Next Article in Journal
Special Issue “Smart Manufacturing Systems for Industry 5.0: Challenges and Opportunities”
Next Article in Special Issue
LSTM Network for the Oxygen Concentration Modeling of a Wastewater Treatment Plant
Previous Article in Journal
A Feasibility Study on Textile Sludge as a Raw Material for Sintering Lightweight Aggregates and Its Application in Concrete
Previous Article in Special Issue
Advanced Electronic and Optoelectronic Sensors, Applications, Modelling and Industry 5.0 Perspectives
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Applied Sciences
Volume 13
Issue 11
10.3390/app13116396
applsci-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Extended Fuzzy-Based Models of Production Data Analysis within AI-Based Industry 4.0 Paradigm

by
Izabela Rojek
*,
Piotr Prokopowicz
,
Piotr Kotlarz
and
Dariusz Mikołajewski
Faculty of Computer Science, Kazimierz Wielki University, 85-064 Bydgoszcz, Poland
*
Author to whom correspondence should be addressed.
Appl. Sci. 2023, 13(11), 6396; https://doi.org/10.3390/app13116396
Submission received: 20 April 2023 / Revised: 11 May 2023 / Accepted: 21 May 2023 / Published: 24 May 2023
(This article belongs to the Special Issue IIoT-Enhancing the Industrial World and Business Processes)
Browse Figures
Versions Notes

Abstract

:

Featured Application

Potential applications of the work include the computerization of production processes and the construction of artificial intelligence-based systems to support and optimize tool selection on existing and newly designed assembly lines in line with Industry 4.0 and Internet of Things paradigms.

Abstract

Fast, accurate, and efficient analysis of production data is a key element of the Industry 4.0 paradigm. This applies not only to newly built solutions but also to the digitalization, automation, and robotization of existing factories and production or repair lines. In particular, technologists’ extensive experience and know-how are necessary to design correct technological processes to minimize losses during production and product costs. That is why the proper selection of tools, machine tools, and production parameters during the manufacturing process is so important. Properly developed technology affects the entire production process. This paper presents an attempt to develop a post-hoc model of already existing manufacturing processes with the increased requirements and expectations resulting from the introduction of the Industry 4.0 paradigm. In particular, we relied on fuzzy logic to support the description of uncertainties, incomplete data, and discontinuities in the manufacturing process. This translates into better controls compared to conventional systems. An analysis of the proposed solution’s limitations and proposals for further development constitute the novelty and contribution of the article.

