraction and application of knowledge [6].
However, despite technological advancements, the manufacturing industry still faces issues with the lack of standardized knowledge guidance in product design, processing, and assembly, which directly affects the stability of process strategies and parameter selection, as well as the final quality of manufacturing [7].In response to the accelerating pace of product updates and the increasing complexity of structures, engineers must spend considerable time integrating manuals, data, documents, and expert knowledge [8] to extract useful process knowledge from the processed data.Timely and accurate capture and representation of this knowledge are key challenges in building intelligent manufacturing process management systems [9].Through a systematic review of the relevant literature, this paper aims to showcase the latest research trends in knowledge acquisition, representation, and application in the field of smart manufacturing and address the aforementioned challenges.The goal is to provide a value framework for the integration and utilization of manufacturing knowledge by synthesizing existing research, and establishing a smart manufacturing system that can quickly respond to market and consumer demand changes while adapting to evolving market trends.
The structure of this article is organized as follows: Section 2 compiles a synthesis of the research corpus on manufacturing process knowledge, outlining the salient features of extant studies.Section 3 expounds upon the methodologies utilized in the acquisition of manufacturing process knowledge.Section 4 embarks on an exhaustive analysis of knowledge taxonomies, concurrently assessing the latest progress in defined research domains.Conclusively, Section 5 predicts the development trends and prospects of knowledge acquisition and representation in complex manufacturing processes from an intelligent manufacturing perspective, based on existing research findings.

Literature Research Methodology
This review aims to critically examine the historical progress and current status in the field of knowledge acquisition and representation in manufacturing processes since 2013.Using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) process [10], this study primarily retrieves the relevant literature from the Web of Science (WOS) core database, supplemented by EI-indexed papers in the China National Knowledge Infrastructure (CNKI) database (as shown in Figure 1).The purpose of this research is to identify the deficiencies and gaps in the existing literature and to emphasize its academic value, providing a clear framework for the academic community and industry practitioners to deepen their understanding of the value of knowledge acquisition and representation in modern manufacturing.Additionally, this review explores the challenges in key technologies and fundamental manufacturing processes, aiming to outline a direction for future research and to promote the development of knowledge management and representation strategies using emerging technologies, which is crucial for sustaining innovation and maintaining a competitive edge in dynamic markets.
to extract useful process knowledge from the processed data.Timely and accurate capture and representation of this knowledge are key challenges in building intelligent manufacturing process management systems [9].
Through a systematic review of the relevant literature, this paper aims to showcase the latest research trends in knowledge acquisition, representation, and application in the field of smart manufacturing and address the aforementioned challenges.The goal is to provide a value framework for the integration and utilization of manufacturing knowledge by synthesizing existing research, and establishing a smart manufacturing system that can quickly respond to market and consumer demand changes while adapting to evolving market trends.
The structure of this article is organized as follows: Section 2 compiles a synthesis of the research corpus on manufacturing process knowledge, outlining the salient features of extant studies.Section 3 expounds upon the methodologies utilized in the acquisition of manufacturing process knowledge.Section 4 embarks on an exhaustive analysis of knowledge taxonomies, concurrently assessing the latest progress in defined research domains.Conclusively, Section 5 predicts the development trends and prospects of knowledge acquisition and representation in complex manufacturing processes from an intelligent manufacturing perspective, based on existing research findings.

Literature Research Methodology
This review aims to critically examine the historical progress and current status in the field of knowledge acquisition and representation in manufacturing processes since 2013.Using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) process [10], this study primarily retrieves the relevant literature from the Web of Science (WOS) core database, supplemented by EI-indexed papers in the China National Knowledge Infrastructure (CNKI) database (as shown in Figure 1).The purpose of this research is to identify the deficiencies and gaps in the existing literature and to emphasize its academic value, providing a clear framework for the academic community and industry practitioners to deepen their understanding of the value of knowledge acquisition and representation in modern manufacturing.Additionally, this review explores the challenges in key technologies and fundamental manufacturing processes, aiming to outline a direction for future research and to promote the development of knowledge management and representation strategies using emerging technologies, which is crucial for sustaining innovation and maintaining a competitive edge in dynamic markets.

The Characteristics of Manufacturing Process Knowledge Acquisition and Representation
In the specialized field of manufacturing process knowledge acquisition and representation, the core objective is to precisely identify and determine key parameters and operational steps through an in-depth analysis of complex manufacturing processes, thereby effectively extracting and presenting information that is both easy to understand and applicable in practical scenarios.Referring to Figure 1, it can be observed that during the research on manufacturing process knowledge acquisition and representation, several salient features typically emerge: (1) Domain Specificity: In manufacturing process engineering, capturing and representing context-specific knowledge is essential.Each process has unique terminologies and logical frameworks, influencing mechanical precision and material understanding.It is crucial to meticulously record process parameters, material interactions, and Subsequently, the framework classifies and represents manufacturing process knowledge based on process basic knowledge, process design knowledge, process management knowledge, process decision-making knowledge, and other related knowledge.Ultimately, on the basis of this knowledge acquisition and representation, the framework meets diverse application requirements such as improving remanufacturing process efficiency, fault diagnosis, process improvement, and process quality control.

The Characteristics of Manufacturing Process Knowledge Acquisition and Representation
In the specialized field of manufacturing process knowledge acquisition and representation, the core objective is to precisely identify and determine key parameters and operational steps through an in-depth analysis of complex manufacturing processes, thereby effectively extracting and presenting information that is both easy to understand and applicable in practical scenarios.Referring to Figure 1, it can be observed that during the research on manufacturing process knowledge acquisition and representation, several salient features typically emerge: (1) Domain Specificity: In manufacturing process engineering, capturing and representing context-specific knowledge is essential.Each process has unique terminologies and logical frameworks, influencing mechanical precision and material understanding.It is crucial to meticulously record process parameters, material interactions, and machine performance to ensure the knowledge representation system accurately reflects and adapts to operational variables in practical applications.Transforming tacit expertise into explicit, actionable directives requires technical precision, versatility, and the system's ability to evolve alongside advancing process technologies [11].(2) Technical and Complex: Manufacturing processes are intricate due to their diverse procedural stages and the extensive range of data and parameters involved [12].From selecting raw materials to final product inspections, each stage integrates complex layers of information, including detailed process parameters and precise control strategies.This complexity demands high accuracy and encompasses variables across multiple disciplines such as material science, mechanical engineering, thermodynamics, electronics, and computer science, creating an interdisciplinary framework.
Effective management of this framework requires a deep understanding of each discipline and the ability to extract essential decision-support information from the vast array of variables, relying on analytical skills to navigate the intricate interrelations within this multidisciplinary environment.(3) Dynamic Updatability: In the complex manufacturing process, dynamic updating is the core element to maintain the competitiveness of manufacturing systems and technologies.Facing the continuous emergence of new technologies and ever-changing technical requirements, manufacturing process systems must demonstrate a high degree of adaptability to maintain their operational efficiency and effectiveness.Dynamic updating goes beyond the simple maintenance of existing knowledge; it represents an active and proactive learning and adaptation mechanism that requires systems to absorb the latest information, reassess existing functions and parameters, and make necessary adjustments accordingly.This dynamic updating ensures that manufacturing processes can flexibly respond to technological innovations and changes in market demand, optimize production processes, enhance product quality, and ultimately strengthen the market competitiveness of enterprises through continuous knowledge acquisition and application [13,14].(4) Multi-faceted Knowledge Sources: Manufacturing knowledge arises from varied sources including design specifications, operational details, engineering expertise, machine data, and quality feedback.Integrating these complementary sources-from theoretical to practical, quantitative to qualitative-requires a systematic approach and advanced data management systems to consolidate information and enable its seamless utilization throughout the manufacturing process [15].(5) Process Knowledge Tacit: Transforming tacit knowledge into explicit forms involves systematically capturing and codifying workers' experiences, judgments, and lessons into documents, manuals, or databases.This process typically utilizes knowledge extraction interviews, collaborative sessions, and specialized tools.Such efforts not only preserve and disseminate knowledge but also lay a foundation for continuous innovation and improvement [16].(6) Hierarchization of Knowledge: Structuring knowledge hierarchically creates a layered framework from operational techniques to macro-management strategies, each layer serving specific functions in decision support systems.This structure ensures rapid access to relevant information, enhancing decision-making quality and efficiency.By systematically implementing this hierarchy, knowledge is preserved, disseminated, and applied methodically across various decision points, improving coordination and efficiency in the production system, and centralizing knowledge within an efficient manufacturing ecosystem [17].

