Journal of Manufacturing Systems最新文献

The interaction between customer demands and manufacturing paradigms is becoming increasingly apparent. As the demand for personalized products grows, the manufacturing industry is evolving towards a socialized manufacturing paradigm. This shift makes the manufacturing system more unstable and complex, necessitating organization of production through a socialized resource service platform. Unlike traditional systems, emerging distributed smart manufacturing system (DSMS) face challenges of trusted collaborative operation and real-time optimal-state control in dynamic operational environments. To overcome these challenges, we propose a trusted optimal-state synchronized control (OSsC) approach suitable for DSMS to ensure optimal operation under dynamic customer demands. This paper introduces a digital twin and blockchain-based trusted optimal-state control framework for reliable decision-making, integrating OSsC approach into a trusted virtual layer to achieve real-time optimal target setting. We also propose a blockchain-based mechanism for trusted synchronized operation in open production logistics, enhancing cross-domain trust and intelligent selection of units under dynamic interruptions. Furthermore, we apply the analytical target cascading method for multi-objective synchronized optimization decision model in complex systems. A case study in the air conditioning manufacturing industry demonstrates the effectiveness of the framework, mechanism, and algorithm in enhancing reliability and reducing costs in dynamic environments, providing valuable insights for the optimization design and reliable operation of future manufacturing systems.

In the assembly sequence planning (ASP) of aviation products, recalibration of components or sufficient space to assemble subsequent components are critical factors for ensuring product quality. To address this need, a fine-grained ASP (FASP) is defined to take assembly operations as units to plan sequences. Lots of operations have complex sequence constraints that are attended unequally in the FASP. A method based on knowledge graph (KG) and deep reinforcement learning is proposed to plan assembly operations. Firstly, continuous and discrete procedures are defined, and a quantitative characterization method is presented to deduce complex constraints objectively. Then, a dynamic KG is designed to establish and update the information model mainly composed of constraints. Finally, a labeled degree centrality algorithm (LDCA) considers edge labels to minimize the number of assembly tool changes and assembly direction changes for sequences. An improved deep Q-network (IDQN) introduces a convolutional layer to extract local features of technical requirements for planning procedures more efficiently. A helicopter structure assembly is used to verify the effectiveness of the proposed method. The improved algorithms have better performance in solving speed, sequence quality, and convergence speed than ordinary ASP methods, respectively. The fine-grained assembly sequence is more reasonable and feasible by comparing it with the ordinary sequence.

The crude protein purification automated workstation has recently resolved the bottlenecks induced by manual operations, paving the way for high-throughput protein biomanufacturing. However, its three interacted constraints consisting of batch processing machines, limited buffer, and transportation present challenges for systematic scheduling. Here, we develop a triply-constrained flow shop model, enabling optimization in scheduling the crude protein purification automated workstation. A batching genetic algorithm is designed, where the flexible decoding resolves contradictions between the triple constraints, and the hybrid population initialization enhances performance by incorporating flow-shop heuristic and batching branch-and-bound. Computational experiments are conducted on 27 instances of varying problem scales ranging from small to large, demonstrating a notable 9.18 % reduction in makespan and enhanced stability when compared to three advanced meta-heuristics. Furthermore, the mechanism of how batching settings, including capacities and layouts, impact the makespan is revealed, offering managerial insights. This marks the first demonstration of modeling and scheduling crude protein purification automated workstations, signifying a significant advancement in biomanufacturing systems.