1. Introduction

Industry 4.0, i.e., the Fourth Industrial Revolution, involves the widespread implementation of artificial intelligence (AI), robotics, and the Internet of Things (IoT) to make processes more flexible, reduce the workforce, and make complex processes easier to control and manage. Industry 4.0 is applicable to all aspects of industry: analysis, diagnostics, manufacturing, technical control at all stages of production, packaging, and product/service evaluation throughout the life cycle [1,2,3,4]. Fast, accurate, and efficient analysis of production data is a key element of the Industry 4.0 paradigm. This applies not only to newly built solutions but also to the digitalization, automation, and robotization of existing factories and production or repair lines. In particular, technologists’ extensive experience and know-how are necessary to design correct technological processes to minimize losses during production and product costs. Tool selection is the cognitive process required to use a tool and is based on distinct semantic knowledge under different conditions [5]. Thus, a human (operator, technologist) selecting a tool performs the task of matching the tool(s) that achieves the objective of the activity and requires the selection of the best tool from among many candidates. The aforementioned procedure is repetitive, subject to errors due to selection fatigue, and, at the same time, should be subject to automation in order to reduce its duration as well as improve accuracy and repeatability. It also often requires complex manipulation procedures for the selected tool [6,7], whereas in some cases, the tool can be fed and assembled automatically. This is always conducted in the same way and with the same accuracy, reducing the preparation time of the machine. That is why the proper selection of tools, machine tools, and production parameters during the manufacturing process is so important.
Artificial intelligence (AI) has already provided massive improvements in many areas of technology, including in the wider economy. Process modeling, machine operation simulations, and entire production lines are expanding, as the development directions of equipment and its control systems are based on calculations carried out within their virtual twins. This fits within the Industry 4.0 paradigm. It also works the other way: the ever-new challenges posed to the process industry, including within the framework of the green deal, have already generated innovative technologies used to optimize processes and designs from this point of view, primarily in the form of intelligent, flexible, and open applications, including those that are fast enough to operate in near real time. So, the data generated from the sensors must be simple enough to not hinder the process, and the algorithms used must be computationally efficient so as not to be acquisitive in terms of time and energy. An additional requirement here may be the increasing processing of data on mobile devices, which is economical in the use of memory and processor computing capacity [8]. A number of previous studies have indicated numerous possible benefits of this process, requiring its planned implementation. These benefits include reducing costs (especially unit costs) by up to 25%, increasing production by up to 100%, increasing profitability by up to 30%, and accelerating the implementation of a new product by up to 6 months, giving a significant competitive advantage. The inclusion of computing technologies will allow for the improvement of enterprise data processing at all levels: material supply, planning, preparation, control of production processes, equipment load, active technical control at all stages of work in progress, and, hence, the presented benefits [9]. Fuzzy logic is often used to manage complex processes, especially with constraint satisfaction, linguistic description, and an uncertain dataset [10,11].
The development of technologies supporting Industry 4.0, including artificial intelligence, already opens up new possibilities for data collection and inference, prediction, and visualization in a way that is easily understood by humans. Properly developed AI technology affects the entire production process. Decision systems are becoming a necessity even in the selection of tools as a key element of the production process. Here, the data-driven approach becomes a key technology that allows for improved planning, the create of models that optimize design, the reduction of risks and implementation barriers, and the facilitation of the introduction of complex and intelligent production systems [12,13], production logistics systems, and multistage processes burdened with uncertainty [14,15,16,17].
The technological process is a fundamental part of the production process directly related to the change of shape, dimensions, surface quality, and physical–chemical properties of the work piece, as well as determining the mutual positioning of parts or assemblies in the product. From this definition follows the task (purpose) of the technological process, which is to change the state of a work piece from the initial state (semi-finished or initial material) into the final one (finished product). The design of a technological process is divided into stages. The first is the selection of a semi-finished product; the secondis the design of the technological process structure, i.e., determining the sequence of operations and procedures. Then for each operation and technological procedure, the following are selected in turn: the tooling of the object, the machine tool, the tool, the tooling of tools, and machining parameters.
In the vast majority of cases, standardized tools should be selected. When this is not possible, special tools should be chosen. When selecting a tool, the cutting material and its geometry must be taken into account; these factors determine the cutting properties and the full utilization of the tool in the machining process. The criteria for the selection of machining tools for operations are as follows: machining method, shape of machined surfaces, type and accuracy of machining, production volume, work piece material, and type of machine tool. The determination of these elements is the starting point for the development of an algorithm, by means of which the following are determined: technological possibilities of the tools, type of semi-finished material, type of machining (roughing, shaping, finishing), tool shape, and dimensions and operating cost.
In order to examine the current state of the art in process design, we reviewed the literature on the subject. In the area of selection of machine tools and machining parameters, Zhang et al. [18] applied fuzzy logic to the selection of machining parameters in a polishing process, while Tan et al. [19] converted an expert system for a carbide tool selection for a CNC lathe. Zhang et al. built a repository of polishing parameters within products and processes. They are used to model experience-based polishing process planning: each case is structured, matched with cases from the repository, and then the most similar cases are used to further analyze for verifiability and adaptability to create an optimal solution. This research combines fuzzy set theory with case-based reasoning to solve problems in polishing process planning: the values of product features and process parameters (polishing force, amount of polishing compound, polishing wheels, rotational speed, feed rate, etc.) cannot be accurately measured and controlled, and the relationships between process parameters and polishing quality indicators (surface roughness, thickness) cannot be scientifically determined mathematically [18]. The aforementioned studies aimed to develop an optimal system for selecting a tool holder, cutting tool, insert, and, in addition, machining parameters: feed and cutting speed. In addition, Igari et al. [20] proposed an optimal selection of tools and machining parameters using rules generated from decision trees. The system works on the basis of an updated machining database and a microprocess planning algorithm. The result of the decision-making process, i.e., the cutting conditions, is determined from a feasible regression equation [20]. Yan and Wang [21] proposed a fast and tailored decision-making method based on an improved support vector machine (ISVM) and production environment, one that is more responsive to dynamic changes and the uncertainty and fuzziness of the production environment. The SVM model, based on the triangular fuzzy number theory, is allowed here to represent uncertain input data or unbalanced samples [21]. The solution uses not only triangular fuzzy numbers representing uncertain inputs, but also independent penalty factors for different categories of unbalanced samples. This makes it possible to better respond to certainty, fuzziness, and randomness, but also to develop a new type of production guidance systems based on self-adaptation, self-learning, self-evolution, self-reconfiguration, and self-service. Saranya et al. demonstrated efforts to reduce human intervention in cutting tool selection and metal cutting process parameters using AI methods and techniques. The selection of a suitable cutting tool, as well as the optimization of process parameters, is realized by artificial neural networks, fuzzy logic, and a genetic algorithm. This is conducted based on multi-objective optimization criteria, including material removal rate, tool life, and tool cost [22]. Shetty et al. developed an expert system model based on response surface methodology (RSM) designed to predict the force generated during the cutting process when machining the Ti–6Al–4V alloy with minimum quantity lubrication (MQL) [23]. We should pay particular attention to the application of a fuzzy approach to modeling, optimization, and the control of continuous production [24,25].
In the area of complex process design, Fichtner et al. [26], Nassehiet et al. [27], Agraval et al. [28], and XieandTu [29] used an agent-based technique for process design. In addition, Joo et al. [30] presented the concept of a process planning system together with the selection of machine tools, machining parameters, toolpath and numerical control (NC) code using neural networks, genetic algorithm, and intelligent search [31]. Rojek [32] presented the use of decision trees for the selection of machine tools and machining parameters. Publications by Kacalak and Majewski [33], as well as Kacalak et al. [34], presented the concept of intelligent, interactive, automated systems for designing machine components and assemblies based on their features described in natural language (linguistically) using artificial neural networks and fuzzy inference. This achieves the normalization of design features as input quantities to an automatic technological classification system. The analysis of the research carried out so far indicates that a significant range of problems related to making optimization decisions for machines remains to be solved. These include: the selection of a machine for processing parts (one or more), an analysis of kinematic and dynamic parameters of the machine, tool selection (in the case of many: selection of the order of their use), selection of the optimal tool path, and the optimization of the selection of control and reinforcement strategies, including tool wear, energy, and costs in general [35].
Intelligent networked AI systems enable a faster and more efficient use of the knowledge in the data, making much more effective decisions. This is based on web-based tools, performance analysis, shared knowledge adaptively tailored to specific organizations, and their business scenarios, AI/machine learning (ML), robotics, chatbots and IoT [36,37].
An analysis of the state of research on computer-aided design of technological processes has shown that AI methods (expert systems with decision rules, neural networks, genetic algorithms), as well as their combinations in hybrid systems, are frequently used in this area. These methods are used to select sequences of technological operations, machine tools, and machining parameters. Attempts are also being made to design larger sequences of technological processes using integrated AI methods, which are particularly useful in the representation and processing of uncertain, imprecise, and incomplete knowledge, advanced data analysis, data mining and knowledge discovery, and intelligent decision support. The application of AI to process design allows the technologist’s experience and knowledge to be extracted into knowledge bases, which in turn allows the system to self-learn and then equip it with human-like reasoning skills.
The computer-aided design of technological processes uses individual technological processes of products that have been developed and implemented in the enterprise. On this basis, the developed models learn to select a semi-finished product, create a technological process structure, and select a work-piece tooling, machine tool, tools, tool fixtures, and machining parameters (including machining time, tool life). The models come in the form of neural networks, trees, and decision rules and are applied in the developed expert system, which allows the technological process of a new product to be designed. Such technological process designs using AI methods and techniques are comprehensive and holistic in nature, from the selection of the semi-finished product to the creation of the technological process structure to the selection of work-piece tooling, machine tools, and tooling, up to and including machining parameters.
This paper presents an attempt to develop a post-hoc model of already-existing manufacturing processes with the increased requirements and expectations resulting from the introduction of the Industry 4.0 paradigm.
The article has the following layout: the Introduction section provides an introduction to Industry 4.0 and its use of artificial intelligence methods and tools with a particular focus on fuzzy logic. The Material and Methods section presents the source and form of the datasets used in the study and the computational tool used, with particular emphasis on where and how to use proprietary fuzzy models. The Results section presents the two variants of the solution, highlighting the differences between them. The Discussion section compares the results of the in-house study with data available in the literature and points out the strengths, weaknesses, and limitations of the in-house solution, as well as directions for further research. The article concludes with the Conclusions section.

2. Materials and Methods

2.1. Material

Cases for analysis consisted of 553 examples of tool selection (553 records).
The input data are a given type of machining (e.g., roughing, finishing), type of machined surface (irregular shape, groove, contour, plane), type of work piece (e.g., 316 L), numerical value of the Ra roughness parameter, type of milling cutter (e.g., monolith), type of mounting-cutter diameter (e.g., socket), cutter diameter in mm, cutter shape (e.g., cylindrical), number of teeth (e.g., 10), total cutter length [mm], cutting speed range in m/min, depth of cut in mm, feed rate in mm/min, milling width mm, and operating cost in PLN/h.
The output data are tools matched to the machining requirements (symbol of the selected cutter).

2.2. Methods

Fuzzy logic reduces the ambiguity of making decisions regarding the selection of a tool for machining [38,39,40,41]. It is realized by transforming descriptive imprecise information expressed by the technologist/operator of the device into numerical data in the Mamdani-type fuzzy model. After such a process, the linguistic data can be processed by computing systems, which improves the exchange of information between humans and the computing system.
This is progress towards the development of computational models of devices, processes, and products (digital twins) that improve the prediction of errors, failures, and downtime while also improving accuracy, efficiency, and safety [42,43].