Manufacturing Process Knowledge Acquisition
The meticulous pre-processing of manufacturing process features is a pivotal initial step that underpins the integrity of data quality and the robustness of ensuing analytical procedures.This foundational stage encompasses a suite of critical tasks.Paramount among these is data cleansing, a process dedicated to the excision of records that are invalid, incomplete, or corrupted by noise.Equally critical is the standardization of data formats-an endeavor that ensures seamless integration across heterogeneous systems and devices.Moreover, normalizing data ranges is essential to attenuating the influence of disparate magnitudes and ratios on the analytical results.Additionally, the strategic application of feature selection and dimensionality reduction techniques plays a crucial role.These methods are instrumental in pinpointing and preserving feature variables that exert considerable influence on the manufacturing process.Concurrently, these techniques deftly streamline the dataset by excluding superfluous and irrelevant information, thus enhancing computational efficiency.The judicious employment of these pre-processing strategies markedly augments the velocity and precision of manufacturing process analysis.This diligent preparation lays a formidable data foundation, one that is instrumental in fortifying subsequent process control and fostering continuous quality enhancement.Such a foundation is indispensable for manufacturing systems aspiring to achieve and maintain operational excellence [18] (the length of the article constrains me from expanding on this here).
The acquisition of knowledge in manufacturing processes involves the systematic identification, collection, and analysis of key attributes stemming from manufacturing activities.These attributes span a diverse set of parameters, such as operating conditions of machinery, properties of raw materials, environmental factors, operational behaviors, and control settings.The goal is to distill valuable insights and information from these parameters [19].For example, Borkar et al. [20] developed an automated system for extracting 3D part features specifically designed for computer-integrated manufacturing settings.This system is capable of independently identifying machining features in a uniform way while utilizing product data exchange standards to infer these features.As a result, it enables the automatic delivery of crucial information needed for the manufacturing process.Meanwhile, Wang et al. [21] proposed a knowledge-driven method for reconstructing multiview bill of materials (BOM) for complex products.By studying the construction of a manufacturing process knowledge base, this method supports BOM modeling and simulation analysis in digital twin workshops, with the goal of reducing cycle time, improving efficiency, and enhancing quality.Wang et al. [22] advanced the field by introducing a feature recognition methodology that addressed the challenges of identifying interactive features in computer-aided designs for cast and rolled parts.This methodology employed attribute adjacency graphs and feature trees, resulting in a substantial improvement in the efficiency of producing complex parts.
These processes are underpinned by advanced data analysis techniques, including data mining, feature recognition, and numerical modeling.These techniques serve to transform raw data into actionable knowledge, thereby bolstering the preprocessing and responsiveness of knowledge acquisition in the production process.

Manufacturing Process Knowledge Acquisition Based on Statistical Analysis and Mathematical Modeling
In the domain of manufacturing, the application of statistical analysis and mathematical modeling for knowledge acquisition elucidates the production process's intrinsic patterns.The aggregation and scrutiny of extensive production data enable the identification of pivotal process variables, the prediction of potential quality issues, the mitigation of superfluous cost burdens, and the provision of a data-driven foundation for unrelenting process refinement.Such data-guided lean manufacturing upholds process stability.In essence, statistical and mathematical methodologies furnish a rigorous and methodical framework for the profound assimilation and deployment of manufacturing knowledge, underscoring data's imperative role in facilitating production efficiency and innovation.Zhang et al. [23] devised a paradigm for multi-source knowledge acquisition, analyzing the design process and various knowledge sources.They developed a heuristic design methodology predicated on the machining process innovation knowledge (MPIK) network to systematically address the intricacies associated with the production of miniaturized components and foster process innovation.Concurrently, Martin et al. [24] refined a product model through a data-centric approach, achieving an accurate depiction of product attributes.The model's efficacy was corroborated via a case study involving simulations and tolerance analyses of automotive brake systems.Scheibel et al. [25] introduced the DigiEDraw system, which leverages a 2D clustering algorithm to automate the extraction of dimensional data from engineering schematics, effectively integrating this data into the automated manufacturing workflow to augment quality control efficacy, achieving a recall rate in excess of 88%.Moreover, Guo et al. [26] posited an innovative strategy that synthesizes graph theory and rule sets, utilizing B-Rep and weighted attribute adjacency matrices for structural representation, with categorical features discerned via inverse modeling-a methodology validated within the context of axle components.Zhou [27] innovated a technique combining sequence comparison and grain calculation for standard process route identification.This approach commences with the construction of a fuzzy similarity matrix through sequence comparison, followed by the application of fuzzy quotient space theory to formulate a multi-level process structure.Cluster analysis is then employed to refine the process hierarchy, culminating in the delineation of standardized process pathways.Chen et al. [28] concentrated on augmenting the reusability of assembly process information, endeavoring to standardize the extraction of assembly case data from multimedia sources via an assembly process case (APC) integration framework.This framework marries hierarchical topologies with semantic constructs to streamline information retrieval.Lastly, Sivakumar et al. [29] introduced an automated technique for CAD/CAM/CAI data transfer, focusing on extracting features from STEP files to facilitate system integration, computer numerical control (CNC) code generation, and product quality assurance.

Manufacturing Process Knowledge Acquisition Based on Natural Language Processing Techniques
The integration of natural language processing (NLP) techniques into the acquisition of manufacturing process knowledge represents a burgeoning domain that capitalizes on advanced linguistic analysis tools.These tools adeptly interpret unstructured textual data prevalent in manufacturing settings, such as operation manuals, quality reports, equipment maintenance logs, and workforce feedback.Leveraging NLP technologies enables the automated distillation of pivotal information from these textual sources, converting it into actionable knowledge and insights to augment production processes.Notably, parametric analysis control utilizing NLP can bolster process stability, while subject identification can unveil recurrent patterns of malfunctions and defects.Kang et al. [30] demonstrated the potential of NLP technologies in transmuting unstructured manufacturing knowledge into structured rules, thereby facilitating the identification of essential process knowledge.Navinchandran et al. [31] applied NLP methodologies to extract critical terms from maintenance documentation and scrutinized the correlation between these terms and key performance indicators (KPIs), pinpointing factors that significantly influence equipment maintenance efficacy, such as time and cost.Further, Yang et al. [32] proposed an NLPbased method for the automatic extraction of system engineering ontologies from textual documents to foster standardization and interoperability.Bhardwaj et al. [33] developed an automated analytical tool that seamlessly combines syntactic parsing with semantic understanding.This tool is designed for the comprehensive examination of industrial equipment maintenance records, aiming to detect critical defective components effectively and augment design and maintenance processes.Yin et al. [34] proposed an industrial named entity extraction method (INERM) for process industries, alleviating the problem that existing entity extraction methods cannot reasonably utilize the semantic information and context of words.This suite of NLP applications in manufacturing underscores the transformative impact of linguistic technology in deciphering and harnessing the wealth of information contained within industry-related text.

