An Improved Stochastic Fractal Search Algorithm: dFDB-SFS

dFDB-SFS is proposed (i) Global Optimization (ii) Constrained Power Systems Optimization (iii) determining the parameters of the PV model

Description ✱

A Dynamic Fitness-Distance Balance based Stochastic Fractal Search for Solving Global Optimization and Determining Accurate Modeling of Photovoltaic Models

Stochastic fractal search (SFS) algorithm, among population-based metaheuristic automation algorithms, is a robust optimization algorithm for solving optimization problems in different fields of science, inspired by the diffusion feature and natural growth phenomenon seen regularly in random fractals. However, as in population-based optimization algorithms, it is a great challenge to effectively design the selection process in the SFS method. In order to imitate the selection process in nature effectively and accurately, the dynamic-fitness-distance-balance (dFDB) selection method has been used in the SFS algorithm in six different versions. In this way, the dFDB-SFS algorithm has been developed, which more effectively mimics nature with exploitation, exploration and balanced search capabilities. Firstly, the performance of the proposed dFDB-SFS algorithm was investigated in CEC 2020 benchmark test functions. Wilcoxon and Friedman statistical analyzes of the results obtained from the test functions were made and according to these analysis results, the best version of the proposed approach was determined. Secondly, the performance of the proposed algorithm is used to determine the photovoltaic module parameters, which is one of the real-world engineering problems. The accurate parameter estimation of the photovoltaic (PV) module is essential for studying and optimizing the energy output of PV systems into electric power systems. The accuracy of these parameters is mainly based on the optimization algorithm implemented and the objective function used. In this paper, dFDB-SFS are proposed and applied to estimate unknown parameters of the single diode model (SDM), double diode model (DDM), and PV module models. The root mean square error (RMSE) between the measured and estimated current datasets, which is widely used in the literature, is adapted to evaluate the method’s effectiveness. These three cases have been used to validate the performance of the best one from these proposed algorithms. A quantitative and qualitative performance evaluation reveals that the sixth proposed algorithm has improved results than other proposed algorithms in terms of accuracy and robustness when obtaining PV parameters.