Cross-condition bearing fault diagnosis suffers from severe performance degradation due to domain shift across different operating conditions, especially when only a few labeled target-domain samples are available. To address this challenge, this paper proposes a progressive attention-guided two-stage transfer learning framework for few-shot bearing fault diagnosis across different fixed operating points. First, the raw time-domain vibration signals are fused with frequency-domain representations extracted by short-time Fourier transform (STFT) to enhance fault feature representation. Then, a progressive attention-guided feature learning strategy is developed by integrating dual efficient channel attention (ECA) modules into a deep one-dimensional convolutional neural network (1D-CNN), enabling the network to adaptively emphasize fault-sensitive features while suppressing redundant information. Subsequently, a two-stage transfer learning strategy is designed, consisting of transferable feature learning from the source domain and few-shot adaptation to the target domain. During target-domain adaptation, key feature extraction layers are frozen, and a sample-balanced optimization mechanism is introduced to alleviate the dominance of source-domain samples during joint training. Experimental results on the Case Western Reserve University (CWRU) bearing dataset demonstrate that the proposed method achieves an average accuracy of 99.96% across three cross-condition transfer tasks. Furthermore, experiments conducted on a self-built shaft system dataset show that the proposed method achieves an average accuracy of 87.11% under three representative transfer scenarios. The results verify that the proposed framework effectively mitigates domain shift and enables accurate bearing fault diagnosis with limited labeled target-domain samples.
Zi-Yi Zhang, Long-Chao Cao, Zhe Wang et al.· Machines· 0 citations
Multi-objective optimization plays an important role in modern design and complex engineering applications. However, achieving an effective balance between the convergence and diversity of Pareto-optimal solutions remains challenging. This paper proposes a Sobol-driven Multi-objective Whale Migration Algorithm (SMOWMA), which extends the Whale Migration Algorithm within a non-dominated sorting and elite-selection framework. A maximin scrambled Sobol initialization scheme is first employed to improve the distribution of the initial population. An archive-guided adaptive Student-t flight mechanism is then incorporated into the leader-whale position update to dynamically balance global exploration and local exploitation. In addition, archive crowding information and archive-entry success feedback are jointly used to adjust the search behavior according to both environmental diversity and recent search performance. SMOWMA is evaluated on five widely used multi-objective benchmark suites, namely ZDT, DTLZ, WFG, UF, and CF, using four performance indicators: generational distance (GD), inverted generational distance (IGD), spacing (SP), and hypervolume (HV). The results, together with Friedman tests and Holm-adjusted Wilcoxon tests, demonstrate that SMOWMA achieves competitive overall performance in terms of convergence, diversity, and objective-space coverage, although its relative advantage remains problem-dependent. The practical applicability of SMOWMA is further examined using multi-objective welded-beam design formulations, a bi-objective four-bar truss design problem, and a five-objective car side-impact design problem. The engineering results show that SMOWMA can obtain competitive and stable approximation sets for constrained design problems with different numbers of objectives, supporting its effectiveness and applicability in multi-objective engineering optimization.
Li-Zhen Du, Dahongnian Zhou, Xiao-Shuang Xiong et al.· Biomimetics· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
