A novel triage-based fault diagnosis method for chemical process

Deep learning-based fault detection and diagnosis (FDD) methods have received considerable attention, and many methods based on convolutional neural network (CNN) have been applied to fault diagnosis for chemical processes. However, current fault diagnosis methods train and detect all faults using a single model and the same feature inputs, resulting in the neglect of correlations and difference between faults and inferior fault diagnosis performance. In this study, a novel fault diagnosis method named triage-based convolutional neural network (TrCNN) for fault diagnosis is proposed. Initially, the fault set is partitioned into distinct triage types. Subsequently, distinct models are formulated and applied to their respective triage types in the sub networks layer, while a triage network is developed in the triage layer. Ultimately, the models from the triage layer and sub networks layer come together to constitute the triage fault diagnosis system. The proposed method can adaptively select suitable models and features for different triage types, leading to improved diagnostic accuracy, especially for similar faults. When applied to the Tennessee Eastman (TE) chemical process the TrCNN demonstrates impressive performance, validating its effectiveness in fault diagnosis.