A combined diagnosis scheme for electric drive system faults

Research on fault diagnosis for electric drive systems has led to the proposal of a joint diagnosis scheme. Firstly, based on nonlinear analytical redundancy residuals, a detection modeling method applicable to faults in electric drive systems is proposed. In this modeling approach, on the one hand, nonlinear analytical redundancy residuals are obtained according to nonlinear system theory, and a general fault detection method is constructed using these residuals. On the other hand, the previously obtained nonlinear analytical redundancy residuals are applied to electric drive systems, and corresponding redundancy residual calculation formulas are generated based on the state space equations of the electric drive systems, thus achieving fault detection in these systems. Secondly, to classify and identify faults, a classification and identification method using a multi-layer perceptron artificial neural network is proposed, and the fault identification principle, architecture design, as well as training and testing based on the multi-layer perceptron artificial neural network are discussed in detail. Finally, simulation experiment results from an actual electric drive system composed of a three-phase asynchronous motor verify the feasibility of the fault diagnosis scheme proposed in this paper.