A Comparative Study on Performance of Fitness Functions for Harmonic Profile Improvement using Parameter-less AI Technique in Multilevel Inverter for Electrical Drives

Abstract

This article proposes the comparative investigation of harmonic profile improvement in the multi-phase multilevel inverter. For the proposed study, three-phase, seven-level cascaded H-bridge (CHB) multilevel inverter (MLI) is considered. Modulation of the stepped waveform output of the multilevel inverter is done using selective harmonic elimination (SHE) method. Many algorithms are proposed for solving the set of nonlinear transcendental trigonometric equations for selective harmonic elimination methods such as Genetic algorithm (GA), Particle Swarm Optimization (PSO) algorithm, and many more. Most of these techniques have controlling parameters, which need to be tuned while optimizing the fitness functions. The teaching learning-based optimization (TLBO) algorithm and JAYA algorithms are parameterless optimization techniques. Due to the lack of controlling parameters, these algorithms are most robust among the family of artificial intelligence (AI) techniques. TLBO algorithm has an added advantage of fast convergence compared to the JAYA algorithm. This advantage attracts most of the researchers to use TLBO in engineering Applications. A novel fitness function is proposed in this paper. The performance of the proposed fitness function is compared with two different popular fitness functions reviewed from various literature. A comparative investigation is carried out on these three fitness functions for controlling total harmonic distortion (THD) in a multilevel inverter. Throughout the article, a comparative exploration of lower order harmonics and THD profile with respect to modulation index is carried out for each fitness function. A most suitable fitness function is concluded after comparing total harmonic distortion profiles of each. In this article, a simulation study is carried out using the parameterless TLBO algorithm, and the performance is compared with the PSO algorithm. It is seen in this study that the proposed novel fitness function with the TLBO algorithm improves the harmonic profile by 17% compared to the PSO algorithm, producing the most optimum result.