Multi-product centralized delivery and kitting assembly present significant challenges to hierarchical co-processing in multi-stage manufacturing systems. The combinations of priority dispatching rules at each level are transiently adaptive, and the performance in online scheduling deteriorates rapidly with changing environment. This paper investigates the selection of rule combinations for sustained high-performance responsive scheduling in two-stage flexible assembly flow shop scheduling problem with asynchronous execution and complex decision correlation. A Multi-Level Action Coupling Deep Q-Network (MALC-DQN) approach is proposed for adaptive integrated scheduling in hybrid processing and assembly shops. Firstly, the problem is skillfully established as an event-triggered integrated decision markov decision process. The prioritized batch experience replay mechanism is employed to retain the complete correlation information of key decision sequences. Then, coupling and sequence feature extraction modules are developed to enhance the agent’s ability to perceive execution process and the environment. Furthermore, the multi-level wait-limit mechanism and efficient action filtering mechanism are designed to mitigate ineffective waiting waste and action space explosion during learning. Finally, a series of sophisticated experiments are conducted to validate the effectiveness of the proposed methodology. In 20 actual instances of different sizes, MLAC-DQN outperformed its closest competitor, with a 26.6% improvement in average tardiness. Moreover, extraordinary robustness is demonstrated in 16 sets of experiments involving different configurations of resources, orders, and arrival concentration levels.

Human-centricity, sustainability, and resilience are becoming core values in modern manufacturing, with human–robot collaboration (HRC) in high demand for flexible automation. However, human–robotic swarms are typically designed to target one specific procedure and cannot fully share their autonomy. The Metaverse, characterized by socialized avatars in a virtual-physical fused world, holds the promise of Proactive HRC. In line with this evolutionary roadmap, this paper presents a futuristic perspective on the industrial Metaverse for Proactive HRC and identifies its six embodiments. A representative universe that supports online and offline human users/operators in the design, machining, and maintenance of aeroengine turbine blades is introduced to spark and accelerate future implementation of the industrial Metaverse for Proactive HRC. The current challenges and future opportunities of this paradigm are also highlighted. It is hoped that this work can attract further investigation and discussions, providing useful insights to both academic and industrial practitioners in smart manufacturing.

The high-stake nature of most manufacturing processes empowers the importance of real-time quality control and assurance. In the event of a failure in production, a decision-making process can be time-consuming for the human and prevent timely actions. The agility can be boosted with a decision-support system based on artificial intelligence. Particularly, multi-objective process optimization can be employed to select the optimal control settings in real-time, and thus enhance relevant key performance indicators, concurrently. However, process optimization in manufacturing scenarios has never been an easy task, due to the complexity, non-convexity, and non-linearity of dependences among process parameters and physical constraints typical for strict production procedures. Precise and high-performative digital replicas of physical systems are required to simulate different scenarios. Physical models are computationally demanding for real-time applications and are usually hard to develop. In that light, this paper brings a novel solution based on multi-objective evolutionary optimization coupled with process surrogate data-driven models, in charge of predicting the relevant process responses. Based on process and quality parameters being streamed from the production plant in real-time, the optimizer can act in timely critical and quality-threatening situations and generate immediate corrective actions. The multi-regime operation of the plant and design space dimensionality can impact the convergence rate and add to execution time. Therefore, production regimes recognition and greedy search of suffix tree-based models of the process have been engaged, aiding in a better-focused and faster space search at an early phase of the algorithm run. Beyond simply reviewing the outputs, the user can leave feedback, which is utilized by the optimizer’s reinforcement learning mechanisms. The process of tire tread production has served as the playground for methodology design and implementation. Validated in this real-world scenario, the solution produced a rise from 81.83% to 90.91% in the tread quality. Thanks to its generic and modular nature, the methodology is applicable to various industrial cases, with the potential to enhance their efficiency and ensure high-quality output.

