19 February 2024 in IEEE Transactions on Industrial Informatics
Fault diagnosis of dynamic multivariate systems is a challenging problem. In this article, a novel fault diagnosis scheme based on variable-wise stacked temporal autoencoder (VW-STAE) is proposed. First, a variable-wise strategy is proposed on the raw industrial data, which sorts the variables for a specific fault by its deviation factor and introduces fault label information during pretraining procedure. Then, temporal autoencoder (TAE) is designed to capture the temporal and spatial feature synchronously and model the complex dependencies of dynamic samples. The stacked TAE is built to enhance the ability of feature extraction by combining multiple TAEs. By inputting the sorted variables sequentially, the VW-STAE is trained as a binary classifier for a specific fault; thereby its input variables and the corresponding network parameters are ultimately selected according to the VW-STAE with the optimal diagnosis performance. Finally, a bank of VW-STAEs is adopted for all faults, which is followed by a fully connected layer to achieve comprehensive fault diagnosis result. The effectiveness of the proposed method is demonstrated in the sensorless drive diagnosis example. The results indicate that the proposed method outperforms other existing deep learning methods.
3 December 2025 in IEEE Transactions on Industrial Informatics
Root cause analysis is critical for identifying fault propagation paths and root cause variables in industrial processes. However, existing methods often overlook deep causal information, require per-variable modeling of temporal dependencies, and offer limited interpretability. To address these challenges, we propose a novel lag-wise temporal squeeze-and-excitation network. This method decomposes the causal inference task along the time lag dimension, employing subnetworks to extract deep-layer causal features. A unified temporal squeeze-and-excitation module then models temporal dependencies across all lags. The proposed scheme integrates lagged inputs, a temporal causal matrix, and interpretable predictions to enable exploration of causality variations and preservation of local causal patterns. Finally, a reachability matrix derived from the causal adjacency matrix quantifies root cause scores. Experimental validation on a real-world industrial mineral process demonstrates the effectiveness and superiority of the proposed method.
10 January 2026 in Journal of the Taiwan Institute of Chemical Engineers