Identification of the Fuzzy Sets for the Inputs

The identification of the input variables is based on the clustering of the reference dataset. We use eight parameters to describe the machining conditions (see the description column in Table 1). Thus, a single reference data (a data row) is a set of eight numerical input parameters that are specific to a particular machining tool. Overall, we use data rows for 17 different tools that cover most of the machining operating conditions.
It is worth noting that some tools are described by one row of data, others by two or four. In summary, we have 47 rows of data that indicate 17 different tools for machining. As can be seen from the above, we have 47 benchmark values for each input parameter. However, although we have many values, they are often repeated for different tools. A cluster is the number of occurrences of a given value (or range) in the reference data. The values and their occurrences shown in Table 1 are the basic clusters for further clustering to define fuzzy sets.
The identification of fuzzy sets for input variables is done by some steps of collecting neighboring initial clusters into extended clusters, which we can call granules. Such granules can be represented by intervals of values of a given input space. These intervals are represented as trapezoidal fuzzy sets.
Generally, we take into consideration two coefficients/characteristics of the available reference data:
  • The relative increase in the values of a given linguistic variable.
  • The total number of occurrences of data representing a given interval of values of that variable.
To summarize, the general process of defining fuzzy sets groups clusters of data and transforms their range into a fuzzy interval that is representative of a given input variable.
Version 1 and Version 2 differ in the assignment of clusters from the reference data.
In Variant 1, the assumptions were intuitive. The grouping of clusters into granules, represented by fuzzy sets (starting from lower values), proceeds as follows:
  • If the total number of occurrences in the clusters is less than 25%, and the interval of values of a given space is greater than 50% (of the entire range of this variable), we indicate the entire group as a new fuzzy set (new quality).
  • If the number of occurrences is between 25 and 50%, and the value interval is greater than 25% of the entire range of this variable, we transform the clustered data into a fuzzy set.
  • If the number of occurrences is above 50% and the value increment in the reference data is 10%, we indicate a new fuzzy set.
It is worth noting that after defining a new fuzzy granule/set, if there are clusters that are not clustered, we start clustering them for the next fuzzy set.
All of the above steps represent intuitive mechanisms for clustering data, which we can present as follows:
  • We group the data into larger clusters if the set of basic clusters is small but represents a wide range of values.
  • We group the data if the set of basic clusters is medium, and they represent a medium range of values.
  • We group the data if the set of basic clusters is large, even if the data they represent has a small range of values.
It should be noted that the roughness variable is authoritatively divided into two sets due to the specificity of the data, i.e., a division practically into two groups, with the benchmark data having the lowest value per unit.
In Variant 2, the division into fuzzy sets is systematic and based on strictly accepted criteria of observed relationships in the reference data. It represents intuitions that we can express, such as:
  • We group the data into larger clusters if the set of underlying clusters is at least average.
  • Additionally, if another cluster minimally extends the current range of values, we still add it to the group regardless of its size.
We practically implement this by starting from the lower boundary. We then group subsequent clusters into a single granular/fuzzy set following the principles:
  • A cumulative increment of at least 25% of occurrences results in a new group (granule)—fuzzy set starting from the next basic cluster;
  • Additionally, the increment of values in the variable space must be at least 5% or more.If not, we still add another basic cluster to the previous group.
Both of these variants are basis for finding the intervals of cores (kernels) of the fuzzy sets (the membership is equal 1).
Figure 1 shows detailed data of the cores of fuzzy sets for input variables of Variant 2, as it is most efficient solution presented in this paper.
Since the reference data represent the parameters of machining tools, their grouping consists mainly in combining the base clusters into larger ranges. Therefore, it is more convenient to describe them with trapezoidal sets, especially triangular sets, which appear in several cases and can be easily represented also as trapezoidal shape.
All the edges in the fuzzy sets are made by adopted constant spread factor of 25%. With this factor, the models can be adjusted in the future for data other than the benchmark data (including measurement/operational data from the real systems).We also used several specific formulas, including interesting functions of fuzzy set memberships.
In summary, the trapezoidal fuzzy sets were used, and their defining was performed as follows.
  • We identify the core (using Variant 1 or Variant 2 on benchmark data) to obtain a set of intervals for a given input space: [kis;kie]—where ks, ke—lower and upper boundary of the i-th interval for given input variable from benchmark data.
  • Definition of trapezoidal shape is described by the fourth [l; ks; ke; r], where l is left (down) and r is right (upper) boundary of support of fuzzy membership. That form is popular in many scientific software like Matlab, Octave, Scilab. As was mentioned earlier for the fuzzy edges, the constant parameter spread = 25% was used. Finally, definition of the trapezoid fuzzy set is done as: [ks − 0.25ks; ks; ke; ke + 0.25 × ke].
As for the output for a classification purpose, it would be enough to use activation levels as indicator of the class. However, for completeness of processing fuzzy rules, we used, for each output, simple linguistic variable Y:
Yi = (Yino, Yiyes),
        Yino = (0;0;0;1), Yiyes = (0;1;1;1),
where:
i—number of output variable (class to recognize)
Yino,Yiyes—trapezoidal fuzzy numbers indicating negation or confirmation of the class recognition.
In fact, in the rule base for classification, we use only confirmation parts of output variables, so with the Middle of Maximum (MOM) defuzzification, it gives us just transformation of rule activation on the output space (Table 2).

2.3. Specificity of the Rules

The specificity of the presented method is that the benchmark data give us indication of the certain elements, but it is not complete knowledge. We accept that is not complete data. Therefore, we do not define linguistic numbers with typical covering with the fuzzy numbers. What we have is the universe of the variable, and the indication of the intervals’ (granules) characteristic for the given output data. Fuzzy sets defined with such intervals do not fulfill typical expectations from the linguistic variables as membership sum to 1 for all elements of the input space. Figure 1 shows one of the generated input variables regarding source data and spread.
As we attempt to recognize known data classes without a complete knowledge about whole input parameters, we may propose only rules that point us to the information we are looking for. Therefore, having 47 reference datasets, we transform them into 47 rules.
The rules came from the reference data and have form:
IFv1isFS1j1 and v2 is FS2j2 and … and vk is FSkjkTHENYl = Ylyes,
where: v—inputs, k—the number of input spaces, FSkjk—j-th fuzzy set of k-th input space, Yl—l-th output space, Ylyes—fuzzy set describing confirmation of recognition a class.
The structure of the system is presented in Figure 2.

3. Results

We have proposed two versions of the solution, applied depending on how the research problem is posed. Both solutions used the configurations of Mamdani fuzzy system type as follows:
  • Aggregation of premises in the rules: MIN.
  • Implication: MIN.
  • Aggregation of results from the rules (accumulation): MAX.
  • Defuzzification: middle of maxima (MOM).
It is worth underlining that we tried other configurations with aggregation and implication operator type PROD, and also with defuzzification Center of Gravity (COG). However, the most efficient configuration was the one listed above.
The results are presented in Table 3 and Table 4. The values are the degree of recognition/adaptation of the tool in question. The columns denote the recognized tool (output class), which represent 17 different classes. The rows represent the different input sets that were used (from the reference data). The last column indicates which input set should recognize the tool from the reference data. A value of 0.99 is understood as full recognition (a value equal to 1) due to the rounding formats of digital calculations. The symbol N means that the given rule did not produce any result for this output (recognised tool/class). For practical purposes, we can treat this as a null value—the default in the absence of recognition.
It can be observed that some inputs generate more than one recognition. Thanks to the fuzzy mechanism, we can interpret that, as in the conditions described by the input values, different tools could have been used, but we still have a degree describing which tool is better.
Analyzing the results presented, it can be seen that for all entries, the correct tool (expected, indicated in the last column) was fully recognized. However, in Variant 1, for some entries such as Rows 2 and 4, among others, other tools were also recognized. In fact, Entry/Row 4 (but also 7, 19, and several others) demonstrates that four different tools were fully recognized. Such a situation is rather less valuable for an expert system in the context presented.
Continuing the analysis, if we look at Option 2, we see that full recognition only applies to the expected tools. If more than one tool is matched for any input set, the remaining values are less than 1. This is valuable behavior for an advisory expert system.