The rapid development and widespread applications of cognitive computing technologies have led to a paradigm shift towards cognitive intelligent development in manufacturing, where knowledge plays an increasingly important role in enabling higher levels of cognition. Knowledge graph (KG) has emerged as one of the essential tools in cognitive intelligent manufacturing and its completion would significantly impact the quality of knowledge. To facilitate effective knowledge management, KG embedding has proven to be an effective approach for KG completion. However, existing models have deficiencies in achieving relation-specific transformations, differentiating the neighbor nodes, and exploiting the intermediate information generated during the KG embedding learning process, which is prone to limit model performance and hinder successful applications. To address these limitations, this paper proposes a new multi-hierarchical aggregation-based graph convolutional network (GCN), consisting of relation-aware, entity-aware, and across-block aggregation. A parallel relation and entity-aware aggregation (PREA) block is established to simultaneously perform relation-specific transformations and entity-differentiated learning. Additionally, an across-block aggregation is constructed to efficiently integrate extracted information from the intermediate stacked block. To demonstrate the effectiveness and superiority of the proposed approach, 3D printing KG is constructed, which is a representative knowledge-intensive industry linking to a variety of aspects like raw materials, adhesion, usages, etc. Extensive experiments are conducted based on the link prediction task. Illustrative examples are provided to reveal the practical implementation of the proposed method, along with technical details and insightful opinions, exhibiting its promising applications.

The manufacturing sector is experiencing a never-before-seen surge in data generation, acquisition, and analytics. The promising potential of fundamental research following a data-driven approach may enable a more comprehensive understanding of forming operations and efficient optimisation of component quality. However, as a crucial component of metal forming operations, the investigation and insights from a data-centric perspective of hot forging processes is still absent. In the present study, the digital characteristics (DC) of the hot forging process was generated based on voluminous metadata extracted from experimentally verified FE simulations and localised sensors. Inherent and distinctive manufacturing nature throughout the life cycle of a hot-forged product have been revealed, spanning over the design, manufacturing, and application stages. The tribological DC was then extracted and analysed, and the data-guided interactive friction modelling was established to enable a digitally enhanced evaluation and improvement scheme of the lubricant product applied during the hot forging process. Significant potential has been demonstrated in implementing data-centric innovation techniques into traditional manufacturing paradigms to improve efficiency and process effectiveness.

Today’s dynamic manufacturing context, characterized by frequent product variations and consistently rising production volumes, forces companies to continuously adapt their systems with frequent reconfigurations. To support effective decision-making in this regard, it is necessary to have performance evaluation methods that can be modified conveniently to represent configuration alternatives while accounting for the intertwined dynamics of different production areas. The objective of this paper is to propose a modular architecture for performance evaluation of manufacturing systems, able to integrate models of different parts of the same system that are built independently from each other with different approaches, as analytical or simulation. The proposed method is based on the decomposition approach and evaluates the performance of manufacturing systems at the steady-state. The method has been validated through comparison with discrete event simulation considering different system layouts and parameters. Results demonstrate the accuracy of the method and the robustness of the underlying evaluation algorithm. The applicability of the method in industry has been proven in a case study involving the reconfiguration analysis of a manufacturing system producing electrical distribution equipment in scenarios with strongly increasing demand of products.

Tool condition recognition is considered an indispensable solution with significant advantages in improving production cost and quality in intelligent manufacturing. However, the emergence of complex problems such as variable cutting conditions and feature engineering further causes the technology to have a low generalization performance, which severely limits its application in engineering practice. To overcome the above problems as much as possible, a technological framework for monitoring and multi-step forecasting of ball-end tool wear based on multi-modal information under different cutting conditions is proposed. Firstly, a two-stage hybrid deep feature extraction method is proposed by monitoring the cutting vibration and power signals of the spindle. Secondly, a tool wear monitoring model based on SBiLSTM_Multihead Self-attention is proposed to adapt to different cutting conditions. On this basis, a multi-step forecasting model with CNN_SBiLSTM_Multihead Self-attention is proposed to realize the future forecasting of tool wear trend. Finally, the generalization performance of the proposed methods is investigated based on three-axis and five-axis milling experiments. The results show that the correlation coefficient of the enhanced features can reach a maximum value of 87 %. The average accuracy of the proposed monitoring model is improved by an average of 23.84 % over the conventional method. In particular, the multi-step forecasting method is more suitable for long-term forecasting under different cutting conditions. Its average accuracy reaches an average of about 0.013 in the 24-step forecasting. Therefore, the study can provide theoretical references for the application of tool condition recognition in complex machining environments in engineering practice to some extent.