Machine Learning and Deep Learning Based Knowledge Acquisition for Manufacturing Processes
Acquisition of process knowledge through deep learning constitutes a cornerstone in the Industry 4.0 epoch, employing sophisticated neural networks to analyze and decipher data within manufacturing contexts.Deep learning frameworks, notably convolutional neural networks (CNNs) and recurrent neural networks (RNNs), possess the capability to unearth patterns and trends from extensive production data sets, discerning even those subtle interrelations that may elude human expertise.The potential applications of these models are vast, encompassing process quality inspection, predictive maintenance, enhancement of process efficiency, and the diagnostic analysis of equipment failures.By monitoring image, acoustic, and vibration data in real-time on production lines, deep learning algorithms can presciently forecast the degradation of equipment performance and impending failures, thus facilitating proactive scheduling of maintenance and operational manufacturing procedures.Kumar et al. [35] introduced a supervised learning construct, merging bidirectional long short-term memory networks (Bi-LSTMs) with conditional random fields (CRFs), to categorize and annotate unstructured texts in manufacturing sciences, demonstrating an accuracy rate of 88%.Bhardwaj et al. [36] developed an unsupervised lexicon-based machine learning algorithm, proven to autonomously pinpoint the operational status of industrial machinery from unstructured texts, a claim substantiated by analyses of maintenance records from oil drilling equipment.Zhang et al. [37] leveraged deep 3D convolutional neural networks to automate the identification of machining features within CAD models, demonstrating both efficiency and accuracy in processing large data sets and intricate geometries, thereby surpassing preceding technologies.Additionally, Zhang et al. [38] introduced a deep reinforcement learning framework designed to optimize process paths in parts manufacturing automatically.This framework, which integrates directed graphs and graph neural networks, has been validated against real-world scenarios, enhancing both the efficiency and quality of aircraft part production.These innovations highlight the profound impact of deep learning on manufacturing process knowledge acquisition, paving the way for advanced smart manufacturing practices.

Knowledge Acquisition for Manufacturing Processes Based on Semantic Web and Knowledge Graphs
In manufacturing process knowledge acquisition, semantic networks and knowledge graphs are central to a robust knowledge management framework.The semantic web underpins this by formalizing data relationships and enhancing machine interpretability and data utility.Knowledge graphs extend this framework by encapsulating entities, concepts, and their interconnections within the manufacturing domain, thereby creating an intricate, multi-layered knowledge network that clarifies embedded expertise and fosters information interconnectivity.
This integration of semantic networks and knowledge graphs enables efficient extraction of insights from distributed data, advancing automated knowledge acquisition and application, and optimizing knowledge use in manufacturing.
Han et al. [39] developed an efficient assembly retrieval method based on semantic information and web ontology, utilizing weighted bipartite graphs for enhanced manyto-many matching.Zhang et al. [40] devised a method to construct a metallic materials knowledge graph (MMKG) using semantic algorithms and ontology, leveraging DBpedia and Wikipedia to demonstrate excellence in performance.Wen et al. [41] introduced a process knowledge graph construction method, applying ontology and a pattern-guided bootstrapping framework with a two-tiered filtering mechanism to prevent overfitting.

Manufacturing Process Knowledge Acquisition Based on Hybrid Methods
In manufacturing process knowledge acquisition, hybrid methodologies integrate multiple technical tools such as NLP, machine learning, deep learning, knowledge graphs, and statistical analysis for a holistic approach.This strategy effectively analyzes various data types, utilizing NLP for text extraction, machine learning for identifying patterns in sensor and log data, and knowledge graphs for mapping complex relationships between entities.This enhances both the interpretability of data and the depth of knowledge discovery.
Wang et al. [42] presented a process innovation framework leveraging collective intelligence, multi-source knowledge, and semantic technologies, fortified by trust mechanisms for maintained information integrity.Kang et al. [43,44] improved rule extraction from manufacturing texts by fusing constraint modeling, NLP, and ontological knowledge.Zhou et al. [45] advanced assembly planning with a knowledge graph and graph convolutional neural network, validated on aero-engine rotor assemblies.Madhusudanan et al. [46] utilized NLP and sentiment analysis for key information extraction and knowledge base construction as a decision-support strategy for smart manufacturing.Hu et al. [47] integrated case-based reasoning with enhanced weighted latent Dirichlet allocation (LDA) techniques for assembly instruction extraction.Guo et al. [48] developed a framework to build an intelligent knowledge base for machining processes.This framework integrates ontology rule mapping with the BERT-Improved TRANSFORMER-CRF (BITC) model and a word vector cosine similarity algorithm, facilitating the efficient capture and integration of machining knowledge.Jiang et al. [49] applied an automatic update method for remanufacturing disassembly process knowledge graphs using an ALBERT-BiLSTM-CRF model combined with NLP.Li et al. [50] integrated manufacturing information via kernel principal component analysis and evolutionary cellular automata for intelligent clustering.Zhang et al. [51] proposed a process route extraction method combining similarity measures with particle swarm optimization.Jeon et al. [52] bridged semantic gaps between design documents and CAD models using domain ontology and shallow NLP.Arista et al. [53] employed an ontology-based approach for integrated enterprise knowledge in aerospace product design optimization.

Research Status of Manufacturing Process Knowledge Representation
In alignment with national standard GB/T39469-2020, which delineates classification and codification protocols for general manufacturing process knowledge [54], we propose a structured taxonomy that encapsulates the multifaceted nature of manufacturing processes.This refined classification model is composed of five principal categories: foundational process knowledge, process design knowledge, process management knowledge, decision-making knowledge, and a compendium of hybrid knowledge.The objective of this framework is to facilitate a profound and methodical understanding of the manufacturing milieu, as well as the core tenets that orchestrate its dynamics.
The suggested taxonomy unfolds as follows: Fundamental process knowledge underpins the interpretation and execution of manufacturing operations.Process design knowledge translates theoretical concepts into tangible manufacturing applications.Process management knowledge underscores the optimization of operational efficacy and the enforcement of quality assurance.Decision-making knowledge in process selection harnesses data analytics to refine decision-making processes; and hybrid knowledge constitutes an integration of multifaceted process-related intelligence, critical for catalyzing technological innovation and sustainable advancement.
The adoption of this systematic and hierarchical knowledge structure promises not only to consolidate and amplify the expertise of practitioners but also to encourage crossdisciplinary collaboration, thereby nurturing a dynamic culture of perpetual innovation within the manufacturing sector.Complementing our discourse, Figure 3 illustrates the conceptual framework that embodies the essence of manufacturing process knowledge as per the established criteria.
conceptual framework that embodies the essence of manufacturing process knowledge as per the established criteria.

