Industrial Process Monitoring and Fault Diagnosis Based on Temporal Attention Augmented Deep Network

Following the intuition that the local information in time instances is hardly incorporated into the posteriorsequence in long short-term memory (LSTM), this paper proposes an attention augmented mechanism for faultdiagnosis of the complex chemical process data. Unlike conventional fault diagnosis and classification methods,an attention mechanism layer architecture is introduced to detect and focus on local temporal information. Theaugmented deep network results preserve each local instance’s importance and contribution and allow theinterpretable feature representation and classification simultaneously. The comprehensive comparativeanalyses demonstrate that the developed model has a high-quality fault classification rate of 95.49%, onaverage. The results are comparable to those obtained using various other techniques for the Tennessee Eastmanbenchmark process. KCI Citation Count: 0