4. Discussion

From the perspective of previous research and working hypotheses, this study extends existing experiments mainly using decision trees and random forests and traditional artificial neural networks with fuzzy logic [42,43,44,45].The use of fuzzy logic for classification of machining tools has so far been rare. This is because fuzzy classification usually means grouping elements into fuzzy sets, where each observation is assigned a degree of membership to a particular fuzzy set. This allows for a linguistic description of a set of numerical data, and the understanding of implicit classification patterns becomes easier to understand than with traditional logic. The use of fuzzy logic seems to increase with the complexity of the problem and patterns. In the medical and health sciences alone, we found 1124 publications addressing the problem of classification using fuzzy logic. These include the classification of rare diseases or the grouping of patients [46,47,48], but also geological [49,50], climatic phenomena [51,52], human behavior [53], or crop maturity classifications [54].
The computerization, automation, and robotization of solutions associated with the growing implementation of the 4.0 paradigm is becoming increasingly complex, requiring the use of increasingly sophisticated methods, tools, and entire fuzzy systems based on fuzzy classifiers using not only data, but also expert opinion or a combination of both.
Fuzzy systems provide greater robustness against inaccurate, distorted, or incomplete input data, including when feeding data from several types of input simultaneously in which precise input data is not required. This also applies to emergency situations (e.g., excessive inaccuracies), which can be anticipated in the system and signaled with alerts to human operators or technologists, preventing material loss or machine damage (Figure 3).
There are many phenomena and mechanisms that require the use of fuzzy logic to be considered in the production process; their number will increase as the complexity of processes, machines, equipment, technologies (e.g., 3D printing), and materials (e.g., shape memory or multi-material products) increases.
Previous approaches on a similar dataset have shown the effectiveness of the dynamic ensemble selection approach based on the value of the score function used in the one-to-one decomposition scheme and its advantages over the static ensemble selection method based on the integration of base classifiers in geometric space (also a one-versus-one decomposition scheme) [44]. Our approach goes one step further, demonstrating effectiveness in the case of uncertain, incomplete, or partially overlapping data (i.e., when more than one processing tool can be used).
Duch points out that complex problems are difficult to analyze with precision, and that the expert’s knowledge in such cases lends itself to fuzzy descriptions. Hence, different types of uncertainties arise:
  • Stochastic uncertainty (based on probability calculus, e.g., dice throw, accident, insurance risk).
  • Measurement uncertainty (based on statistical analysis of the problem, e.g., about 2.5 cm or about 10 points).
  • Informational uncertainty (based on data mining. e.g., a reliable borrower, a good candidate, i.e., one who collectively meets certain conditions).
  • Linguistic uncertainty (based on fuzzy logic, e.g., favorable price, nice weather).
In doing so, we must bear in mind that thinking, inference, and decision-making is not a universal process and is usually based on patterns that depend on the field of knowledge [55].

4.1. Limitations of the Proposed Solution

A major limitation of the study is the actual origin of the data, which reflect one particular assembly line and may be specific. This requires keeping the model flexible for the future. We also assume that the number of parameters is finite, and that their range is defined, which may limit the use of materials and tools from other manufacturers. On the other hand, it shows the possibility of creating dedicated models within model frameworks for digital twins equipped with fuzzy logic.
At the moment, there are no process models with many different data sources/forms for which a fuzzy system with many inputs can be built.

4.2. Directions for Further Research

The basic directions of further research within Industry 4.0 paradigm include:
  • Use of directed fuzzy numbers.
  • Development of hybrid methods, combining the advantages of different solutions (See, for example, the classification of textile products and fiber-reinforced polymer composites using artificial neural networks, genetic algorithm, and fuzzy logic [56], or aerogels combining enhanced physicochemical properties and structural features with sensory and energetic materials [57].)
  • Development of a framework for the development of various models of industrial processes, adapted on site.
  • Algorithms based on fuzzy data to reduce subjectivity in estimation [58], including uncertainty in multi-criteria decision analysis [59,60,61].
  • Assessing the performance of the logistic chain, including the food chain, and preventing threats to supply continuity and product safety [62,63].
  • Simultaneous analysis of product/service quality, price, and consumer satisfaction.
  • Incorporating this group of solutions into ERP systems.
The use of fuzzy logic and other AI methods increases the scope of possibilities for using data contained in technological databases [64,65]. The experiments carried out have confirmed the usefulness of fuzzy logic in the selection of tools for a technological operation, which is a stage of designing the technological process of a product.
AI methods, including fuzzy logic, introduce a new quality of technological process redesign using computer systems. A technologist’s knowledge is largely based on his or her years of experience. As a result, the decisions he or she makes are often “intuitive”—the expert himself or herself is sometimes unable to concretely express the rule he or she followed when making a particular decision. Gathering knowledge in the form of examples of already-designed technological processes verified during the production of products is, therefore, the best way to discover knowledge from the technologist’s experience. This knowledge can then be encapsulated in computer systems using AI methods. In turn, it can be transformed into rules that can be used to design technological processes for new products. We are aiming here at so-called learning systems, which will automatically design the technological process (for example, the use of fuzzy logic here).
This is particularly important when developing computer-aided process planning (CAPP) systems for complex real-world systems. However, the disadvantage of this approach is the complexity of the resulting CAPP system. Models for the selection of work-piece tooling, machine tooling, and machining parameters must be developed for each technological operation separately (for milling, grinding, turning, etc.). The neural network learning file contains quantitative and qualitative parameters, so their normalization and coding are necessary. There is also the problem of acquiring new knowledge. The developed models need to be trained periodically, i.e., according to incoming new data. This data include both new machine tools and tools that appear in the company and further examples of developed technological processes. In conclusion, it is clear that the advantages of the selected artificial intelligence methods—neural networks and decision trees—clearly outweigh the disadvantages. In addition, the practical knowledge contained in the computer system that takes into account the experience of technologists and specific data from the company makes it possible to design technological processes also for less experienced technologists. A new employee does not know the capabilities or conditions of the company’s machinery stock. If he or she were to attempt to design a technological process using only catalogue values, the end result could prove unsatisfactory. As a result of many years of operation of a machine tool, its parameters specified in the technical and operational documentation are completely different from those in reality. The expert system, by training the models on an ongoing basis, takes these changes into account, which makes it possible to avoid errors in the design of the technological process and, consequently, to minimize the company’s losses. It should be emphasized that the computer system has an advisory role, and that the final decisions always belong to the technologist.
The developed methodologies and models (neural network structures and decision trees) can be applied to various enterprises as they constitute a universal tool that only needs to be individualized, i.e., adapted to the expectations of a given user. To this end, it is necessary to equip prototype expert systems with knowledge based on the technological data of a specific enterprise.
The theoretical considerations and experimental results presented in this paper expand the possibilities of designing and supervising the technological process of machining, and the developed methodologies, models, and prototype expert systems based on neural networks, trees, and decision rules are the author’s contribution to the development of research on artificial intelligence in machine construction and operation.

5. Conclusions

In the article, we relied on fuzzy logic to support the description of uncertainties, incomplete data, and discontinuities in the manufacturing process. This translates into better controls compared to conventional systems. An analysis of the proposed solution’s limitations and proposals for further development constitute the novelty and contribution of the article.
The classification proposed here fully recognizes the tools and is based on a much smaller (reduced set of) reference data used previously. In addition, the fuzzy approach allows the identification of similar tools that may be the closest alternative when the most suitable ones are not available. This is an important added value, especially important in practical applications of expert systems.

