Baigang Du

Precast component production is an important part of industrialization in construction. However, in actual production, it often faces problems such as high resource occupation, long production cycles, and high delivery costs caused by unreasonable scheduling arrangements. This study proposes a hybrid multi-objective optimization framework that combines Q-learning with the non-dominated sorting genetic algorithm, termed QL-NSGA-II, to address the highly coupled problem of component grouping on pallets and production scheduling in precast component manufacturing. The model aims to collaboratively minimize the number of pallets used, the makespan, and the penalty cost. The algorithm constructs a dual Q-table Markov decision architecture, in which the agent adaptively coordinates PC component grouping and scheduling strategies based on reward feedback from the evolutionary state. Through extreme-value tests on multiple benchmark instances from the literature, the results show that the proposed QL-NSGA-II not only achieves the accuracy of the benchmark solutions, but also, in some cases, reduces the makespan and penalty cost by 34.77% and 60.24%, respectively, compared with existing benchmark algorithms. In a real industrial order experiment with 87 components, the algorithm notably outperforms the comparison algorithms across comprehensive convergence metrics such as hypervolume and inverted generational distance, confirming its favorable practical application potential for balancing pallet resource consumption and flow cycle.

Fuzzy control decision-making (FCDM) is an intelligent paradigm that leverages linguistic variables to model knowledge, explicitly addressing epistemic uncertainty (incomplete system knowledge) and aleatoric uncertainty (stochastic noise), thereby enabling resolution of multi-objective optimization conflicts in complex industrial systems. Following the PRISMA protocol, this study conducts a systematic literature review of 123 peer-reviewed publications retrieved from IEEE Xplore, Web of Science, ScienceDirect, and Google Scholar over the period 1965–2026, with emphasis on developments in the past 15 years. Existing reviews predominantly focus on isolated subdomains (e.g., scheduling, maintenance, energy systems), lacking a unified cross-scenario synthesis and implementation framework for industrial FCDM. To address scalability challenges such as rule base explosion in high-dimensional spaces, the literature is analyzed with respect to hierarchical fuzzy architectures, rule pruning, and dimensionality reduction techniques. The primary contribution is a structured synthesis of FCDM mechanisms across four industrial domains, combined with a systematic examination of integration with Industrial Internet of Things (IIoT), Digital Twins, and Edge Analytics. Furthermore, a three-stage closed-loop framework is formalized as a unified optimization protocol and modular architecture with technical specifications for Industry 4.0 integration, comprising data preprocessing, fuzzy inference, and optimization-driven decision output with iterative feedback. Comparative evaluation against MILP, MPC, and DRL highlights the conditions under which FCDM provides superior robustness and interpretability.

Detecting defects on critical surface products is essential in manufacturing for ensuring high-quality standards. However, defect detection faces challenges due to the random variation in defect location and content, as well as inconsistent annotation quality. To address these issues, we propose the CCW-Net for defect detection on critical surface products. To manage positional variation, positional information is embedded within channel attention, emphasizing defect location features through coordinate attention (CA). For handling content variation, a content-aware reassembly of features (CARAFE) aggregates contextual information across a broad receptive field, enabling the capture of comprehensive defect size and shape characteristics. For the uneven annotation quality, the regression loss with dynamic focusing mechanism (Wise-IOU) module is introduced to dynamically allocate gradient gain, which enhances the model’s generalization ability. Experiments conducted on public datasets—NEU-DET (83.9% mAP) and Magnetic-Tile (94.6% mAP)—demonstrate the efficacy, real-time performance, and practical feasibility of the proposed approach.

Precast concrete components, as one of the important structural systems in prefabricated buildings, have received widespread attention due to their efficient manufacturing characteristics on the production line. Their production sequence and layout on the mold table have a crucial impact on production energy consumption. However, a critical constraint is often overlooked in the first step of precast concrete manufacturing: the production sequence and layout of molds are planned without considering the limited availability of molds for each component type. Therefore, this article proposes a mixed-integer programming model for the production sequence and layout of precast concrete components under a limited number of molds, aiming to simultaneously minimize production energy consumption, fluctuation coefficients of mold table utilization, and mold switching time. To obtain high-quality solutions for production sequence and mold layout, a multi-objective genetic flatworm algorithm with a Tabu mapping mechanism is developed to efficiently determine the production sequence and the positions of molds on the mold tables. Through three production cases of precast concrete components with different scales, the proposed model and algorithm have been demonstrated to be highly effective in assisting decision-makers in quickly formulating the optimal production sequence and layout schemes for precast concrete components.

In response to the multifactory collaborative distributed characteristics of building material equipment during the manufacturing process, this paper studies the two‐stage (production–assembly) distributed assembly flexible job shop scheduling problem (DAFJSP) and constructs a corresponding mathematical model. To address this complex NP‐hard problem, a hybrid genetic algorithm with variable neighborhood search (HGA‐VNS) is proposed. The algorithm integrates multiple crossover and mutation strategies, thereby significantly enhancing global search capabilities. Concurrently, the implementation of four neighborhood operations has been demonstrated to enhance local search efficiency. To validate the effectiveness and superiority of the proposed algorithm, comparative experiments with other scheduling algorithms were conducted. The findings indicate that the proposed algorithm demonstrates notable advantages in addressing the DAFJSP.

Most existing data-driven methods for remaining useful life (RUL) prediction of rolling bearings rely on the prerequisite for the existence of massive labeled data with similar distributions. However, in actual industrial scenarios, the sufficient life-cycle degradation data are difficult to collect for a single edge client. Meanwhile, there is the problem of isolated data island among different clients due to the data privacy preservation and commercial rivalry. Hence, it is an urgent challenge to train a prediction model via the decentralized data from multiple clients, especially with different data feature distribution. To fill this gap, a novel federated transfer learning (TL)-based model is proposed for RUL prediction of rolling bearing across edge clients in distributed scenarios. Firstly, a pre-training scheme based on the temporal convolutional network model is employed to learn and preserve the feature representations and temporal dependencies in the bearing degradation data. Then, a knowledge sharing framework based on federated learning facilitates knowledge sharing by transmitting gradient information instead of raw data, thereby ensuring data privacy. Finally, a personalized experience transfer strategy based on TL enhances the model’s prediction accuracy on the target domain through personalized fine-tuning and the introduction of domain loss. By applying this method to experimental data and comparing it with other methods, the superiority and effectiveness of the proposed model are verified. This research provides a promising solution to prognostic problems in limited data with severe distribution drifts and privacy-preserving scenarios.