Process Basic Knowledge
The research content as shown in Table 1 can be divided into the following three parts:

Process Basic Knowledge
The research content as shown in Table 1 can be divided into the following three parts: Process base ontology model: Zangeneh et al. [55] proposed a unified ontology model to strengthen the representation and analysis of industrial project knowledge with logical reasoning and semantic techniques.Saha et al. [56] constructed a core domain ontology for joining processes (CDOJP) that utilizes web ontology language (OWL) to address semantic inconsistencies present in the welding process.Liang et al. [57] proposed a new method of multi-entity knowledge joint extraction (MEKJE) for faulty communication devices in Industrial IoT.The method constructs a multi-task tightly coupled model for fault entity and fault relationship extraction.Zhang et al. [58] employed a knowledge hypergraph to manage intricate multivariate relationships, devising a hypergraph embedding approach for representing such relationships within traveling fault knowledge.This method builds an ontological framework for multivariate couplings using traveling fault data, harnessing the bidirectional encoder representation from transformers (BERT) model and hypergraph convolutional networks to generate knowledge embedding vectors that facilitate the identification of analogous faults.Jin et al. [59] introduced an ontology-based semantic modeling approach for knowledge representation and intelligent decision support for roof collapse accidents in coal mines.
Design feature recognition and transformation: Guo et al. [26] introduced a novel hybrid 3D feature recognition method that automates the identification of machining features on shaft parts by integrating graphs and rules, thereby enhancing the method's versatility and efficiency.Manafi et al. [60] addressed the process of converting design details into manufacturing specifications, which involved tool orientation and dimensional tolerances and conducted empirical validation of the proposed method.Sanderson et al. [61] developed an expanded function-behavior-structure (FBS) framework, specifically designed for evolvable assembly systems (EAS), to optimize manufacturing processes oriented by product functionality.Zhang et al. [62] proposed a bio-inspired adaptive growth method employing semi-supervised learning with tree-hereditary axioms and constraints to emulate biological growth patterns, aiming to improve the accuracy of design object extraction from engineering documents.
Fundamental knowledge capture and optimization: Qin et al. [63] proposed the RFBSE model (reuse of functional block structure elicitation), which is designed to capture and reuse engineering knowledge and experience throughout the design process.Xu et al. [64] introduced a process route clustering algorithm that blends pseudo-longest metric sequences with Jaccard similarity, incorporating operation priority and similarity factors.The method's effectiveness was demonstrated experimentally by implementing numerical and size constraints and optimizing the K-medoids algorithm.
Process basic knowledge encompasses a range of critical domains, including the properties of materials, the functions of mechanical equipment, parameters of processing procedures, as well as standards for product quality.Currently, this knowledge is meticulously organized and preserved within comprehensive databases and knowledge bases, where the application of data models and ontologies facilitates the structured representation of knowledge.With the ongoing evolution of IoT technology, the collection of real-time data and the implementation of feedback mechanisms have become increasingly important.This necessitates the rapid transmission and integration of relevant information into the process knowledge base, thereby promoting the dynamic updating and continuous optimization of processes.

Process Design Knowledge
The content of this part, as shown in Table 2, can be divided into the following four components:    Extraction of typical process route

Build different granularity information
Discovering and obtaining valuable process knowledge from existing process data, obtaining typical process routes [92] Design process ontology knowledge database: Wu et al. [65] devised an ontological database for assembly process knowledge, enabling both standardization and systematization.Zhong et al. [66] offered an ontologically based assembly sequence algorithm with integrated inference rules to streamline knowledge application.Šormaz et al. [67] introduced the semantic integrated manufacturing planning model (SIMPM) ontology to articulate manufacturing constraints, utilizing axioms for problem-solving in process design.Qiao et al. [68] applied ontology web language description logics (OWL-DL) and semantic web rule language (SWRL) for processing assembly-related geometric and associative data.Guo et al. [69] conceptualized process design knowledge via a "function + flow + case" ontology, abstracting workflows and resources.Lastly, Qiao et al. [70] developed a geometrically focused ontology for assembly, enhancing the semantic description of product geometry within assembly processes.
CNC process planning and design features: Han et al.
[71] introduced a synergistic data-driven and knowledge-guided methodology for CNC process planning, employing structured models and advanced grammatical constructs to iteratively optimize NC machining solutions.Ma et al. [72] automated process and resource selection for complex rotational part machining, leveraging multi-level feature recognition and ontology knowledge bases to enhance efficiency.Eum et al. [73] proposed a process planning strategy focusing on ontological knowledge modeling to enhance the flexibility and adaptability of operation selection in multi-axis machining.Wan et al. [74] developed an inverse modeling technique for 3D machining processes, streamlining knowledge extraction and reuse.Shao et al. [75] proposed a machining metabody-based framework to facilitate the automatic identification and localization of manufacturing features in CAD models for process model generation.Wang et al. [76] utilized a graphical convolutional neural network to efficiently plan machining features, achieving a high predictive accuracy of 93.31% for machining routes.
Smart manufacturing and knowledge graphs: Shafiq et al. [77] presented an efficient approach for operational and resource information exchange in distributed manufacturing, validated through virtual engineering and bio-heuristics.Jing et al. [78] optimized complex part machining routes using hybrid reasoning and genetic algorithms, with demonstrated success in marine diesel engine component manufacturing.Li et al. [79] enhanced CAM model unity and efficiency with a multi-level knowledge mapping and inference system.Chhim et al. [80] introduced an industry standard ontology framework that facilitates knowledge reuse in product design and manufacturing.Bharadwaj et al. [81] proposed a structured knowledge mapping method for extracting assembly and similarity data from 3D models.
Automation and optimization approach: Zheng et al. [82] developed an automated robot programming system enhanced by knowledge-driven methodologies for efficient code generation.Qian et al. [83] fused knowledge bases with optimization techniques to streamline assembly scheduling.Yang et al. [84] utilized CAD data to inform ontology models for intelligent assembly planning.Ye et al. [85] advanced CNC process planning through a cloud-based knowledge system with proven effectiveness.He et al. [86] employed ontology methods for cost-effective remanufacturing process modeling.Liu et al. [87] crafted a digital twin process model focused on process knowledge and feature evolution.Liu et al. [88] leveraged digital twin technologies for autonomous robot assembly task planning, using data-driven and knowledge-based reasoning.Ling et al. [89] applied knowledge graphs to construct geometric semantic models for assembly, demonstrating utility in wind turbine assembly.
Decision support and design process: Li et al. [90] introduced a method to capture and deliver machining process knowledge based on evolving part geometry, using complex network theory to push relevant knowledge, with effectiveness demonstrated through examples.Wang et al. [91] created the "PEI-X diagram", a decision-centric design process visualization tool, proven to aid in understanding and simulating decision-making in design, validated in hot rolling system design.Zhou et al. [92] employed granular computing and bioinformatics to layer information from process analysis, extracting valuable insights for process optimization.
In the realm of manufacturing design, the representation of process knowledge is multifaceted and intricate, encompassing product design, process planning, and the integration of simulation and modeling data.Advanced CAD and CAE tools enable the structured representation of design knowledge, laying a solid foundation for design precision.The incorporation of AI technologies, such as machine learning and deep learning, further enhances the representation and application of design knowledge, facilitating virtual validation of complex designs and expediting the iterative design process.Consequently, the representation of process knowledge in the modern manufacturing design phase is a hightech, efficient, and systematic academic field, involving design knowledge management, optimization of decision support systems, and comprehensive performance analysis of the product development process.

