Knowledge-based and Data-driven Approach based Fault Diagnosis for Power-Electronics Energy Conversion System

Abstract

Recently, power electronic converters have been widely used since more renewable energy systems have been interconnected with the power grid, among which three-phase PWM rectifier is one of the most widely used in drives of electrical motors, AC and DC transmission, and other energy conversion fields. Like any other power electronic converter, three-phase PWM rectifier may be affected by various faults like open-circuit faults. Therefore, fault diagnosis is extremely important to reduce the maintenance costs and improve the stability of the system. A novel open-circuit faults diagnosis method is proposed. The fault diagnosis method only requires the three-phase AC currents, and then Concordia transform is used to calculate the slopes of the current trajectories (knowledge-based). After that the data-driven method of random forest algorithm is used to train the fault diagnosis classifier with slopes data. Finally the knowledge-based and data-driven fault diagnosis methods are combined to achieve fault diagnosis and location. Experiments are carried out and the experimental results are presented to validate effectiveness, robustness of the proposed fault diagnosis method. Furthermore, the proposed method is suitable for vast majority of three-phase energy conversion systems.