Comparative performance on photovoltaic model parameter identification via bio-inspired algorithms

Photovoltaic (PV) models are usually composed by nonlinear exponential functions, where several unknown parameters must be identified from a set of experimental measurements. Owing to the ability to handle nonlinear functions regardless of the derivatives information, bio-inspired algorithms for parameter identification have gained much attention. In this work, six bio-inspired optimization algorithms, i.e. genetic algorithm, differential evolution, particle swarm optimization, bacteria foraging algorithm, artificial bee colony, and cuckoo search are compared statistically by testing over single-diode models to evaluate their performance in terms of accuracy and stability under uniform solar irradiance and various environmental conditions. Various parameter settings of these algorithms are used in the study. Results indicate that cuckoo search algorithm is more robust and precise among these bio-inspired optimization algorithms. In addition, this paper shows that bio-inspected algorithms are capable of improving the existing PV models by using optimized parameters.