Advancing Industrial Automation: Highlights from Recent Research

1. Introduction

Industries today face several challenges due to rapidly evolving market demands, stringent environmental regulations, and the growing integration of renewable energy sources. Addressing these issues requires the adoption of innovative technologies that foster adaptability, operational efficiency, and resilience [1]. Within this context, key research areas include robust control strategies, advanced fault detection systems, dynamic optimization techniques, and the development of sophisticated real-time monitoring models. These advancements, often based on learning algorithms and real-time problem-solving approaches, hold significant promise for transforming industrial processes. This Special Issue, “Recent Advances in Process Control and Monitoring”, marks the 10th anniversary of Processes by showing a curated selection of contributions. These works highlight notable progress in robust control [2,3,4], fault detection [5,6,7], dynamic optimization [2,8], and the application of cutting-edge monitoring and detection models for real-time solutions [9,10,11].

2. Overview of the Contributions

The selected papers cover a wide range of topics, highlighting both the diversity and depth of contemporary research in process automation.

Keßler and Kienle [2] propose a robust optimization approach for operating conditions in a multistage reactor producing green methanol, explicitly accounting for fluctuations in hydrogen supply. The integration of renewable energy sources poses significant challenges, which have been addressed by various researchers through solutions such as intermediate hydrogen storage [12] and advanced control strategies [13,14]. Keßler and Kienle [2] identify that robust designs relying solely on steady-state optimization across different process scenarios often result in conservative operations that neglect dynamic behavior, potentially causing infeasible conditions. To overcome this, they develop a feed-forward control scheme that employs dynamic optimization over a short time horizon to adjust feed distribution and shell temperatures, effectively mitigating transient effects.

Accurately describing real physical phenomena is a fundamental challenge in process control, especially when dealing with dynamic complex systems exhibiting nonlinear behaviors [15]. Nonlinear descriptor systems offer a powerful and general framework to model such systems, capturing both differential and algebraic relationships that arise in many practical applications [16]. Lin et al. [3] introduce an observer-based proportional–derivative fuzzy controller for uncertain discrete-time nonlinear descriptor systems, which extends the capability of robust control methods to more general nonlinear systems. By employing the Takagi–Sugeno fuzzy framework and parallel distributed compensation, the method avoids issues such as noncausal behavior and impulse responses. Furthermore, it incorporates a fuzzy observer for unmeasured states and uses advanced techniques such as singular value decomposition to enhance stability analysis while reducing conservatism.

Digital hydraulic technology aligns closely with the global industry’s drive for sustainability and intelligent systems [17]. This connection is highlighted by Mitov et al. [4], who address control challenges in hydraulic drives and introduce a promising new approach. The authors compare two advanced control strategies, Linear–Quadratic Regulator and H-infinity controllers, applied to an axial piston pump. Unlike traditional hydro-mechanical controllers [18], their study employs an electro-hydraulic proportional valve controlled by a microcontroller capable of embedding various algorithms. Through experimental validation, both controllers demonstrate robust stability under significant uncertainties, even though the H-infinity controller proves superior, attenuating disturbances.

The other important aspect in process engineering is fault detection, and the use of digital twins can be a valid support [19]. Gómez-Coronel et al. [5] develop the digital twin of an experimental hydraulic system designed for single- and multi-leak detection. The system integrates real-time data, a hydraulic model, and a genetic algorithm for diagnosis, along with a remote operator interface for system control.

A fault detection and isolation method in a circulating fluidized bed boiler is investigated by Kim et al. [6]. Their study proposes a novel approach based on shared nearest neighbor (SNN) analysis. Unlike traditional k-nearest neighbor (kNN) methods, the SNN-based method improves fault detection accuracy by weighting distances according to shared neighbor information. This approach is independent of data distribution and avoids the smearing effects often seen in PCA and ICA techniques. The SNN method’s application aligns with general advancements in kNN methodologies, as highlighted in the comprehensive review by Halder et al. [20]. The proposed approach is validated through two case studies and a real boiler failure, demonstrating earlier fault detection and more effective identification of fault variables compared to conventional methods.

Another contribution related to fault detection focuses on the common rail system of diesel engines, which, due to its complexity, presents challenges in diagnosing and maintaining optimal pressure control, which is essential for its efficient operation [21]. Bacci di Capaci and Pannocchia [7] propose a data-driven methodology, using a Hammerstein model that combines a nonlinear component for control valve behavior with an extended linear model for process dynamics. This approach effectively identifies and quantifies oscillations caused by various fault sources. The methodology’s effectiveness is demonstrated through selected real case studies.

Regarding optimization, in Jitchaiyapoom et al. [8], the production of biodiesel is considered and used as a case study for developing a procedure to predict production capacity and optimize the process. In the paper, an advanced approach using few-shot learning and LSTM models is employed to address challenges associated with feed uncertainty and limited data. The choice of LSTM is driven by its ability to handle complex relationships, such as those found in industrial processes, by capturing temporal dynamics and long-term dependencies [22]. The methodology not only demonstrates significant improvements in predictive accuracy and process optimization for glycerin production but also highlights its potential adaptability to similar industrial scenarios.

Real-time monitoring is addressed in the work of Xiang et al. [9], where a detection algorithm for identifying solidification layers of scattered coal during transportation is proposed. The procedure is based on an improved YOLOv8 model, integrated with fluorescence detection, and it successfully identifies coal solidification layers during transportation, enhancing detection accuracy and demonstrating potential for practical applications in monitoring and managing coal transport.

Advancements in sensor technology have enhanced the study of complex systems such as machining processes. Chatter, due to self-excited tool vibrations, is a major problem as it can lead to poor surface finish and tool/machine damage [23]. In Wu et al. [10], a DAE-BiLSTM model is developed to improve tool chatter recognition in turning processes and enhance workpiece quality and production efficiency. The model combines DAE for dimensionality reduction with BiLSTM for bidirectional signal processing, addressing the limitations of previous methods such as SVMs and unidirectional LSTMs. Experiments on cylindrical cutting demonstrate higher classification accuracy when compared to traditional models. It is important to highlight that the approach has been successfully implemented in industrial settings, with future work focusing on multi-sensor fusion and closed-loop control strategies.

Asset integrity management is crucial for petrochemical plants, focusing on maintaining the reliability, availability, maintainability, and safety of assets while minimizing risks and costs [24]. A systematic framework for managing risks and ensuring asset integrity in real time is proposed in Han et al. [11], with application to petrochemical plants. By integrating dynamic risk-based inspection (RBI) with integrity operating windows (IOWs), the authors propose a framework that allows for real-time monitoring and adjustment of operational parameters to ensure safety and reliability. The dynamic RBI methodology enables continuous risk assessment, adapting to changing conditions and providing timely insights for inspections, while IOWs establish clear boundaries for safe operation, reducing the likelihood of equipment failure. This integrated strategy not only optimizes inspection schedules and reduces costs but also improves overall plant performance by aligning operations with risk tolerance levels.

3. Conclusions

By emphasizing the importance of advanced control and monitoring systems in the next evolution of industrial processes, this Special Issue promotes the future of sustainable, intelligent processing through a synergy of control theory, digitalization, and eco-efficiency. The diversity of topics—ranging from robust optimization in methanol production to advanced fault detection systems, and from nonlinear control frameworks to real-time risk management—illustrates the multidimensional challenges and opportunities in process automation. Key contributions demonstrate the practical relevance and transformative potential of these advancements, reinforcing the need for continuous innovation to foster resilience, sustainability, and operational excellence in addressing the evolving demands of modern industry.