Recently, the research team led by Professor Jiao Weidong at the College of Engineering of Zhejiang Normal University, in collaboration with experts and scholars from Qinghua University, has achieved significant progress in the field of intelligent fault diagnosis for rolling bearings. The related findings were published in the SCI Q1 Top journal *Advanced Engineering Informatics* on January 30, 2025, under the title: "Double attention-guided tree-inspired grade decision network: A method for bearing fault diagnosis of unbalanced samples under strong noise conditions" (https://www.sciencedirect.com/science/article/pii/S1474034624006554). The paper was also honored with the "Best Researcher Award" by the journal's organizing committee.
Achieving hierarchical multi-class fault diagnosis for imbalanced bearing datasets under strong noise conditions presents a challenging problem. To address this issue, this paper develops a hierarchical multi-class fault diagnosis model named the Double Attention-Guided Tree-inspired Grade Decision Network (DATGDN). First, a top-level ternary attention mechanism and an innovative multi-head convolutional attention mechanism were designed to capture bearing fault features. Furthermore, these mechanisms can be integrated into standard convolutional neural networks to form a dual attention-guided backbone network. Finally, by incorporating an innovative tree-structured decision network, DATGDN enables hierarchical decision-making for determining the location and severity of bearing faults. The developed model was tested on two sets of bearing data with varying signal-to-noise ratios and multiple imbalance ratios. Experimental results demonstrate that, compared to several state-of-the-art algorithms, the proposed method not only achieves higher recognition rates across various tasks but also enables hierarchical decision-making for determining the location and severity of bearing faults.
This paper proposes a DATGDN model that combines diagnostic performance with interpretability, designed for diagnosing unbalanced bearing datasets in noisy environments. (a) Within the DATGDN framework, the DACNN serves as the backbone network, leveraging TA and multi-head convolutional attention mechanisms to effectively extract fault-related features, particularly in noisy environments. (b) DATGDN incorporates a TGDL architecture designed for hierarchical decision-making, which starts with fault classification and proceeds to severity assessment, reflecting the systematic evaluation process in human cognition. (c) Across multiple test tasks involving various UR and SNR conditions in two experimental datasets, the proposed GDN consistently outperforms several state-of-the-art algorithms in recognition rates, confirming the effectiveness and superiority of the proposed model. In future work, a thorough examination of the interpretability of mechanisms within the hierarchical architecture will be conducted, potentially using techniques such as Integrated Gradients or Gradient-weighted Class Activation Mapping. Additionally, the noise robustness of the model will be further investigated.
Professors Yonghua Jiang and Weidong Jiao from Zhejiang Normal University, along with Professor Feibin Zhang from Qinghua University, served as co-corresponding authors of the paper, with Zhejiang Normal University being the primary affiliated institution, and Dr.Zhilin Dong, a young faculty member at Zhejiang Normal University, is the first author of the paper, which received support from the Key and Youth Programs of Zhejiang Provincial Natural Science Foundation, as well as the Key Scientific and Technological Project of Jinhua City.
Professor Jiao Weidong's research team (Institute of Equipment Condition Monitoring and Intelligent Maintenance Technology) primarily focuses on intelligent detection and signal processing, mechanical dynamics, equipment condition monitoring, and fault diagnosis, having led numerous research projects including those under the National "863" Program, the General Program and Youth Science Fund of the National Natural Science Foundation of China, the Outstanding Youth Foundation, Key Program, and General Program of the Zhejiang Provincial Natural Science Foundation, as well as major scientific and technological projects in Jinhua City.