IET Collaborative Intelligent Manufacturing最新文献

The reentrant flowshop scheduling problems (RFSP) are ubiquitous in high-tech industries such as semiconductor manufacturing and liquid crystal display (LCD) production. Given the complexity of RFSP, it is significant to improve the production efficiency using effective intelligent optimisation techniques. In this study, four meta-heuristics assisted by two reinforcement learning (RL) algorithms are proposed to minimise the maximum completion time (makespan) for RFSP. First, a mathematical model for RFSP is established. Second, four meta-heuristics are improved. The Nawaz–Enscore–Ham (NEH) heuristic is utilised for population initialisation. Based on the problem characteristics, we design six local search operators, which are integrated into the four meta-heuristics. Third, two RL algorithms, Q-learning and state–action-reward–state–action (SARSA), are employed to select the appropriate local search operator during iterations to enhance the convergence in a local space. Finally, the results of solving 72 instances indicate that the proposed algorithms perform effectively. The RL-guided local search can significantly enhance the overall performance of the four meta-heuristics. In particular, the artificial bee colony algorithm (ABC) combined with SARSA-guided local search yields the highest performance.

Cognitive workload (CWL) assessment has gained increasing importance in Industry 4.0 and 5.0 settings where human–machine interactions are becoming more complex. Despite growing attention, a comprehensive CWL assessment that integrates methodologies, technologies and case studies is still lacking. This study reviews 69 articles related to the CWL assessment, selected from the Scopus database. The review identifies five primary methodologies for the CWL assessment: physiological measures, subjective evaluation (e.g., NASA-TLX), performance evaluation, cognitive load models and multimodal approaches. The analysis shows an increasing trend towards multimodal approaches that combine subjective assessment methods with physiological measures obtained from electroencephalography, eye tracking and heart rate monitoring devices. Additionally, emerging technologies such as advanced sensors and specialised equipment are increasingly considered in case studies that address the CWL assessment in current work environments. Results reveal significant advancements in physiological and multimodal assessment methods, particularly emphasising real-time monitoring capabilities and context-specific applications. Case studies underscore the key role of CWL management in assembly, maintenance and construction tasks, demonstrating its impact on performance, safety and adaptability in dynamic environments. This review establishes a framework for advancing CWL research by addressing methodological limitations and proposing future research directions, including the development of personalised, adaptive systems for real-time workload management.

Multiagent cooperative control enhances system efficiency through the facilitation of distributed collaboration, demonstrating significant applications in intelligent manufacturing. As a fundamental issue of cooperative control, multiagent consensus has been implemented extensively in numerous domains. Therefore, this paper studies the asymptotic consensus issue of a nonlinear system under switching topologies. The changeable topological structures hinder the system's ability to stabilise or require a substantial amount of time for stabilisation. To address this issue, we have incorporated topological information into the traditional Riccati equation. Subsequently, a topology-based dynamic event-triggered mechanism is presented by introducing an internal dynamic variable based on the solution of the Riccati equation. Furthermore, this research proposes a novel control protocol that utilises the full information of the switching topologies. This protocol contains a changeable control gain, which allows for the adjustment of the control law in response to the communication topology. Then, the Lyapunov stability theory guarantees that the nonlinear system reaches an asymptotic consensus under the proposed control law. This study also proves that the system does not exhibit Zeno behaviour. Ultimately, the simulation results confirm the viability of the control protocol.

Employing advanced technology in manufacturing will improve productivity and resource efficiency as well as reduce long term operating cost. Storage drive manufacturers focus on the advanced technology adoption as a way to reduce their operating cost. Despite many benefits of Industry 4.0, integration and implementation are not easily achievable. This research aims to identify the barriers of Industry 4.0 implementation in storage drive industry in the context of hard disk drive (HDD) and solid-state drive (SSD) manufacturing and to suggest guidelines to overcome the barriers. Starting from extensive literature review, followed by expert justification, 15 barriers for the implementation of Industry 4.0 in storage drive manufacturing industry were identified. The fuzzy AHP approach was used to prioritise the barriers. The study found that for both HDD and SSD industries, ‘economic’ criteria is the priority followed by ‘technology’ and ‘organisation’ criteria. The result suggests that decision makers should find avenues to overcome these three barriers before implementing Industry 4.0 in the storage drive manufacturing industry. Getting sufficient financial fund for capital investment, being technological-oriented organisation and getting strong management support for new technology are the main guideline for the industry. The research methodology in this study could be applied in other manufacturing industries to identify barriers and plan for strategic actions before the intelligent manufacturing implementation.

