Scholarlink

This paper presents the implementation of an Adaptive Neuro-Fuzzy Inference System (ANFIS) optimized with Particle Swarm Optimization (PSO) for the minimization of Total Harmonic Distortion (THD) in a Multilevel Inverter. The study focuses on optimizing the switching angles to achieve the lowest possible THD, considering the variations in the input voltage sources. The training phase consisted of generating random angle values for ANFIS training. The results show that the ANFIS + PSO model achieves accurate angle predictions, with minimum errors of ± 0.0017 during training and ± 0.0016 during testing. When applied in simulation, the THD values obtained remained within ±0.2% of the calculated values, demonstrating the effectiveness of the trained model. The THD values achieved range from 10.40% to 11.36%, with the largest observed deviation being 1%. Comparison with other works shows that studies on seven-level inverters are limited, and many do not explicitly focus on THD minimization. Factors such as the number of switching angles, including harmonic filters and the total harmonic order, significantly influence the reported THD values. The results confirm that PSO integration improves the internal learning process of ANFIS, making it a promising approach to optimize switching angles in multilevel inverters.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Intelligent Optimization of Multilevel Inverter Switching Angles with ANFIS and PSO: A Simulation-Based Approach

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Oscar Vargas,
  • Susana Estefany De León Aldaco,
  • Jesús Aguayo Alquicira,
  • Víctor Hugo Olivares Peregrino,
  • Ricardo Eliú Lozoya Ponce,
  • Eligio Flores Rodríguez

摘要

This paper presents the implementation of an Adaptive Neuro-Fuzzy Inference System (ANFIS) optimized with Particle Swarm Optimization (PSO) for the minimization of Total Harmonic Distortion (THD) in a Multilevel Inverter. The study focuses on optimizing the switching angles to achieve the lowest possible THD, considering the variations in the input voltage sources. The training phase consisted of generating random angle values for ANFIS training. The results show that the ANFIS + PSO model achieves accurate angle predictions, with minimum errors of ± 0.0017 during training and ± 0.0016 during testing. When applied in simulation, the THD values obtained remained within ±0.2% of the calculated values, demonstrating the effectiveness of the trained model. The THD values achieved range from 10.40% to 11.36%, with the largest observed deviation being 1%. Comparison with other works shows that studies on seven-level inverters are limited, and many do not explicitly focus on THD minimization. Factors such as the number of switching angles, including harmonic filters and the total harmonic order, significantly influence the reported THD values. The results confirm that PSO integration improves the internal learning process of ANFIS, making it a promising approach to optimize switching angles in multilevel inverters.