Author Contributions

Conceptualization, I.R., P.P. and D.M.; methodology, I.R., P.P. and D.M.; software, P.P.; validation, I.R., P.P. and D.M.; formal analysis, I.R., P.P. and D.M.; investigation, I.R., P.P. and D.M.; resources, I.R., P.P., P.K. and D.M.; data curation, I.R., P.P. and D.M.; writing—original draft preparation, I.R., P.P., P.K. and D.M.; writing—review and editing, I.R., P.P., P.K. and D.M.; visualization, I.R., P.P., P.K. and D.M.; supervision, I.R. and P.P.; project administration, I.R.; funding acquisition, I.R. All authors have read and agreed to the published version of the manuscript.

Funding

The work presented in the paper has been financed under grant to maintain the research potential of Kazimierz Wielki University.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Not applicable.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Tepljakov, A. Intelligent Control and Digital Twins for Industry 4.0. Sensors 2023, 23, 4036. [Google Scholar] [CrossRef] [PubMed]
  2. Sharma, D.; Patel, P.; Shah, M. A comprehensive study on Industry 4.0 in the pharmaceutical industry for sustainable. Environ. Sci. Pollut. Res. Int. 2023. [Google Scholar] [CrossRef] [PubMed]
  3. Werbińska-Wojciechowska, S.; Winiarska, K. Maintenance Performance in the Age of Industry 4.0: A Bibliometric Performance Analysis and a Systematic Literature Review. Sensors 2023, 23, 1409. [Google Scholar] [CrossRef]
  4. Mazzei, D.; Ramjattan, R. Machine Learning for Industry 4.0: A Systematic Review Using Deep Learning-Based Topic Modelling. Sensors 2022, 22, 8641. [Google Scholar] [CrossRef]
  5. Tobia, M.J.; Madan, C.R. Tool selection and the ventral-dorsal organization of tool-related knowledge. Physiol. Rep. 2017, 5, e13078. [Google Scholar] [CrossRef] [PubMed]
  6. Finkel, L.; Hogrefe, K.; Frey, S.H.; Goldenberg, G.; Randerath, J. It takes two to pantomime: Communication meets motor cognition. Neuroimage. Clin. 2018, 19, 1008–1017. [Google Scholar] [CrossRef] [PubMed]
  7. Lee, C.L.; Huang, H.W.; Federmeier, K.D.; Buxbaum, L.J. Sensory and semantic activations evoked by action attributes of manipulable objects: Evidence from ERPs. Neuroimage 2018, 167, 331–341. [Google Scholar] [CrossRef]
  8. Rojek, I.; Macko, M.; Mikołajewski, D.; Saga, M.; Burczynski, T. Modern methods in the field of machine modelling and simulation as a research and practical issue related to Industry 4.0. Bull. Pol. Acad. Sci. Tech. Sci. 2021, 69, e136717. [Google Scholar] [CrossRef]
  9. Chlebus, E. CAx Computer Techniques in Production Engineering; WNT: Warsaw, Poland, 2000. [Google Scholar]
  10. Prokopowicz, P.; Mikołajewski, D.; Mikołajewska, E.; Kotlarz, P. Fuzzy system as an assessment tool for analysis of the health-related quality of life for the people after stroke. Lect. Notes Comput. Sci. 2017, 10245, 710–721. [Google Scholar]
  11. Mikołajewska, E.; Prokopowicz, P.; Mikołajewski, D. Computational gait analysis using fuzzy logic for everyday clinical purposes—Preliminary findings. Bio-Algorithms Med.-Syst. 2017, 13, 37–42. [Google Scholar] [CrossRef]
  12. De Paula Ferreira, W.; Armellini, F.; de Santa-Eulalia, L.A. Simulation in industry 4.0: A state-of-the-art review. Comput. Ind. Eng. 2020, 149, 106868. [Google Scholar] [CrossRef]
  13. Cioffi, R.; Travaglioni, M.; Piscitelli, G.; Petrillo, A.; Parmentola, A. Smart Manufacturing Systems and Applied Industrial Technologies for a Sustainable Industry: A Systematic Literature Review. Appl. Sci. 2020, 10, 2897. [Google Scholar] [CrossRef]
  14. Li, G.; Yang, S.; Xu, Z.; Wang, J.; Ren, Z.; Li, G. Resource allocation methodology based on object-oriented discrete event simulation: A production logistics system case study. CIRP J. Manuf. Sci. Technol. 2020, 31, 394–405. [Google Scholar] [CrossRef]
  15. Davis, R.; Vochozka, M.; Vrbka, J.; Neguriţă, O. Industrial Artificial Intelligence, Smart Connected Sensors, and Big Data-driven Decision-Making Processes in Internet of Things-based Real-Time Production Logistics. Econ. Manag. Financ. Mark. 2020, 15, 9–15. [Google Scholar] [CrossRef]
  16. Santos, C.H.D.; de Queiroz, J.A.; Leal, F.; Montevechi, J.A.B. Use of simulation in the industry 4.0 context: Creation of a Digital Twin to optimise decision making on non-automated process. J. Simul. 2022, 16, 284–297. [Google Scholar] [CrossRef]
  17. Sun, H.; Pedrielli, G.; Zhao, G.; Zhou, C.; Xu, W.; Pan, R. Cyber coordinated simulation for distributed multi-stage additive manufacturing systems. J. Manuf. Syst. 2020, 57, 61–71. [Google Scholar] [CrossRef]
  18. Zhang, Y.; Huang, G.Q.; Ngai, B.K.K.; Chen, X. Case-based polishing process planning with Fuzzy Set Theory. J. Intell. Manuf. 2010, 21, 831–842. [Google Scholar] [CrossRef]
  19. Tan, C.F.; Kher, V.K.; Ismail, N. An expert system carbide cutting tools selection system for CNC lathe machine. Int. Rev. Mech. Eng. 2012, 6, 1402–1405. [Google Scholar]
  20. Igari, S.; Tanaka, F.; Onosato, M. Customization of a Micro Process Planning System for an Actual Machine Tool based on Updating a Machining Database and Generating a Database-Oriented Planning Algorithm. J. Control. 2013, 26, 87–94. [Google Scholar] [CrossRef]
  21. Yan, H.S.; Wang, Y.F. Matching decision method for knowledgeable manufacturing system and its production environment. J. IntellManuf. 2019, 30, 771–782. [Google Scholar] [CrossRef]
  22. Saranya, K.; John RozarioJegaraj, J.; Ramesh Kumar, K.; Rao, G.V. Artificial Intelligence Based Selection of Optimal Cutting Tool and Process Parameters for Effective Turning and Milling Operations. J. Inst. Eng. India. Ser. C 2018, 99, 381–392. [Google Scholar] [CrossRef]