Detection and identification of defects in manufactured products, a task related to the basic requirements of quality management systems. By moving to higher levels, under the right conditions, these defects can be avoided, for example, by preventing manufacturing defects from occurring. Quality control and monitoring of welds are closely linked to the requirements of Industry 4.0. In the case of welding processes, quality assurance is a multifaceted area, including not only the analysis of input parameters but also the quality of the weld surface. By superimposing the point clouds of the parts under test, geometric features are generated to the initial manufacturing parameters to help increase manufacturing efficiency. In our work, the information data recorded by the data acquisition framework, which is captured during the welding process, is integrated with the outputs of the point cloud characteristics of the examined by the structured light scanning technology, as well as the value of the seam width magnitude extracted by the image recognition algorithms. This contributes to the possibilities of broadening the seam detection processes.

Efficient sequencing of processes and resource allocation are critical in production planning scenarios, such as manufacturing workshops and construction projects, to enhance efficiency and reduce operational costs. Resource allocation in such environments is often challenged by temporal constraints, process interdependencies, and resource limitations, which complicate scheduling and increase the risk of delays. This study presents a multi-agent-based simulation system to address these challenges. A scheduling optimisation model is developed to simulate and optimise resource allocation in complex processes with network structures and temporal constraints. The primary objective is to minimise production completion time while ensuring effective resource allocation. Additionally, an adaptive, partially distributed Agent-Based Modelling and Simulation framework is proposed to simulate the execution logic of real-world processes, integrating key factors such as resource limitations, process interdependencies, and real-time decision-making. A priority-based genetic algorithm is also designed and embedded into the multi-agent system to further optimise process sequencing and resource distribution. Simulation experiments across varying case scales validate the model and algorithm. This study highlights the potential of agent-based simulation for solving complex engineering challenges and provides new insights for addressing resource allocation problems in network-structured, time-constrained environments.

Complex industrial processes usually exhibit multimode characteristics, meaning that statistical features of process data, such as mean, variance, and correlation, vary across different modes. Extracting critical information from these distinct modes can significantly enhance the accuracy and robustness of data-driven models in process monitoring, condition evaluation, and quality improvement. Consequently, the multimode identification of industrial data becomes a paramount concern in data-driven modelling. However, existing methods for multimode identification require prior knowledge to predetermine the number of modes and struggle to describe the similarity between high-dimensional samples effectively. To address this issue, this study introduces an automatic multimode identification method based on complex network community detection. In this approach, each data sample is considered as a node, and manifold similarity is calculated to construct the complex network model. The method leverages weighted geodesic distances to capture the data's manifold structure and potential density, enabling better distinction between high-dimensional samples in different modes. The greedy search algorithm with modularity maximisation is employed to partition nodes into modes without manual selection of the number of modes. Furthermore, a node degree-based indicator is developed for online mode monitoring. Experimental studies on two examples demonstrate the effectiveness of the proposed method in uncovering multimode characteristics of complex industrial processes, highlighting its promising application potential.

The deep integration of digital twins (DT) and agents is expected to open up new collaborative machine intelligence solutions. A new concept, namely, agent-based digital twins (ADT), is proposed to establish a novel machine intelligence framework with automatic perception, self-evolution and autonomous collaboration.

Machine breakdowns are alarming threats to factories. They can substantially decrease productivity, cause financial losses, and create unsafe work environments for operators. Early detection of system anomalies is crucial to prevent and fix machine threats before they become fatalities. With the advent of digitalisation and smart manufacturing, various artificial intelligence (AI) and machine learning (ML) techniques contribute to implementing efficient anomaly detection systems with more accurate results. In this research, the design of an experimental anomaly detection platform (ADP) was suggested for a conveyor motor system. The ADP analyses time-series conveyor motor parameters and accurately classifies whether they would cause a faulty system. The authors build a classification ML model using dendritic gated neural networks (DGNN) to achieve better accuracy. Dendritic Neural Networks are highly immune to forgetting, contributing to better performance than regular artificial neural networks (ANNs) using backpropagation. The ADP also includes a fault detection platform section for the conveyor motors' time-series parameters with recurrent neural networks (RNN) ML regression models to predict motor sensor values. When training ML classification models, the predicted time-series parameters can also serve data augmentation purposes. This regression section contributes to a more robust and double-layered ADP, preventing threats from the time-series inputs to the output classification level. The ADP solution suits small traditional factories with limited historical data records. The experimental results show the benefits of using our ADP built on the DGNN ML model over several classification models such as ANN, convolutional neural network (CNN), and support vector machine (SVM).

First carbon hit rate (FCHR) is an essential indicator of steel converter smelting, reflecting the proportion of steel tapping completed without additional oxygen blowing. However, significant data loss has occurred due to equipment ageing and worker operations, resulting in difficulties in analysing the FCHR. This paper uses mechanism analysis and feature screening to determine the model input, predicts and fills in abnormal data through ensemble learning, and then optimises it through data transformation. Finally, the Stacking model predicts the FCHR, with a training accuracy of up to 94.5% and a test set accuracy of 90.5%. In addition, the authors also conducted a predictive study on oxygen consumption, and the hit rate performed well under different error thresholds, with a maximum of 97.9%. These results provide powerful decision support for steel production and effectively overcome the challenges of data missingness.