Critical evaluation of soft computing methods for maximum power point tracking algorithms of photovoltaic systems

Abstract

With the proliferation of numerous soft computing (SC)–based maximum power point tracking (MPPT) algorithms for photovoltaic (PV) systems, determining which algorithm performs better than others is becoming increasingly difficult. This is primarily due to the absence of standardized methods to benchmark their performances using consistent and systematic procedures. Moreover, the module technology, power ratings, and environmental conditions reported by numerous publications all differ. Based on these concerns, this paper presents a critical evaluation of the five most important and recent SC-based MPPTs, namely, genetic algorithm (GA), cuckoo search (CS), particle swarm optimization (PSO), differential evolution (DE), and evolutionary programming (EP). To perform a fair comparison, the initialization, selection, and stopping criteria for all methods are fixed in similar conditions. Thus, the performance is determined by its respective reproduction process. Simulation tests are performed using the MATLAB/SIMULINK environment. The performance of each algorithm is compared and evaluated based on its speed of convergence, accuracy, complexity, and success rate. The results indicate that EP appears to be the most promising and encouraging SC algorithm to be used in MPPT for a PV system under the multimodal partial shading condition.