Research on engine multiple fault diagnosis method based on cascade model

Abstract

The engine is the core component of the power system, and the health status of the components of the engine is very important for the normal operation of the power system. Most of the exist-ing fault diagnosis methods diagnose the engine fault type without further analysis of the sever-ity of the fault. Different fault severity requires different maintenance measures. Therefore, this paper proposes a cascading fault diagnosis model based on Gated Recurrent Unit (GRU) to di-agnose the fault type and the severity of the corresponding fault type.Firstly, the effective fea-tures are extracted from the vibration signals, and then the features are input into the GRU for fault type diagnosis to obtain the sub-fault diagnosis model. After training, one fault type diag-nosis model and four fault severity diagnosis models are obtained. Then the obtained model is cascaded to obtain the total fault diagnosis network. Fault type diagnosis is located at the first level, and four fault severity diagnosis is located at the second level. The effectiveness of the proposed method is verified by experimental data