Comparison of PSO, AG and Krill Herd algorithms for the optimization of a PI regulator applied to photovoltaic MPPT

Abstract

Optimizing PI controller parameters is crucial for improving the performance of photovoltaic systems, especially in the context of maximum power point tracking (MPPT). Meta-heuristic algorithms such as Particle Swarm Optimization (PSO), Genetic Algorithm (GA), and Krill Herd Algorithm (KHA) are widely used for this type of optimization. PSO is valued for its fast convergence and ease of implementation, but it can sometimes get trapped in local optima. GA, while efficient in global search, has higher algorithmic complexity and longer computation time. In contrast, KHA, inspired by the behavior of krill schools, offers a good balance between exploration and exploitation, allowing for more stable convergence and better accuracy in finding the maximum power point. Comparative studies show that KHA provides better dynamic control, with faster transient response, reduced overshoot, and effective error minimization (ITAE), surpassing in several cases the performance achieved with PSO and GA. The choice of algorithm, therefore, depends on the desired compromise between speed, accuracy, and robustness