Development and Reliability Assessment of a Novel 13 Level Inverter with Fault Switch Prediction Using Machine Learning

Abstract

This thesis introduces a novel multilevel inverter (MLI) design utilizing nearest level Modulation technique to achieve maximum voltage levels with reduced component count and voltage stress. The proposed topology consists of eight IGBTs and two DC sources in a 1:2 voltage ratio, generating thirteen distinct voltage levels. Compared to recent MLI designs, the architecture demonstrates superior performance in key metrics, including switch count, DC source requirements, Total Standing Voltage (TSV), and Total Harmonic Distortion (THD). The system achieves a voltage THD of 4.13% and a current THD of 0.81%, ensuring high-quality output suitable for photovoltaic applications. newlineExperimental validation and MATLAB/Simulink simulations were used to assess performance. FFT analysis and loss estimations under various load circumstances were part of simulation research. To validate the simulation results, a hardware prototype was built, and with 0.5 kW of output power, it achieved 95.3% efficiency under RL load conditions. Eight SK25GH12T4 IGBTs with TPL250 gate drivers and an FPGA Spartan XE3S250E controller were used in the prototype to generate pulses. Input voltages of 100V and 200V were supplied by regulated power supply, and output waveforms under resistive, highly inductive, and balanced load conditions were recorded using a DSOX3034T oscilloscope. The results confirm that the suggested MLI is a dependable and affordable option for solar power systems due to its high efficiency, low harmonic distortion, and straightforward design newline