  23. Shetty, R.; Kumar, C.R.S.; Ravindra, M.R. RSM based expert system development for cutting force prediction during machining of Ti–6Al–4V under minimum quantity lubrication. Int. J. Syst. Assur. Eng. Manag. 2021. [Google Scholar] [CrossRef]
  24. Orazbayev, B.; Zhumadillayeva, A.; Orazbayeva, K.; Iskakova, S.; Utenova, B.; Gazizov, F.; Ilyashenko, S.; Afanaseva, O. The System of Models and Optimization of Operating Modes of a Catalytic Reforming Unit Using Initial Fuzzy Information. Energies 2022, 15, 1573. [Google Scholar] [CrossRef]
  25. Orazbayev, B.B.; Orazbayeva, K.N.; Utenova, B.E. Development of mathematical models and modeling of chemical engineering systems under uncertainty. Theor. Found. Chem. Eng. 2014, 48, 138–147. [Google Scholar] [CrossRef]
  26. Fichtner, D.; Nestler, A.; Dang, T.N.; Schulze, A.; Carlsen, U.; Schreiber, S.; Lee, S.W. Use of agents and neural networks for acquisition and preparation of distributed NC information to support NC planning. Int. J. Comput. Inte.-Grated. Manuf. 2006, 19, 581–592. [Google Scholar] [CrossRef]
  27. Nassehi, A.; Newman, S.T.; Allen, R.D. The application of multi-agent systems for STEP-NC computer aided process planning of prismatic components. Inter.-Natl. J. Mach. Tools Manuf. 2006, 46, 559–574. [Google Scholar] [CrossRef]
  28. Agraval, R.; Shukla, S.K.; Kumar, S.; Tiwari, M.K. Multi-agent system for distributed computer-aided process planning problem in e-manufacturing environment. Int. J. Adv. Manuf. Technol. 2009, 44, 579–594. [Google Scholar] [CrossRef]
  29. Xie, S.Q.; Tu, Y. An Agent-based Sheet Metal Process Planning System. In Rapid One-of-a-kind Product Development: Strategies, Algorithms and Tools; Springer: Berlin/Heidelberg, Germany, 2011; pp. 125–148. [Google Scholar] [CrossRef]
  30. Joo, J.; Park, S.; Cho, H. Adaptive and dynamic process planning using neural networks. Int. J. Prod. Res. 2010, 39, 2923–2946. [Google Scholar] [CrossRef]
  31. Salehi, M.; Tavakkoli-Moghaddam, R. Application of genetic algorithm to computer-aided process planning in preliminary and detailed planning. Eng. Appl. Artif. Intell. 2009, 22, 1179–1187. [Google Scholar] [CrossRef]
  32. Rojek, I. MLP Neural Nets in Design of Technological Process. Adv. Manuf. Sci. Technol. 2015, 39, 87–95. [Google Scholar]
  33. Kacalak, W.; Majewski, M. Inteligentne, interaktywne, zautomatyzowane systemy projektowania elementówi zespołów maszyn na podstawie ich cech opisywanych w języku naturalnym. Pomiary Autom. Kontrola 2012, 58, 471–474. [Google Scholar]
  34. Kacalak, W.; Majewski, M.; Stuart, K.D.; Budniak, Z. Interactive systems for designing machine elements and assemblies. Manag. Prod. Eng. Rev. 2015, 6, 10–20. [Google Scholar] [CrossRef]
  35. Chryssolouris, G.; Alexopoulos, K.; Arkouli, Z. Artificial Intelligence in Manufacturing Equipment, Automation, and Robots. In A Perspective on Artificial Intelligence in Manufacturing. Studies in Systems, Decision and Control; Springer: Cham, Switzerland, 2023; Volume 436. [Google Scholar] [CrossRef]
  36. Turban, E.; Aronson, J.E.; Liang, T.-P.; McCarthy, R.V. Decision Support Systems and Intelligent Systems, 7th ed.; Prentice Hall: Hoboken, NJ, USA, 2004. [Google Scholar]
  37. Sharda, R.; Delen, D.; Turban, E. Analytics, Data Science & Artificial Intelligence: Systems for Decision Support, 11th ed.; Pearson: London, UK, 2019. [Google Scholar]
  38. Kosiński, W.; Prokopowicz, P.; Rosa, A. Defuzzification Functionals of Ordered Fuzzy Numbers. IEEE Trans. FuzzySyst. 2013, 21, 1163–1169. [Google Scholar] [CrossRef]
  39. Ahmadi, H.; Gholamzadeh, M.; Shahmoradi, L.; Nilashi, M.; Rashvand, P. Diseases diagnosis using fuzzy logic methods: A systematic and meta-analysis review. Comput. Methods Programs Biomed. 2018, 161, 145–172. [Google Scholar] [CrossRef] [PubMed]
  40. Hernández-Julio, Y.F.; Prieto-Guevara, M.J.; Nieto-Bernal, W.; Meriño-Fuentes, I.; Guerrero-Avendaño, A. Framework for the Development of Data-Driven Mamdani-Type Fuzzy Clinical Decision Support Systems. Diagnostics 2019, 9, 52. [Google Scholar] [CrossRef]
  41. Farzandipour, M.; Nabovati, E.; Saeedi, S.; Fakharian, E. Fuzzy decision support systems to diagnose musculoskeletal disorders: A systematic literature review. Comput. Methods Programs Biomed. 2018, 163, 101–109. [Google Scholar] [CrossRef] [PubMed]
  42. Rojek, I.; Mikołajewski, D.; Dostatni, E. Digital twins in product lifecycle for sustainability in manufacturing and maintenance. Appl. Sci. 2021, 11, 31. [Google Scholar] [CrossRef]
  43. Rojek, I.; Mikołajewski, D.; Dostatni, E.; Macko, M. AI-optimized technological aspects of the material used in 3D printing processes for selected medical applications. Materials 2020, 13, 5437. [Google Scholar] [CrossRef]
  44. Rojek, I.; Burduk, R.; Heda, P. Ensemble selection in one-versus-one scheme—Case study for cutting tools classification. Bull. Pol. Acad. Sci. Tech. Sci. 2021, 69, e136044. [Google Scholar] [CrossRef]
  45. Cruz, R.M.; Sabourin, R.; Cavalcanti, G.D. Dynamic classifier selection: Recent advances and perspectives. Inf. Fusion. 2018, 41, 195–216. [Google Scholar] [CrossRef]
  46. Arji, G.; Ahmadi, H.; Nilashi, M.; Rashid, T.A.; Hassan Ahmed, O.; Aljojo, N.; Zainol, A. Fuzzy logic approach for infectious disease diagnosis: A methodical evaluation, literature and classification. Biocybern. Biomed. Eng. 2019, 39, 937–955. [Google Scholar] [CrossRef]
