A Comparative Study of Evolutionary Computation Techniques for Solar Cells Parameter Estimation

Abstract

Recently, the use of renewable energy has attracted the interest of several scientific communities due to the environmental consequences of fossil fuels. Many different technologies have been proposed for the exploitation of clean energies. One of most used is the solar cells considering their unlimited source power characteristics. The estimation of solar cell parameters represents a critical task since its efficiency directly depends on their operative values. However, the determination of such parameters presents several difficulties because of the non-linearity and the multimodal properties from the estimation process. The problem of solar cell parameter estimation has been widely faced through Evolutionary Computation (EC) techniques. Essentially, these methods have produced better results than those obtained by classical methods regarding accuracy and robustness. Each EC algorithm has been designed to fulfill the conditions of specific problems since no one approach can optimize all problems effectively. Under such circumstances, the performance of each EC approach must be correctly assessed considering the application context. Several proposals of EC methods to estimate the parameters of solar cells have been reported in the literature. However, most of them report only a single EC technique considering a minimal number of solar cell models. In this paper, a comparative study of EC techniques used for solar cells parameter estimation is proposed. In the study, the most popular EC approaches currently in use are considered, evaluating their performance over the complete set of solar cell models.

Estimation of solar cell parameters through utilization of adaptive sine–cosine particle swarm optimization algorithm

“…The movements are updated using Eqs. (10) and (11) where (y pbest i ) represents the best solution that y i of particle I have over all previous iterations. y gbest is the best global solution between whole search agents in the top and bottom layers.…”

“…Several approaches have been proposed in recent years to optimize the estimation of SC model parameters [9,10]. The methods for estimating SC model parameters can be categorized into three types: numerical techniques, analytical techniques, and soft-computing techniques.…”