Fault Diagnosis of Cascaded H-Bridge Inverter Using Model Predictive Control

This paper proposes a new fault detection and localization (FDL) algorithm based on modulated model predictive control (M2 PC) method for Cascaded H-bridge (CHB) inverters. Model predictive control (MPC) algorithms have been widely applied in medium-voltage (MV) power converters for its advantages in dynamic performance and control of multiple objectives. However, it has some limits for multilevel topologies, including variable switching frequency and large computation burden. In addition, in cascaded H-bridge (CHB) multilevel inverters, fault detection and localization need to be incorporated into the control system to make it work normally when there is an open fault in power cells. This paper employs M2 PC approach to calculate output voltage references with reduced computation burden in every sampling time. Also, M2 PC features a fixed switching frequency in the system which helps improving harmonic spectrum and simplify filter design. Furthermore, FDL scheme can detect open switch fault in CHB multilevel inverter by comparing actual and predicted phase voltages. With the fault detection method, open switches can be localized through the fault matrix with the measurement of load currents and phase voltages in various open-fault scenarios. Simulation results verify the performance of proposed M2 PC and FDL algorithms.