Process Management Knowledge
Based on the general summary of Table 3, it can be summarized in the following four sections.Integration of terminology components for production ontologies

Heterogeneous data fusion (Semi) automatic integration in ontology language
Smart factory [114] Mechanical processing and CAD model management: Liu et al. [93] implemented digital twin technology for real-time evaluation of mechanical processes, using data modeling to develop a digital twin for marine diesel engine components, demonstrating instant data updates and method efficacy.Guo et al. [94] developed a knowledge management framework for processes, automating the creation of a knowledge base via graphs, and streamlining knowledge operations.Liu et al. [95] utilized hierarchical data structures for part evolution, creating descriptors for machining feature lookup and mismatch resolution.Liu et al. [96] introduced a method for rapid process information reuse, correlating information to features through a three-tier model.Zhou et al. [97] applied case-based reasoning to remanufacturing, enhancing solution retrieval and precision with nearest neighbor matching.Zhou et al. [98] employed the BERT model to construct a knowledge graph for the injection molding domain, facilitating knowledge input and expansion via a web platform.Collectively, these methodologies propel advancements in intelligent process planning and knowledge management for machining and CAD models.
Health maintenance and fault mode management: Wang et al. [99] crafted a four-level ontology for maintenance engineering, optimizing declarative and procedural knowledge management to improve complex product maintenance efficiency.Khosravani et al. [100] developed a case-based intelligent fault detection system for injection molding dropper production that reduces downtime and increases productivity by matching failure cases.Li et al. [101] constructed a reliability analysis model based on a non-uniform Poisson process by using a fishbone diagram and the 5M1E method, which provided an analysis basis for early troubleshooting and reliability improvement of machine tools.Zhou et al. [102] developed the configurable method for fault diagnosis knowledge of machine tools (CMFDK-MT) for machine tool fault diagnosis and the knowledge-based configurable fault diagnosis platform for machine tools (KCFDP-MT) framework, integrating diagnostic technologies and validating its effectiveness on a CNC gear hobbing prototype.Wang et al. [103] proposed a framework for acquiring, representing, and reasoning gas turbine health problem knowledge, and combined with ontology modeling and reasoning, aimed at improving gas turbine maintenance and diagnosis.Collectively, these pioneering endeavors represent noteworthy strides in the domains of health maintenance and fault mode management, showcasing innovative approaches and methodologies for the advancement of knowledge and practices in these critical areas.
Assembly processes and total lifecycle management: Wu et al. [104] innovated a knowledge representation framework, drawing on genetic structures to intricately capture and organize complex data in remanufacturing.Wan et al. [105] conceived a sophisticated prototype apparatus to facilitate synergistic tool machine maintenance scheduling, adeptly orchestrating the manufacturing process's dynamic, unstructured knowledge corpus.Liu et al. [106] introduced an ontology-driven strategy, utilizing OWL coding to craft a knowledge base for the full lifecycle management of electromechanical product assembly, enhancing process knowledge through semantic and logical analysis.These advancements collectively streamline assembly process knowledge handling and bolster support across product lifecycles.
Process and production ontology management: Kestel et al. [107] devised an ontologybased knowledge management tool for simulation engineering, automating knowledge capture and enhancing decision-making, leading to improved simulation precision and reduced design errors.Huang et al. [108] created a data-centric framework integrating shop-floor monitoring with enterprise verification, establishing a knowledge management system that aligns manufacturing information, thus unifying knowledge management.Zhang et al. [109] leveraged semantic web technologies to refine the unit manufacturing process for mechanical products, incorporating SWRL for knowledge reasoning and OWL for retrieval, boosting knowledge efficiency.Mou et al. [110] proposed a fuzzy logic approach for process planning, utilizing historical data analysis for informed decision-making.Mandolini et al. [111] developed a formal framework to manage manufacturing and cost knowledge, streamlining production cost estimation and knowledge asset utilization.Zhai et al. [112] enhanced the dual-layer knowledge model with Sentence-BERT, significantly increasing the efficiency of aerospace product assembly design.Adamczyk et al. [113] investigated knowledge-based systems in smart manufacturing, aiming to improve semantic interoperability and data exchange.Ocker et al. [114] presented a framework for the (semi-)automatic integration of production ontologies, ensuring data fusion and consistency within smart factories.Together, these initiatives advance process and production ontology management, underpinning informed decision-making in complex production settings.
Manufacturing management knowledge encompasses production planning, information flow, product lifecycle process, and quality control and maintenance.These elements are typically managed via enterprise resource planning (ERP) systems, with advanced AI algorithms increasingly applied to enhance complex manufacturing process information management.Enhancing the flexibility and scalability of knowledge representation is a critical research focus, aiming to swiftly adapt to market and technological changes and manage growing data complexity.Integrating advanced computational models and algo-rithms, such as machine learning and natural language processing, into ERP systems is pivotal for improving manufacturing management efficiency and effectiveness, driving the industry towards greater intelligence and efficiency.

Process Decision Knowledge
According to the main research methods and application fields of the literature, and as shown in Table 4, it is proposed that the literature can be divided into the following four parts.Laboratory methods for decision testing and evaluation

Analyze the correlation among indicators and assign weights reasonably
Green manufacturing process decision [136] Dual model process knowledge structure Enhance the capability of intelligent process design system

