Multiple open-switch faults detection for grid connected inverter

Abstract

This paper proposes a new multiple open-switch faults (OSFs) detection method for grid connected inverter. Frequency deviation and changing loads of the grid interference are mainly concerned. Only line currents of the inverters are used as the inputs. The method uses extended Kalman filter (EKF) to extract the fault current features and random forests (RF) to recognize the features for fault detection. Because the EKF is a time varying method and load variation does not change the shape of current waveforms, the features can be extracted without influence of load changes. By taking fundamental frequency into the filter model, impact of frequency deviation of the grid on fault detection is also discussed and solved. OSFs detections in bias current and overcurrent situation are also discussed. Motor load influence is discussed too. Results of simulation and experiment demonstrate that the new method can be used on multiple open- switch faults within 2 cycles delay in grid in various situations including load changes, bias current, overcurrent, frequency deviation and motor load interference of the grid.