FRA-Based Parameter Estimation for Fault Diagnosis of Three-Phase Voltage-Source Inverters

This paper presents a fault detection and location identification method for single and double switch Open Circuit Fault (OCF) in three phase voltage source inverters (VSIs) based on current model parameter estimation. The proposed method requires the measurement of the inverter currents to build a dynamic model. A fast recursive algorithm (FRA) is used to estimate the model parameters under either normal and various fault conditions, hence generating a set of fault diagnosis vectors (FDVs) which form a base matrix. A simplified K-Nearest Neighbour (KNN) algorithm is designed to detect the nearest distance, in this case, the Manhattan distance, between the monitored FDV to the normal FDV. When an open-circuit fault occurs, the distance between the two will be significantly increased than a set threshold, hence the fault occurrence can be effectively detected. A simple and effective function based on the analysis of the identified FDV and those in the base matrix is designed to locate the faulty switches. Experimental results under different fault cases are presented to confirm the effectiveness of the method.

Automated summary

This paper presents a method for detecting and locating single and double switch open circuit faults in three phase voltage source inverters based on current model parameter estimation. The method utilizes the measurement of inverter currents to build a dynamic model and uses a fast recursive algorithm to estimate the model parameters. A simplified K-Nearest Neighbour algorithm is employed to detect the closest distance between the monitored fault diagnosis vectors and the normal vectors. A simple and effective function is designed to locate the faulty switches based on the analysis of the identified vectors and those in the base matrix.