Multihead Self-Attention Incremental Network for Rotating Machinery Fault Diagnosis Across Nonstationary Operating Conditions

The advancement of artificial intelligence has enabled significant progress in data-driven fault diagnosis for rotating machinery. However, in real-world industrial environments, such machinery often operates under time-varying and nonstationary conditions, leading to continuous changes in data distributions. These variations pose a major challenge to conventional deep learning (DL) models, which are prone to catastrophic forgetting when sequentially learning new diagnostic tasks. To address this, this article proposes a novel multihead self-attention incremental network (MSIN) for fault diagnosis across nonstationary operating conditions. MSIN employs a stack of multihead self-attention encoder layers to effectively extract fault-relevant features from frequency-domain signals. To alleviate catastrophic forgetting, the model integrates a selective exemplar replay (ER) strategy and introduces a joint cross-head knowledge distillation (KD) loss, enabling effective knowledge retention across tasks. The proposed method is evaluated on two benchmark datasets under both stationary and nonstationary conditions. Experimental results demonstrate that MSIN achieves superior diagnostic accuracy and knowledge retention compared to existing incremental learning (IL) methods, offering a promising solution for continuous fault diagnosis in dynamic industrial settings. The source code of the MSIN model is available at: https://github.com/bankuaimianbao/MSIN-Non-stationary-Code.