Combine multiple forms of knowledge and information
Complex process parameter decision [137] Decision-making systems based on knowledge graphs and ontology: The referenced literature emphasizes the application of knowledge graphs, ontologies, and case-based reasoning to enhance decision-making in manufacturing.These methodologies excel in intricate decision-making scenarios.Guo et al. [115] introduced a knowledge graph detailing components, operations, and assets, devised an inferencing structure, and assimilated semantic analysis and attribute prioritization algorithms to manage decision-making knowledge diversity.Mabkhot et al. [116] designed a decision support system that blends case-based reasoning with ontology to meet specialized product needs in Industry 4.0.The system enhances decision-making by automating the correlation of product attributes, materials, and processes, demonstrating the value of ontologies.Shen et al. [117] crafted an ontology-based knowledge modeling framework for custom clothing manufacture, integrating process, resource, and feature aspects, and introduced a bidirectional knowledge mapping approach for instant knowledge representation and integration.Collectively, these studies highlight the crucial role of knowledge graphs and ontology-informed decision-making frameworks in navigating the complexities of sophisticated manufacturing landscapes, underlining their critical influence in realizing tailored product offerings and augmenting decision-making diversity.
Deep learning-and machine learning-based decision-making methods: This category harnesses deep and machine learning to enhance decision-making in manufacturing, aiming to boost productivity and quality.Zhang et al. [118] developed a machine learningbased system for optimizing assembly planning and management.Huang et al. [119] utilized tool position analysis with deep learning for process intention models.Zhang et al. [120] employed knowledge mapping and deep learning for decision support in machining quality and cost.Zhang et al. [121] created a deep learning-driven method for automating machining route design.Wu et al. [122] applied deep reinforcement learning to process planning, improving strategies with Monte Carlo and deep learning techniques, outperforming traditional methods.These studies attest to the advanced capabilities and effectiveness of deep and machine learning in process decision-making optimization.
Automated methods based on process evaluation: This literature segment explores the automation of process evaluation to streamline the planning and assessment of complex manufacturing processes, enhancing efficiency and precision.Liu et al. [123] suggested an automated feature-matching strategy for process efficiency using 3D designs and hierarchical data.Nonaka et al. [124] presented an automated process planning method to maximize productivity from part models and resources.Jahr et al. [125] developed a semi-automatic system for equipment planning based on building information modeling.Li et al. [126] applied case-based reasoning for grinding process optimization.Single et al. [127] introduced an automated knowledge framework for HAZOP worksheet creation.Li et al. [128] offered a process planning method for multifactorial decision-making.Xu et al. [129] employed fuzzy image sets for managing evaluation discrepancies.Li et al. [130] enhanced remanufacturing decisions with a Takagi-Sugeno fuzzy neural network.Li et al. [131] devised an intelligent planning framework for aerospace components, integrating data, knowledge, and quality prediction.Liu et al. [132] utilized digital twin technology for dynamic process assessment.Huang et al. [133] proposed a blockchain-based automated decision method for optimizing process solutions.These studies advance process evaluation automation by utilizing new tech and approaches, thus improving manufacturing efficiency and decision quality.
Decision-making methods based on environmentally friendly and green manufacturing: The research underscores green manufacturing and eco-conscious decision-making, aiming to minimize environmental impact by optimizing resource utilization and reducing pollution.Lv et al. [134] developed a multi-criteria decision-making model that harmonizes the interplay between productivity, quality, resource utilization, and emissions reduction.Zhang et al. [135] created a low-carbon remanufacturing model that optimizes emissions, time, and cost using a hybrid algorithm.Wu et al. [136] employed the decision-making trial and evaluation laboratory (DEMATEL) method to assess and interlink green manufacturing indicators.Wang et al. [137] developed a dual-model knowledge structure to enhance intelligent system decisions for process design using templates and selected knowledge.These strategies provide a scientific basis for sustainable production and advance green manufacturing initiatives.
Process decision-making knowledge plays a critical role in the manufacturing sector, with its core being the precise evaluation and decision-making regarding issues such as material selection, processing methods, tool application, and operational sequence in the manufacturing process.As research in this field continues to deepen, particularly concerning the integration and application of this knowledge to optimize manufacturing processes and enhance the quality of final products, academic interest is steadily growing.
The current literature predominantly focuses on the collection and analysis of process data, aiming to determine the optimal manufacturing path.This process involves the handling of complex data and relies on intelligent algorithms, such as machine learning and optimization algorithms, which are increasingly being utilized in process decisionmaking.These algorithms are instrumental in pattern recognition, outcome prediction, and providing data-based decision support, thereby enhancing the flexibility and efficiency of the manufacturing process.With the increasing demand for automated and highly integrated process decision-making systems, it is particularly important to scientifically systematize and standardize process knowledge.Additionally, this knowledge must seamlessly integrate with smart manufacturing technologies (such as the Internet of Things, big data analysis, and cloud computing) to meet the requirements of Industry 4.0 standards.

Other Hybrid Knowledge
The complexity of manufacturing processes requires diverse knowledge representations.Shafiq et al. [138] innovated a DNA technique for decision-making based on virtual engineering objects, processes, and factory concepts for preventive problem management.Ye et al. [139] developed a CNC machining knowledge base using cloud technology and web ontology language and realized cloud-based process solution acquisition through the NoSQL database and MapReduce model.Guan et al. [140] explored welding knowledge graph building, using the BiLSTM + attention model and CR-CNN model to extract relationships from documents and applying association rule models to process specific relationships.Xiao et al. [141] comprehensively analyzed the key techniques of process knowledge graphs and considered their integration with large-scale language models.Lin et al. [142] proposed a knowledge representation and reuse model for civil aircraft structural maintenance cases based on a web ontology language to improve the accuracy of historical case retrieval.Pelzer et al. [143] acquired process knowledge in extrusion-based additive manufacturing through an interpretable machine learning approach.Sun et al. [144] investigated an object-oriented process knowledge representation and implemented a process manufacturability analysis system.Liu et al. [145] proposed a fault diagnosis and cause analysis model by combining fuzzy evidential reasoning and dynamic adaptive fuzzy Petri nets to accurately diagnose the causes of anomalies.These studies have shown that the integration of process fundamentals, design, management, and decision-making knowledge from multiple perspectives is essential to improve the efficiency and intelligence of the manufacturing process.
Hybrid knowledge involves the integration of cross-disciplinary knowledge, particularly in the context of cross-departmental product development, where it necessitates the fusion of foundational scientific knowledge, design principles, and management strategies.To effectively represent and utilize this knowledge often requires the reliance on multidisciplinary collaboration platforms and integrated tools to ensure the full utilization of diverse types of knowledge.This is not merely a challenge of technology and tools, but also a complex process involving a deep understanding of various disciplines, communication, and collaboration, serving as a key driver for innovation and addressing cross-disciplinary challenges.As delineated in Figure 5, the scholarly focus on knowledge acquisition and representation within the manufacturing domain over the preceding decade has garnered considerable interest in the academic community.The research trajectory within this sphere has exhibited a pattern of oscillation followed by a notable crescendo.Correspondingly, Figure 6 elucidates the impetus provided by technological innovations, especially the assimilation of cloud technologies, big data analytics, and artificial intelligence, on these research trends.The progressive refinement in knowledge representation mechanisms, evolving towards enhanced intelligence and automation, is a direct consequence of these advancements.Hybridized approaches have secured a prominent position within the research landscape, constituting approximately 23.23% of the investigative efforts.The ascendancy of such methodologies is attributed to their adept amalgamation of assorted techniques, spanning natural language processing, machine learning, deep learning, and traditional artificial intelligence strategies.These composite methods are tailored to navigate the complexities inherent in manufacturing process information, tackling multifaceted challenges including the amalgamation of heterogeneous knowledge reservoirs, the expeditious and precise retrieval of linguistic data, the automation of knowledge updates, and the sustained management of knowledge databases.As delineated in Figure 5, the scholarly focus on knowledge acquisition and representation within the manufacturing domain over the preceding decade has garnered considerable interest in the academic community.The research trajectory within this sphere has exhibited a pattern of oscillation followed by a notable crescendo.Correspondingly, Figure 6 elucidates the impetus provided by technological innovations, especially the assimilation of cloud technologies, big data analytics, and artificial intelligence, on these research trends.The progressive refinement in knowledge representation mechanisms, evolving towards enhanced intelligence and automation, is a direct consequence of these advancements.As delineated in Figure 5, the scholarly focus on knowledge acquisition and representation within the manufacturing domain over the preceding decade has garnered considerable interest in the academic community.The research trajectory within this sphere has exhibited a pattern of oscillation followed by a notable crescendo.Correspondingly, Figure 6 elucidates the impetus provided by technological innovations, especially the assimilation of cloud technologies, big data analytics, and artificial intelligence, on these research trends.The progressive refinement in knowledge representation mechanisms, evolving towards enhanced intelligence and automation, is a direct consequence of these advancements.Hybridized approaches have secured a prominent position within the research landscape, constituting approximately 23.23% of the investigative efforts.The ascendancy of such methodologies is attributed to their adept amalgamation of assorted techniques, spanning natural language processing, machine learning, deep learning, and traditional artificial intelligence strategies.These composite methods are tailored to navigate the complexities inherent in manufacturing process information, tackling multifaceted challenges including the amalgamation of heterogeneous knowledge reservoirs, the expeditious and precise retrieval of linguistic data, the automation of knowledge updates, and the sustained management of knowledge databases.Hybridized approaches have secured a prominent position within the research landscape, constituting approximately 23.23% of the investigative efforts.The ascendancy of such methodologies is attributed to their adept amalgamation of assorted techniques, spanning natural language processing, machine learning, deep learning, and traditional artificial intelligence strategies.These composite methods are tailored to navigate the complexities inherent in manufacturing process information, tackling multifaceted challenges including the amalgamation of heterogeneous knowledge reservoirs, the expeditious and precise retrieval of linguistic data, the automation of knowledge updates, and the sustained management of knowledge databases.As shown in Figure 7, despite the relative maturity of the research on manufacturing basic knowledge and the simplicity and generality of its representation, the research on process design knowledge, process decision-making knowledge, and process management knowledge is still a hotspot of current research.The knowledge representation in these areas is not only related to specific manufacturing processes but also goes deep into the nature of problem-solving and logical relationships, which are highly demanded by both industrial applications and user needs.Specifically, in the realm of complex manufacturing process knowledge acquisition, several significant challenges exist.Foremost among these is the issue of data quality and accuracy, which severely hinders knowledge gathering and codification.The presence of incomplete data [146], random noise, and inconsistency often undermine the reliability of analytical results [147].Additionally, merging data from varied sources and formats adds another layer of complexity to data processing.Addressing these issues requires strict quality control measures and advanced data-cleaning techniques to enhance the accuracy and truthfulness of knowledge extraction and analysis.
In the aspect of knowledge acquisition for manufacturing processes, as summarized in Table 5, different methods each have their respective advantages, disadvantages, and application scenarios.Given the complexity of current manufacturing processes, existing As shown in Figure 7, despite the relative maturity of the research on manufacturing basic knowledge and the simplicity and generality of its representation, the research on process design knowledge, process decision-making knowledge, and process management knowledge is still a hotspot of current research.The knowledge representation in these areas is not only related to specific manufacturing processes but also goes deep into the nature of problem-solving and logical relationships, which are highly demanded by both industrial applications and user needs.As shown in Figure 7, despite the relative maturity of the research on manufacturing basic knowledge and the simplicity and generality of its representation, the research on process design knowledge, process decision-making knowledge, and process management knowledge is still a hotspot of current research.The knowledge representation in these areas is not only related to specific manufacturing processes but also goes deep into the nature of problem-solving and logical relationships, which are highly demanded by both industrial applications and user needs.Specifically, in the realm of complex manufacturing process knowledge acquisition, several significant challenges exist.Foremost among these is the issue of data quality and accuracy, which severely hinders knowledge gathering and codification.The presence of incomplete data [146], random noise, and inconsistency often undermine the reliability of analytical results [147].Additionally, merging data from varied sources and formats adds another layer of complexity to data processing.Addressing these issues requires strict quality control measures and advanced data-cleaning techniques to enhance the accuracy and truthfulness of knowledge extraction and analysis.
In the aspect of knowledge acquisition for manufacturing processes, as summarized in Table 5, different methods each have their respective advantages, disadvantages, and application scenarios.Given the complexity of current manufacturing processes, existing Specifically, in the realm of complex manufacturing process knowledge acquisition, several significant challenges exist.Foremost among these is the issue of data quality and accuracy, which severely hinders knowledge gathering and codification.The presence of incomplete data [146], random noise, and inconsistency often undermine the reliability of analytical results [147].Additionally, merging data from varied sources and formats adds another layer of complexity to data processing.Addressing these issues requires strict quality control measures and advanced data-cleaning techniques to enhance the accuracy and truthfulness of knowledge extraction and analysis.
In the aspect of knowledge acquisition for manufacturing processes, as summarized in Table 5, different methods each have their respective advantages, disadvantages, and application scenarios.Given the complexity of current manufacturing processes, existing knowledge acquisition tools and technologies may not fully meet all demands.Therefore, it is necessary to further develop flexible tools that can adapt to specific environmental requirements.Additionally, transforming large amounts of data into practical knowledge is a crucial aspect; this requires not only efficient data processing technologies but also profound domain expertise to ensure the accuracy of data transformation and the practicality of the knowledge acquired [148].In knowledge representation for manufacturing processes, the specific methods of knowledge representation involve how to vividly and accurately present information within the manufacturing process.The following, Table 6, summarizes several commonly used knowledge representation methods in manufacturing processes, analyzes their main strengths and limitations, and discusses their practicality in different application scenarios.