  47. Al-Dmour, J.A.; Sagahyroon, A.; Al-Ali, A.R.; Abusnana, S. A fuzzy logic-based warning system for patients classification. Health Inform. J. 2019, 25, 1004–1024. [Google Scholar] [CrossRef] [PubMed]
  48. Alcan, V.; Canal, M.R.; Zinnuroğlu, M. Using fuzzy logic for diagnosis and classification of spasticity. Turk. J. Med. Sci. 2017, 47, 148–160. [Google Scholar] [CrossRef] [PubMed]
  49. Ghorbani, S.; Barari, M.; Hoseini, M. Presenting a new method to improve the detection of micro-seismic events. Environ. Monit. Assess. 2018, 190, 464. [Google Scholar] [CrossRef] [PubMed]
  50. Azarafza, M.; Akgün, H.; Feizi-Derakhshi, M.R.; Azarafza, M.; Rahnamarad, J.; Derakhshani, R. Discontinuous rock slope stability analysis under blocky structural sliding by fuzzy key-block analysis method. Heliyon 2020, 6, e03907. [Google Scholar] [CrossRef]
  51. Keramitsoglou, I.; Kiranoudis, C.T.; Maiheu, B.; De Ridder, K.; Daglis, I.A.; Manunta, P.; Paganini, M. Heat wave hazard classification and risk assessment using artificial intelligence fuzzy logic. Environ. Monit. Assess. 2013, 185, 8239–8258. [Google Scholar] [CrossRef]
  52. Sathyendranath, S.; Brewin, R.J.W.; Brockmann, C.; Brotas, V.; Calton, B.; Chuprin, A.; Cipollini, P.; Couto, A.B.; Dingle, J.; Doerffer, R.; et al. An Ocean-Colour Time Series for Use in Climate Studies: The Experience of the Ocean-Colour Climate Change Initiative (OC-CCI). Sensors 2019, 19, 4285. [Google Scholar] [CrossRef]
  53. Ordoñez-Avila, R.; Salgado Reyes, N.; Meza, J.; Ventura, S. Data mining techniques for predicting teacher evaluation in higher education: A systematic literature review. Heliyon 2023, 9, e13939. [Google Scholar] [CrossRef]
  54. Khumaidi, A.; Purwanto, Y.A.; Sukoco, H.; Wijaya, S.H. Using Fuzzy Logic to Increase Accuracy in Mango Maturity Index Classification: Approach for Developing a Portable Near-Infrared Spectroscopy Device. Sensors 2022, 22, 9704. [Google Scholar] [CrossRef]
  55. Duch, W. Concept Representation and the Geometric Model of Mind. Stud. Log. Gramm. Rhetor. 2022, 67, 151–167. [Google Scholar] [CrossRef]
  56. Amor, N.; Noman, M.T.; Petru, M. Classification of Textile Polymer Composites: Recent Trends and Challenges. Polymers 2021, 13, 2592. [Google Scholar] [CrossRef] [PubMed]
  57. Noman, M.T.; Amor, N.; Ali, A.; Petrik, S.; Coufal, R.; Adach, K.; Fijalkowski, M. Aerogels for Biomedical, Energy and Sensing Applications. Gels 2021, 7, 264. [Google Scholar] [CrossRef] [PubMed]
  58. Pinto, A.; Ribeiro, R.A.; Nunes, I.L. Fuzzy approach for reducing subjectivity in estimating occupational accident severity. Accid. Anal. Prev. 2012, 45, 281–290. [Google Scholar] [CrossRef]
  59. Broekhuizen, H.; Groothuis-Oudshoorn, C.G.; van Til, J.A.; Hummel, J.M.; IJzerman, M.J. A review and classification of approaches for dealing with uncertainty in multi-criteria decision analysis for healthcare decisions. Pharmacoeconomics 2015, 33, 445–455. [Google Scholar] [CrossRef] [PubMed]
  60. Prokopowicz, P. Adaptation of Rules in the Fuzzy Control System Using the Arithmetic of Ordered Fuzzy Numbers. In Artificial Intelligence and Soft Computing—ICAISC 2008. ICAISC 2008, Lecture Notes in Computer Science; Rutkowski, L., Tadeusiewicz, R., Zadeh, L.A., Zurada, J.M., Eds.; Springer: Berlin/Heidelberg, Germany, 2008; p. 5097. [Google Scholar] [CrossRef]
  61. Prokopowicz, P. Analysis of the changes in processes using the kosinski’s fuzzy numbers. In Proceedings of the 2016 Federated Conference on Computer Science and Information Systems, FedCSIS 20916, Gdansk, Poland, 11–14 September2016; pp. 121–128. [Google Scholar]
  62. Price, A.R. The marine food chain in relation to biodiversity. Sci. World J. 2001, 1, 579–587. [Google Scholar] [CrossRef] [PubMed]
  63. Park, E.Y. Recent progress in microbial cultivation techniques. Adv. Biochem. Eng. Biotechnol. 2004, 90, 1–33. [Google Scholar] [CrossRef]
  64. Zadeh, L.A. Fuzzy sets as a basis for a theory of possibility. Fuzzy. Sets Syst. 1978, 1, 3–28. [Google Scholar] [CrossRef]
  65. Zadeh, L.A. Fuzzy sets. Inf. Control. 1965, 8, 338–353. [Google Scholar] [CrossRef]
Applsci 13 06396 g001 550
Figure 1. The example of the input variable.
Figure 1. The example of the input variable.
Applsci 13 06396 g001
Applsci 13 06396 g002 550
Figure 2. The structure of the fuzzy system.
Figure 2. The structure of the fuzzy system.
Applsci 13 06396 g002
Applsci 13 06396 g003 550
Figure 3. SWOT analysis of the proposed computational solution.
Figure 3. SWOT analysis of the proposed computational solution.
Applsci 13 06396 g003
Table
Table 1. Basic clusters of reference data—start point for grouping data into fuzzy sets.
Table 1. Basic clusters of reference data—start point for grouping data into fuzzy sets.
Input No.Input DescriptionValues and Their Occurrences Form the Initial Clusters
1Surface roughnessValues6.31020
No. of occur.12323
2Shape of milling cutterValues10162025406380100125250
No. of occur.45106224644
3Milling tool lengthValues32354050556392100104125165175
No. of occur.122828644424
4Average cutting speedValues6067.5747582119120127.5148156180210222254270318328375
No. of occur.222222424225254221
5Cutting depthValues2.54566.25899.5101212.515161920253037.575
No. of occur.1141321161271131722
6Milling widthValues0.150.250.30.51
No. of occur.11711117
7Cutting feedValues1781871982162252292402492702852862893003183213263383613723854144455095967227957961032
No. of occur.2622121212211112212112212112
