Product quality is crucial to the competitiveness of manufacturing enterprises. However, existing quality-aware scheduling studies mainly approximate product quality evolution using indirect factors such as processing time, making it difficult to characterize the integrated effects of equipment health conditions and operation precedence relationships on product quality. To address this issue, a quality-aware hybrid flow shop scheduling problem with multilevel assembly (QHFSP-MA) is formulated by incorporating multilevel assembly processes and processing-assembly quality propagation mechanisms of complex precision products. Due to the large solution space and multi-objective characteristics of QHFSP-MA, existing multi-objective reinforcement learning methods still struggle to obtain high-quality Pareto fronts. Therefore, a preference-guided graph reinforcement learning framework (PGRL) is proposed. A heterogeneous graph is first constructed to represent the processing-assembly system. Then, a preference-guided meta-relation heterogeneous graph neural network is developed to dynamically adjust the importance of different meta-relations according to objective preferences, enabling generalized policy learning under diverse preferences. Furthermore, an adaptive preference weight mechanism (APWM) is designed to reduce the mismatch between uniform preference sampling and irregular Pareto front distributions, thereby improving Pareto front approximation. Experimental results demonstrate that PGRL outperforms dispatching rules, metaheuristic algorithms, and two advanced multi-objective deep reinforcement learning methods in terms of optimization performance, generalization, and computational efficiency. An automotive engine manufacturing case further validates its practical effectiveness, reducing makespan, quality loss cost, and maintenance cost by 27.3%, 25.7%, and 8.8%, respectively.
Reliable fault detection and health monitoring of industrial robots are important for smart manufacturing in Industry 4.0. However, robotic platforms differ in their mechanical design, sensing characteristics, and operating conditions. These differences make it difficult for data-driven models to generalize to unseen robots and working conditions. To address this challenge, this paper proposes a cross-robot framework for open-set fault detection using only normal data from the target robot. The framework uses a condition-normalized residual adaptation mechanism in the latent space. First, a structured representation is learned from the source robot. The learned representation is then adapted to the target robot using only normal samples. Compactness-based constraints and statistical modeling of normal behavior are used to detect unknown faults as deviations from the normal representation. The framework is evaluated on two industrial robots under variations in load, speed, and motion. Two evaluation strategies are also considered: sequential normal-only adaptation to welding motion and direct open-set detection without further adaptation. The results show consistent fault detection across different robots and operating conditions. A comparative benchmarking analysis further demonstrates the effectiveness of the proposed approach and its ability to maintain reliable fault detection performance under different loading and motion conditions. This demonstrates the potential of the framework for practical industrial health monitoring.
Mid-spatial-frequency (MSF) errors induced by repetitive toolpaths remain a major limitation in robotic polishing of precision optical surfaces. This study proposes a closed-loop robotic polishing cell with toolpath orchestration that integrates surface metrology, removal-function management, a parameterized toolpath library, path-based material-removal prediction, and multistage robotic execution. The proposed approach replaces isolated single-path optimization with stage-specific coordination of complementary trajectories. A NURBS-interpolated Gosper path is incorporated as a functional toolpath within the proposed framework, where the space-filling characteristics of the Gosper curve provide multidirectional coverage and NURBS interpolation improves trajectory continuity for MSF-error suppression. Numerical verification and two sets of polishing experiments were conducted on aspheric optical components. In the representative Hilbert–NURBS-interpolated Gosper strategy, the Hilbert path provided efficient baseline correction, whereas the proposed path further suppressed residual periodic textures. Numerical results reduced PV/RMS from 403.216/70.968 nm to 146.891/18.255 nm, corresponding to reductions of 63.6% and 74.3%, respectively. Experimental PSD analysis showed that this coordinated strategy produced the lowest final PSD value among the three tested configurations, with an overall convergence rate of 86.849% at the characteristic MSF wavelength. Additional experiments comparing spiral–NURBS-interpolated Gosper and NURBS-interpolated Gosper–pseudo-random sequences provided further experimental evidence for the effectiveness of toolpath coordination under the tested conditions. The former achieved PSD convergence rates of 99.99%, 99.96%, and 99.02% at characteristic wavelengths of 4, 2.5, and 1.28 mm, respectively, and produced the lowest final spectral levels. These results demonstrate that coordinated use of complementary paths can simultaneously improve form convergence, suppress characteristic MSF peaks, and enhance residual-surface uniformity, providing a framework for integrating metrology, toolpath planning, and robotic precision finishing.
Additive manufacturing (AM) faces critical challenges, particularly in achieving real-time quality control and precision. These challenges are heightened in complex geometries and high-deposition-rate robotic AM (HDR-RAM) processes such as robotic cold spray. Traditional quality control methods generally detect defects after the entire part is produced, leading to material waste and increased lead time and cost. The lack of real-time insight during production limits the ability to identify and correct the process variations as they occur. To address this, digital twin technology has emerged as a powerful tool within the intelligent manufacturing paradigm. This study proposes a novel geometric digital twin (gDT) framework, gDT-AM, specifically designed for HDR-RAM. It continuously captures and maps the part’s geometry in real-time during printing. It incorporates two experimentally validated alternative methods for precise, fast surface reconstruction from sparse spatio-temporal 2D laser profiler scans. This supports the monitoring of geometric anomalies during deposition by providing timely, actionable information for intervention to reduce material waste and process inconsistency. The proposed framework provides a basis for online process-parameter optimization and toolpath correction. Ultimately, integration of the digital twin aligns with the principles of intelligent manufacturing through connectivity, automation, data-driven decision-making, and adaptive process control.
