Fault detection method for IGBT open-circuit faults in the modular multilevel converter based on predictive model

In order to improve the robustness of fault diagnosis method based on model in the modular multilevel converter, an improved predictive model of the circulating current is established. A correction method is proposed to compensate for the uncertainties of the parameters and the measurement errors, and the disturbances are eliminated by using multi-step prediction method. The fault detection method detects faults through comparing the average value of difference between the measured value and the predictive one with the threshold value, and the fault location pinpoints faults by comparing the difference between the capacitor voltage of submodule and the average voltage of submodules with the threshold value. The simulation results verify that the proposed method is characterized with strong robustness.