A Sensor-Based System for Fault Detection and Prediction for EV Multi-Level Converters.

Abstract

Power electronic converters and alternating current motors are the actual driving solution applied to electric vehicles (EVs). Multilevel inverters with high performance are modern and the basis for powering and driving EVs. Fault component detection in multilevel power converters requires the use of a smart sensor-based strategy and an optimal fault analysis and prediction method. An innovative method for the detection and prediction of defects in multilevel inverters for EVs is proposed in this article. This method is based on an algorithm able to determine in a fast and efficient way the faults in a multilevel inverter in different possible topologies. Moreover, the fault detection is achieved not only for a single component, but even for several components, if these faults occur simultaneously. The detection mechanism is based on the analysis of the output current and voltage from the inverter, with the possibility of distinguishing between single and multiple faults of the power electronic components. High-performance simulation programs are used to define and verify the method model. Additionally, with this model, harmonic analysis can be performed to check the correctness of the system's operation, and different fault scenarios can be simulated. Thus, significant results were obtained by simulation on various topologies of multilevel converters. Further, a test bench was developed in order to verify some failure situations on a three-level inverter.