Abstract: This paper investigates the fault diagnosis of electric drive systems and proposes an integrated fault diagnosis scheme combining residual analysis and neural network identification. Firstly, based on nonlinear analytical redundant residuals, a detection modeling method applicable to faults in electric drive systems is proposed. In this modeling method, on the one hand, based on the theory of nonlinear systems, nonlinear analytical redundant residuals are obtai ned and a general fault detection method is constructed using nonlinear analytical redundant residuals. on the other hand, the previously obtained nonlinear analytical redundant residuals are applied to electric drive system, and the corresponding redundant residual calculation formula is generated based on the state space equation of electric drive system, thereby achieving fault detection in electric drive system. Secondly, in order to realize fault classification, a fault identification method based on multi-layer perceptron is proposed. The fault discrimination mechanism, network architecture, as well as model training and testing procedures of the multi-layer perceptron model are analyzed in detail. Finally, based on the simulation results of a real electric drive system composed of three-phase asynchronous motors, the feasibility of the proposed fault diagnosis scheme is verified.