Research Shortcomings and Prospects
According to the analysis of the existing literature in the field of industrial applications, the production and manufacturing process is highly customized.As a result, the knowledge acquired and represented in a specific environment is often difficult to be directly transferred for use in different manufacturing processes.This phenomenon has caused difficulties in the generalization of knowledge application, that is, there is an urgent need to explore effective ways to flexibly adjust and reuse knowledge.
As a result, current research on manufacturing process knowledge acquisition and representation faces multiple challenges.The first challenge is the high degree of fragmentation of knowledge representation.Various manufacturing knowledge domains such as fundamentals, design, decision-making, and management lack a unified framework for the orderly integration of information and collaborative work.This fragmentation has led to a lack of depth and breadth of knowledge, particularly in dealing with the application of new technologies and materials.Secondly, the dynamic nature of the manufacturing environment creates significant difficulties in updating and maintaining the knowledge base, limiting the level of automation.Interoperability and compatibility issues of knowledge representation also hinder the effective sharing and reuse of knowledge across different systems, tools, and organizations.
Furthermore, the inherent complexity of manufacturing processes is challenging to manage effectively in current knowledge representations, especially regarding interdisciplinary and cross-sectoral knowledge integration.At the same time, quality issues in manufacturing data collection, such as accuracy, completeness, and consistency, directly impact the core quality of knowledge representation.Although the introduction of AI technology has brought new development opportunities for knowledge representation, its extent of application in automated reasoning, knowledge discovery, and decision support remains limited.This indicates that further research is needed to optimize the combination of AI technology and knowledge engineering to enhance the efficiency and effectiveness of knowledge representation for manufacturing processes.
Figure 8 illustrates the potential sources of process knowledge in a smart manufacturing environment, encompassing input element content, value-added activities, smart manufacturing environment, smart manufacturing equipment, smart manufacturing objects, and process quality monitoring and control.Building upon this framework, this article envisages the evolution of knowledge acquisition and representation within the smart manufacturing process, considering the multidisciplinary intersections, the shift towards personalized and customized production driven by market demands, and the profound convergence of big data and artificial intelligence technologies.The following trends are further explored: (1) Interactive and collaborative knowledge creation: Future research will explore how computer-supported collaborative tools and platforms can foster knowledge creation and sharing within cross-functional teams.This interdisciplinary collaboration, grounded in advanced networking and communication technologies, aims to spark innovation within teams and facilitate knowledge development.The next generation of collaborative mechanisms and tools will enable instant communication, and efficient knowledge management, and utilize artificial intelligence technologies for recommending, identifying, and integrating interdisciplinary knowledge, thereby accelerating innovation.This process enhances knowledge exchange among experts from different fields, strengthening professional knowledge and broadening perspectives through cross-border cooperation, ultimately improving the innovation capacity and efficiency of manufacturing processes.Such interactivity and collaborative knowledge creation will propel the intelligent manufacturing industry toward greater intelligence, efficiency, and customization.(2) Adaptive evolution of process knowledge: The adaptive process knowledge evolution plays a critical role in intelligent manufacturing systems, which must possess the capability to adapt to continuously changing production demands and environmental conditions.To this end, the knowledge representation methods must exhibit the characteristic of self-evolution, continuously learning and adjusting to reflect new data, experiences, and environmental variables.Consequently, it is foreseeable that more advanced algorithms and models will be developed to support the dynamic up-dating and optimization of knowledge.These algorithms and models will be capable of processing and integrating information from diverse sources in real time, ensuring that intelligent manufacturing systems can flexibly address various production challenges while maintaining their efficiency and accuracy.Through this approach, intelligent manufacturing systems not only maintain their current performance levels but also continuously evolve over time to meet the increasingly complex and variable industrial demands.(3) Application of emerging technologies in knowledge acquisition and representation: Emerging technologies such as the IoT, blockchain, and virtual reality (VR) hold immense potential for enhancing the mechanisms of knowledge capture and representation within the intelligent manufacturing process.These technologies can provide precise real-time data, thereby improving the transparency of workflows and supply chains in the manufacturing environment.Furthermore, they can elevate operational efficiency and user satisfaction through enhanced user experiences and immersive interactions.(4) Enhancing the scope and accuracy of process knowledge representation: As intelligent manufacturing systems increasingly evolve towards more complex and sophisticated production processes, there is a heightened demand for a broader and more precise representation of process knowledge.To meet this requirement, it is imperative to adopt advanced data analytics methodologies, such as machine learning and deep learning techniques, to handle and interpret large-scale complex datasets.These methods enable the transformation of data into actionable knowledge, thereby facilitating a deeper understanding and optimization of the manufacturing processes.The application of these sophisticated technologies not only enhances the capability to extract insights from complex data but also plays a critical role in improving manufacturing decision-making processes and increasing the accuracy of predictive maintenance.
Machines 2024, 12, x FOR PEER REVIEW 24 of 30 production challenges while maintaining their efficiency and accuracy.Through this approach, intelligent manufacturing systems not only maintain their current performance levels but also continuously evolve over time to meet the increasingly complex and variable industrial demands.(3) Application of emerging technologies in knowledge acquisition and representation: Emerging technologies such as the IoT, blockchain, and virtual reality (VR) hold immense potential for enhancing the mechanisms of knowledge capture and representation within the intelligent manufacturing process.These technologies can provide precise real-time data, thereby improving the transparency of workflows and supply chains in the manufacturing environment.Furthermore, they can elevate operational efficiency and user satisfaction through enhanced user experiences and immersive interactions.(4) Enhancing the scope and accuracy of process knowledge representation: As intelligent manufacturing systems increasingly evolve towards more complex and sophisticated production processes, there is a heightened demand for a broader and more precise representation of process knowledge.To meet this requirement, it is imperative to adopt advanced data analytics methodologies, such as machine learning and deep learning techniques, to handle and interpret large-scale complex datasets.These methods enable the transformation of data into actionable knowledge, thereby facilitating a deeper understanding and optimization of the manufacturing processes.The application of these sophisticated technologies not only enhances the capability to extract insights from complex data but also plays a critical role in improving manufacturing decision-making processes and increasing the accuracy of predictive maintenance.In realizing these directions of development, both technical and managerial challenges will be faced.Technically, it is necessary to develop efficient knowledge representation and processing frameworks to meet the challenges posed by big data and complex systems.At the managerial level, it is necessary to establish mechanisms for interdisciplinary cooperation in order to integrate knowledge and technology from different fields.The only literature available shows that addressing these issues will be the key to successful implementation and continued progress in the field of smart manufacturing in the future.

