Optimizing Matrix Converter Performance Using Hybrid Particle Swarm Optimization and Gradient Descent Techniques

This paper presents a comprehensive comparative analysis of three optimization algorithms—Particle Swarm Optimization (PSO), Modified PSO (MPSO), and Hybrid PSO with Gradient Descent (HPSOGD)—for matrix converter optimization, validated through hardware-in-loop (HIL) implementation. The experimental results demonstrate HPSOGD's superior performance, achieving a Total Harmonic Distortion (THD) of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{6. 1 \%}$</tex>, marking a substantial <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{3 0 \%}$</tex> improvement over conventional PSO's 8.7% THD and outperforming MPSO's 6.7% THD. The convergence analysis across multiple operating frequencies (5 Hz, 50 Hz, and 100 Hz) shows HPSOGD consistently achieving optimal fitness values of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$6.0,7.0$</tex>, and 5.0 respectively, compared to PSO's higher values of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$8.0,9.0$</tex>, and 7.0. HIL implementation validates these results through real-time waveform analysis, demonstrating stable voltage regulation and effective harmonic suppression, with higher-order harmonics maintained below 15% of the fundamental magnitude. MPSO, utilizing adaptive parameters and enhanced neighborhood topology, achieves a 23% THD improvement over conventional PSO, positioning it as an intermediate solution. The oscilloscope measurements confirm superior switching pattern quality in HPSOGD, with 50% duty cycle optimization and minimal switching noise. Convergence rates average 2.33 units per iteration across all algorithms, with HPSOGD demonstrating enhanced stability in reaching optimal solutions.