A Simple ANN-Based Diagnosis Method for Open-Switch Faults in Power Converters

Abstract

This paper presents a new diagnosis method for open-switch faults in power converters based on Artificial Neural Network (ANN). The ANN inputs comprise both sampled signals and control signals. Only the signals of one switching period are used in the method. The combination of control signals and output signals enables the trained ANN to represent the internal characteristics of converter behaviors, which is crucial for fault diagnosis. Compared with other data-driven methods, the ANN approach is simpler, making it easier to be applied in microcontrollers. Besides, the ANN responds quickly to the fault due to the training with instant signals. Therefore, easy operation and fast diagnosis can be both achieved. Finally, the open-switch fault diagnosis in a two-level three-phase converter is studied for method validation. In this case, an ANN is trained with 9 input elements, 7 output elements, and 10 neurons in the hidden layer. Simulation results are given to demonstrate the good performance of the ANN method.