A Diagnosis Algorithm for Multiple Open-Circuited Faults of Microgrid Inverters Based on Main Fault Component Analysis

This paper presents a new open-circuit fault diagnosis algorithm for multiple switch of microgrid inverter in different load change condition. From the analysis of the waveform features and frequency composition for the main fault component under the different switch states, a low frequency sampling principle of the main fault component is developed. For the proposed fault diagnosis algorithm, first, based on above principle, small sampling data are obtained by the second low frequency mean sample processing, which can be used to reflect the main component features of different switch states and reduce the amount of the processed data. Because the data of the second low frequency mean sample processing comes from the obtained system data, it does not add any additional hardware and has the lower cost. Moreover, the obtained low frequency sampling data are divided into different feature data to adopt corresponding data processing scheme. Thus, the information of the main fault component is retained and the impact of load change is removed for any feature data. Furthermore, the feature values of the processed low frequency sampling data are extracted from the viewpoint of data attenuation. Finally, neural network is used to implement intelligent classification. The dependency and the number of threshold can be reduced, and low frequency sampling data are fully utilized to reflect the overall features. Compared with the existing fault diagnosis algorithms, the actual computational quantity is effectively reduced without affecting the accuracy and stability of the diagnosis results for microgrid inverter under load change condition. Finally, the effectiveness of fault diagnosis algorithm is verified through the detailed simulation and experiment results.