8Operating costValues751201962502603153504104554865957307607651700
No. of occur.412444422442424
Table
Table 2. Detailed data of input variables for variant 2.
Table 2. Detailed data of input variables for variant 2.
Input No.Input DescriptionNumeric Values and Corresponding Linguistic Values
1Surface roughness6.31020
Fuzzy sets—coreslowlowhigh
2Shape of milling cutter10202580100250
Fuzzy setlowlowmediummediumhighhigh
3Milling tool length32505592100125165175
Fuzzy sets—coreslowlowmediummediumhighhighvery highvery high
4Average cutting speed60120127.5210222270318375
Fuzzy sets—coreslowlowmediummediumhighhighvery highvery high
5Cutting depth2.58912.515202575
Fuzzy sets—coreslowlowmediummediumhighhighvery highvery high
6Milling width0.150.250.30.51
Fuzzy sets—coreslowlowmediummediumhigh
7Cutting feed1782492703383617227951032
Fuzzy sets—coreslowlowmediummediumhighhighvery highvery high
8Operating cost752603154554867651700
Fuzzy sets—coreslowlowmediummediumhighhighvery high
Table
Table 3. Results for Variant 1.
Table 3. Results for Variant 1.
Tool 1Tool 2Tool 3Tool 4Tool 5Tool 6Tool 7Tool 8Tool 9Tool 10Tool 11Tool 12Tool 13Tool 14Tool 15Tool 16Tool 17Expected Recognition
No. of input data10.990.7050.645N0.675NNNNNNNNNNNNTool 1
20.990.990.645N0.675NNNNNNNNNNNNTool 1
30.990.7050.645N0.7050.705NNNNNNNNNNNTool 1
40.990.990.645N0.990.99NNNNNNNNNNNTool 1
5N0.990.580.510.675NNNNNNNNNNNNTool 2
60.990.990.580.510.675NNNNNNNNNNNNTool 2
70.990.990.580.510.990.99NNNNNNNNNNNTool 2
8N0.99NNNNNNNNNNNNNNNTool 2
9NN0.99NNNNNNNNNNNNNNTool 3
10NN0.99NNNNNNNNNNNNNNTool 3
11NN0.99NNNNNNNNNNNNNNTool 3
12NN0.99NNNNNNNNNNNNNNTool 3
13NNN0.99NNNNNNNNNNNNNTool 4
14NNN0.99NNNNNNNNNNNNNTool 4
15NNN0.99NNNNNNNNNNNNNTool 4
16NNN0.99NNNNNNNNNNNNNTool 4
17NNNN0.99NNNNNNNNNNNNTool 5
18NNNN0.99NNNNNNNNNNNNTool 5
190.990.99NN0.990.99NNNNNNNNNNNTool 5
200.990.99NN0.990.99NNNNNNNNNNNTool 5
210.990.99NN0.990.99NNNNNNNNNNNTool 6
220.990.99NN0.990.99NNNNNNNNNNNTool 6
23NNNNNN0.990.99NNNNNNNNNTool 7
24NNNNNN0.990.99NNNNNNNNNTool 8
25NNNNNNNN0.99NNNNNNNNTool 9
26NNNNNNNNN0.990.99NNNNNNTool 10
27NNNNNNNNN0.5250.525NNNNNNTool 11
28NNNNNNNNNN0.525NNNNNNTool 11
29NNNNNNNNNNN0.990.99N0.5450.545NTool 12
30NNNNNNNNNNN0.990.99N0.5450.545NTool 12
31NNNNNNNNNNN0.990.99NN0.545NTool 12
32NNNNNNNNNNN0.990.99NN0.545NTool 12
33NNNNNNNNNNN0.990.99NNNNTool 13
34NNNNNNNNNNN0.990.99NNNNTool 13
35NNNNNNNNNNN0.990.99NNNNTool 13
36NNNNNNNNNNN0.990.99NNNNTool 13
37NNNNNNNNNNNNN0.99NNNTool 14
38NNNNNNNNNNNNN0.99NNNTool 14
39NNNNNNNNNNNNN0.99NNNTool 14
40NNNNNNNNNNNNN0.99NNNTool 14
41NNNNNNNNNNNNNN0.990.99NTool 15
42NNNNNNNNNNNNNN0.990.515NTool 15
43NNNNNNNNNNN0.630.63N0.990.99NTool 16
44NNNNNNNNNNN0.630.63N0.990.99NTool 16
45NNNNNNNNNNN0.630.63NN0.99NTool 16
46NNNNNNNNNNN0.630.63NN0.99NTool 16
47NNNNNNNNNNNNNNNN0.99Tool 17
Tool 1Tool 2Tool 3Tool 4Tool 5Tool 6Tool 7Tool 8Tool 9Tool 10Tool 11Tool 12Tool 13Tool 14Tool 15Tool 16Tool 17
Table
Table 4. Results for Variant 2.
Table 4. Results for Variant 2.
Tool 1Tool 2Tool 3Tool 4Tool 5Tool 6Tool 7Tool 8Tool 9Tool 10Tool 11Tool 12Tool 13Tool 14Tool 15Tool 16Tool 17Expected Recognition
No. of input data10.99N0.645NNNNNNNNNNNNNNTool 1
20.99NNNNNNNNNNNNNNNNTool 1
30.99NNNN0.735NNNNNNNNNNNTool 1
40.99NNNNNNNNNNNNNNNNTool 1
5N0.99NNNNNNNNNNNNNNNTool 2
6N0.99NNNNNNNNNNNNNNNTool 2
7N0.99NNNNNNNNNNNNNNNTool 2
8N0.99NNNNNNNNNNNNNNNTool 2
9NN0.99NNNNNNNNNNNNNNTool 3
10NN0.99NNNNNNNNNNNNNNTool 3
11NN0.99NNNNNNNNNNNNNNTool 3
12NN0.99NNNNNNNNNNNNNNTool 3
13NNN0.99NNNNNNNNNNNNNTool 4
14NNN0.99NNNNNNNNNNNNNTool 4
15NNN0.99NNNNNNNNNNNNNTool 4
16NNN0.99NNNNNNNNNNNNNTool 4
17NNNN0.99NNNNNNNNNNNNTool 5
18NNNN0.99NNNNNNNNNNNNTool 5
19NNNN0.99NNNNNNNNNNNNTool 5
20NNNN0.99NNNNNNNNNNNNTool 5
210.8350.545NNN0.99NNNNNNNNNNNTool 6
220.7550.595NNN0.99NNNNNNNNNNNTool 6
23NNNNNN0.99NNNNNNNNNNTool 7
24NNNNNN0.7950.99NNNNNNNNNTool 8
25NNNNNNNN0.99NNNNNNNNTool 9
26NNNNNNNNN0.99NNNNNNNTool 10
27NNNNNNNNNN0.99NNNNNNTool 11
28NNNNNNNNNN0.99NNNNNNTool 11
29NNNNNNNNNNN0.990.815NNNNTool 12
30NNNNNNNNN0.595N0.990.815NNNNTool 12
31NNNNNNNNNNN0.990.815NNNNTool 12
32NNNNNNNNNNN0.990.815NNNNTool 12
33NNNNNNNNNNNN0.99NNNNTool 13
34NNNNNNNNNNNN0.99NNNNTool 13
35NNNNNNNNNNNN0.99NNNNTool 13
36NNNNNNNNNNNN0.99NNNNTool 13
37NNNNNNNNNNNNN0.99NNNTool 14
38NNNNNNNNNNNNN0.99NNNTool 14
39NNNNNNNNNNNNN0.99NNNTool 14
40NNNNNNNNNNNNN0.99NNNTool 14
41NNNNNNNNNNNNNN0.99NNTool 15
42NNNNNNNNNNNNNN0.990.515NTool 15
43NNNNNNNNNNN0.5950.595N0.5950.99NTool 16
44NNNNNNNNN0.595N0.5950.595NN0.99NTool 16
45NNNNNNNNNNN0.5950.595NN0.99NTool 16
46NNNNNNNNNNN0.5850.585NN0.99NTool 16
47NNNNNNNNNNNNNNNN0.99Tool 17
Tool 1Tool 2Tool 3Tool 4Tool 5Tool 6Tool 7Tool 8Tool 9Tool 10Tool 11Tool 12Tool 13Tool 14Tool 15Tool 16Tool 17
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Rojek, I.; Prokopowicz, P.; Kotlarz, P.; Mikołajewski, D. Extended Fuzzy-Based Models of Production Data Analysis within AI-Based Industry 4.0 Paradigm. Appl. Sci. 2023, 13, 6396. https://doi.org/10.3390/app13116396

AMA Style

Rojek I, Prokopowicz P, Kotlarz P, Mikołajewski D. Extended Fuzzy-Based Models of Production Data Analysis within AI-Based Industry 4.0 Paradigm. Applied Sciences. 2023; 13(11):6396. https://doi.org/10.3390/app13116396

Chicago/Turabian Style

Rojek, Izabela, Piotr Prokopowicz, Piotr Kotlarz, and Dariusz Mikołajewski. 2023. "Extended Fuzzy-Based Models of Production Data Analysis within AI-Based Industry 4.0 Paradigm" Applied Sciences 13, no. 11: 6396. https://doi.org/10.3390/app13116396

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop