Abstract:
Multilevel Inverter have become a research hotspot in high voltage and high power applications because of their many advantages, such as their low voltage stress on power switches, low harmonic and EMI output. However, the increasing of the power devices improves the system's fault rate. How to ensure stable operation of the system has become an important research question.
In this thesis work, a fault diagnostic system in a multilevel-inverter using Fuzzy inference System and an Adaptive Neuro Fuzzy Inference System are developed. These techniques are applied to the fault diagnosis of a Multi Level Inverter (MLI) system to avoid the difficulties in using mathematical models. This thesis work presents a fault detection method for open-circuit and short circuit faults of a switching device in diode-clamped inverter systems, which is based on the inherent characteristic of continuous pulse width modulation and its reconfiguration method to avoid feeding faulted output to load.
The phase-to-phase and phase-to-Neutral voltages include information of switching states in the inverter system corresponding to their respected legs but not affected by the load. Therefore, a fault condition of the inverter system itself can be diagnosed through analysis of any of these voltages. Compared to conventional fault detection methods, the present fault detection method has faster detection capability that is within 1 cycle period it can detect the fault and is much simpler to implement. Therefore, the use of the method presented could minimize harmful effects such as imbalance of dc-link voltage and overstress on other switching devices.
