In the cross-machine fault diagnosis task of bearings for rotating machinery, inherent physical differences among different mechanical devices cause severe domain distribution shift of fault features. Consequently, the diagnostic model trained on the source machine suffers a dramatic decline in generalization performance when applied to diagnosis tasks of target machines, making it difficult to meet the requirements of practical industrial diagnosis. To address this issue, this paper proposes a heterogeneous dual-stream contrastive adversarial network (HDCAN). Firstly, one-dimensional (1D) vibration signals are converted into two-dimensional images via the symmetric dot pattern, constructing a signal-image multimodal input system. Secondly, a parallel dual-stream network architecture is designed, and a cross-modal contrastive learning strategy is introduced simultaneously to effectively eliminate the negative transfer effect between heterogeneous modalities and realize the pre-alignment of multimodal features. On this basis, the domain adversarial learning mechanism is integrated to further extract generalized features with both fault discriminability and domain invariance, so as to improve the cross-machine adaptation capability of the model. To verify the effectiveness of the proposed method, experiments are conducted on six cross-machine diagnosis tasks from three public bearing fault datasets. The experimental results show that the average diagnostic accuracy of the HDCAN model reaches 92.10%, which significantly outperforms the current mainstream domain adaptive fault diagnosis methods. The experimental results fully demonstrate that the proposed HDCAN can effectively alleviate the distribution shift problem of cross-machine fault data and possesses great application potential in industrial practical scenarios.
Haoran Liu, Delong Cui, Zhiping Peng et al.· Structural Health Monitoring· 0 citations
Automated visual recognition in industrial environments has become a key enabler for intelligent monitoring systems, particularly in safety‐critical domains such as chemical plants. Unlike general industrial vision tasks, chemical instrument monitoring operates under stringent constraints, including harsh environments, legacy non‐digital devices, and high reliability requirements for decision‐making. In this context, deep learning‐based visual recognition should not be viewed as an isolated perception task, but rather as a critical component within expert systems that support operational decision‐making. This paper presents a comprehensive scoping review of deep learning–driven visual intelligence for industrial applications, with a specific focus on chemical instrument monitoring from an expert systems perspective. A total of 127 studies published between 2015 and 2025 are systematically analysed, covering object detection, image segmentation, and optical character recognition (OCR), along with their integration into practical monitoring pipelines. Beyond categorizing models, this review emphasizes the evolution from standalone perception models toward system‐level solutions that incorporate edge deployment, data‐efficient learning, and explainability. Key challenges are identified, including data scarcity due to proprietary industrial datasets, limited model generalization across dynamic environments, computational constraints in real‐time deployment, and the lack of interpretability required for safety‐critical decision support. Importantly, this review highlights the gap between high‐performing vision models and their reliable integration into expert systems for chemical operations. Emerging trends are discussed, including synthetic data generation, domain adaptation, foundation models (e.g., SAM and open‐vocabulary detection), and multimodal fusion, with a critical evaluation of their applicability and limitations in chemical instrument scenarios. By reframing industrial visual recognition within the broader context of expert systems, this work provides a structured understanding of current capabilities and outlines future directions for developing trustworthy, efficient, and human‐centric intelligent monitoring systems in chemical industries.