Optimizing DC-DC converter topologies for enhanced efficiency in hybrid renewable energy systems using hybrid techniques

Abstract

This paper proposes an innovative hybrid optimization approach to enhance the efficiency of DC-DC converters in hybrid renewable energy systems (HRESs). The method combines Artificial Rabbits Optimization (ARO) for optimal converter design with pseudo-Hamiltonian neural networks (PHNNs) for accurate parameter prediction. Together, ARO and PHNN significantly improve energy utilization and system performance. Implemented in MATLAB, the ARO-PHNN method was benchmarked against Jellyfish Search with Random Decision Forest (JS-RDF), Improved Non-Dominated Sorting Genetic Algorithm II (INSGA-II), Earthquake Algorithm (EA), Particle Swarm Optimization (PSO) with Lightning Attachment Procedure Optimization (PSO-LAPO), and Galactic Swarm Optimization (GSO). Experimental results show that ARO-PHNN outperforms all competitors, achieving a peak efficiency of 98.1%. It also demonstrates strong statistical performance, with a mean of 0.9421, a median of 0.8612, and a low standard deviation (SD) of 0.0065, indicating high accuracy, stability, and consistency. These results validate the ARO-PHNN method as a robust solution for efficient and reliable energy conversion in HRES.