Conclusions
In this review, we delve into the current state and future trends of manufacturing process knowledge acquisition and representation technologies in the context of smart manufacturing.Through meticulous analysis, this article not only elucidates the core principles and methodologies of data-driven knowledge extraction but also provides a multifaceted perspective on constructing an efficient knowledge representation model, encompassing key aspects such as manufacturing fundamentals, design, management, and decision-making.With the continuous evolution and deepening of intelligent analytics, there is an increasingly pressing need for high-quality, highly flexible knowledge discretization in the field of intelligent manufacturing.
It is evident that the acquisition and representation of manufacturing process knowledge are pivotal drivers of innovation and efficiency improvement in smart manufacturing.Currently, while large-scale manufacturing data can be processed and transformed with the aid of intelligent analytics, further optimization of the depth and breadth of knowledge acquisition, as well as enhancement of the accuracy and utility of knowledge representation, remains imperative.Future developments must concentrate on integrating innovative tools such as advanced machine learning algorithms, IoT technologies, digital twins, and AI-assisted design to maximize the exploitation of information resources and intelligent management of knowledge.This will propel the advancement of intelligent manufacturing to greater heights and contribute to the transformation and upgrading of the global manufacturing industry.

Figure 1 .
Figure 1.PRISMA flowchart for statistical analysis of the literature.Figure 1. PRISMA flowchart for statistical analysis of the literature.

Figure 1 .
Figure 1.PRISMA flowchart for statistical analysis of the literature.Figure 1. PRISMA flowchart for statistical analysis of the literature.

Figure 2 .
Figure 2. Research system for manufacturing process knowledge acquisition and representation.

Figure 2 .
Figure 2. Research system for manufacturing process knowledge acquisition and representation.

Figure 3 .
Figure 3. Classification framework for manufacturing process knowledge representation.

Figure 3 .
Figure 3. Classification framework for manufacturing process knowledge representation.

6 .
Research Summary, Shortcomings and Prospects 6.1.Research Summary In summation, knowledge acquisition and representation in manufacturing processes constitute a multidimensional and dynamically evolving field.The analysis of 123 literature studies indicates a consistent upward trend in research activity over the past decade, reflecting the sustained interest of both industry and academia in enhancing the technology of knowledge acquisition and representation in the manufacturing process.As illustrated in Figure 4, the growing research interest and technological advancements in this domain are evident through the increasing number of research papers published in specialized journals such as The International Journal of Advanced Manufacturing Technology, Advanced Engineering Informatics, Computers in Industry, and Computer Integrated Manufacturing Systems.Machines 2024, 12, x FOR PEER REVIEW 20 of 30

Figure 5 .
Figure 5. Distribution of years of publication.

Figure 5 .
Figure 5. Distribution of years of publication.

Figure 5 .
Figure 5. Distribution of years of publication.

Figure 6 .
Figure 6.Percentage of papers with different methods.

Figure 7 .
Figure 7. Distribution of different manufacturing process knowledge representation types.

Figure 6 .
Figure 6.Percentage of papers with different methods.

Figure 7 .
Figure 7. Distribution of different manufacturing process knowledge representation types.

7 .
Distribution of different manufacturing process knowledge representation types.

Figure 8 .
Figure 8. Knowledge acquisition and representation of the smart manufacturing process.Figure 8. Knowledge acquisition and representation of the smart manufacturing process.

Figure 8 .
Figure 8. Knowledge acquisition and representation of the smart manufacturing process.Figure 8. Knowledge acquisition and representation of the smart manufacturing process.

Table 1 .
Advances in research on manufacturing process fundamentals.

Table 1 .
Advances in research on manufacturing process fundamentals.

Table 2 .
Advances in process design knowledge research.

Table 3 .
Summary of process knowledge management.

Table 4 .
Summary of process decision-making knowledge studies.

Table 5 .
Advantages and disadvantages of knowledge acquisition approaches.

Table 6 .
Advantages and disadvantages of knowledge representation approaches.