Parameters estimation of solar photovoltaic models via a self-adaptive ensemble-based differential evolution
Introduction
Electricity has entered millions of households and people's lives. In general, electricity is converted from coal, but with the burning of coal, a large number of greenhouse gases cause serious pollution to the environment (Chen et al., 2020, Pourmousa et al., 2019). In addition, coal is non-renewable and its conversion efficiency is not very high. Therefore, it is always necessary to find a clean, pollution-free, and efficient alternative energy source for generating electricity (Kannan and Vakeesan, 2016). To achieve this aim, numerous alternative energy sources have been found, such as wind, solar, water, and tidal energy (Long et al., 2020). Among them, wind, water, and tidal energy put strict requirements on the site, and the construction cost of hydropower plants and windmills is extremely high. On the contrary, solar energy exists everywhere. Through photovoltaic (PV) power generation systems, the electricity generated can be transmitted to millions of households (Chen et al., 2019, Zhang et al., 2020). In order to make the PV system have a higher conversion efficiency under various weather and temperature (Yang et al., 2019), performing simulation, optimization, and control on the corresponding PV model is vital and helpful.
For PV models, there exist several commonly used models, for example, the single diode (SD) and double diode (DD) models (Kler et al., 2019). It is expected to find the models’ parameter values that approach the experimental data to maximize the performance of PV models under specific conditions. Accurate parameters are always desired for obtaining high performance of PV models. Hence, searching for parameters of the PV model can be regarded as an optimization problem solved by a powerful optimization method. Recently, as the population-based search engines, the heuristic method has made great achievements in extracting parameters of PV models. They are differential evolution (DE) (Li et al., 2019), artificial fish swarm algorithm (AFSA) (Han et al., 2014), JAYA algorithm (Yu et al., 2019), genetic algorithm (GA) (Zagrouba et al., 2010), imperialist competitive algorithm (ICA) (Fathy and Rezk, 2017), bacterial foraging algorithm (BFA) (Rajasekar et al., 2013), artificial immune system (AIS) (Jacob et al., 2015), cat swarm optimization (CSO) (Guo et al., 2016), particle swarm optimization (PSO) (Liang et al., 2020a), whale optimization algorithm (WOA) (Oliva et al., 2017), grasshopper optimization algorithm (GOA) (Elazab et al., 2020), cuckoo search algorithm (CSA) (Kang et al., 2018), grey wolf optimization (GWO) (Nayak et al., 2019), wind-driven optimization (WDO) (Mathew et al., 2018), backtracking search algorithm (BSA) (Yu et al., 2018), teaching–learning-based optimization (TLBO) (Yu et al., 2017a), and flower pollination algorithm (FPA) (Xu and Wang, 2017). The reasons why heuristic algorithms can be widely employed in practical applications include: 1) they only need to define the parameter search range and an objective function of the problem and do not need to know the specific information of the problem (Liang et al., 2020b), thus they can be easily implemented; 2) they are not sensitive to the initial solution of the problem, which is their advantage over other mathematical methods (Jordehi, 2016); 3) they adopt a model in which multiple individuals evolve simultaneously in the search domain, in this case, the optimal solution is not easily fall into a local optimum. Meanwhile, this allows them to better deal with the problems with multimodal property and discrete search domain (Das et al., 2016).
Inspired by the observation that the abovementioned algorithms still cannot obtain very accurate parameter values within a limited computation burden, this paper develops a new self-adaptive ensemble-based differential evolution (SEDE) to acquire more accurate parameters. Specially, SEDE employs multiple mutation strategies to compensate for the limitations of a single strategy. Meanwhile, these strategies form two overlapping groups, the first group (group1) is dedicated to making the population search more spaces to avoid losing excellent solutions while the second group (group2) aims at making individuals conduct a local search for accelerating the convergence rate. Meanwhile, each strategy is matched with one parameter combination from the parameters pool to produce promising solutions. Moreover, considering that the requirements for the population diversity and convergence are different in different evolution stages, a reasonable strategy selection is established on the basis of the proposed self-adaptive scheme.
To sum up, the main contributions of this paper are:
- ●
A new self-adaptive ensemble-based differential evolution (SEDE) is proposed to identify the accurate parameters of varied PV models. In SEDE, the ensemble of multiple different strategies and parameters is employed to compensate for the limitations of a single strategy and fixed-parameter settings in the basic DE.
- ●
A self-adaptive scheme without increasing the computing complexity significantly is designed to meet the requirements of the population in different evolution stages by adjusting the relationship between diversity and convergence.
- ●
The proposed SEDE is tested on different PV models, and experimental results show that SEDE has better or competitive performance in comparison with other fifteen state-of-the-art algorithms.
The remainder of this paper is arranged as follows. The PV models, as well as the problem formulation, are reviewed in Section 2. The basic DE algorithm is described in Section 3, which is followed by Section 4 to introduce the proposed algorithm SEDE. Section 5 presents the simulation results. Conclusion and future studies are given in Section 6.
Section snippets
Mathematical model and problem formulation
For the PV system, one accurate mathematical model is particularly important to describe its nonlinear characteristics. Generally, the widely used models include SD and DD, and their model descriptions are presented in this Section.
Differential evolution
Duo to its easy operations and high efficacy, DE (Storn, 1996) has become a widely accepted optimizer. In DE, NP solutions of the Gth generation form a population p is represented by , and each individual in the p is represented by , where D is the decision variable dimension. In the evolutionary process, each individual is updated through the following mutation, crossover, and selection operations.
Motivation
The parameter space of PV models has a large number of locally optimal solutions, which is a challenge for basic DE that has strong global exploration ability. In addition, in the different evolutionary stages, it requires the algorithm has different functions (Wu et al., 2019). For example, the exploration ability is needed to help the population to search for more spaces to prevent missing partial solutions. Meanwhile, the excellent exploitation ability is important to conduct the local
Simulation
In this Section, a common test set includes SD, DD, and PV module is established to conduct the comparison experiment among SEDE and other advanced algorithms. For the SD and DD models, prior researchers (Easwarakhanthan et al., 1986) used a 57 mm diameter commercial R.T.C. France silicon solar cells at a temperature of 33℃ to acquire true Current-Voltage values. For Photowatt-PWP201, 36 polycrystalline silicon cells in series form were operated at 45℃ with the irradiance of 100 W/m2. From (
Conclusions
A self-adaptive ensemble-based DE (SEDE) optimizer is proposed in this paper to estimate the parameters of different PV models. In SEDE, the ensemble of three distinct mutation strategies with complementarity assists the algorithm to solve complex problems. Meanwhile, for the control parameters which influence the evolutionary trend of the algorithm, different parameter combinations are established. By combing the different strategies and parameter settings, individuals can exhibit different
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This work was supported by the National Natural Science Foundation of China (61806179, 61922072, 61876169, 61976237, and 61673404), and Key R&D and Promotion Projects in Henan Province (192102210098).
References (51)
- et al.
Identification of unknown parameters of solar cell models: A comprehensive overview of available approaches
Renew. Sustain. Energy Rev.
(2018) - et al.
Parameter identification for solar cell models using harmony search-based algorithms
Sol. Energy
(2012) - et al.
An opposition-based sine cosine approach with local search for parameter estimation of photovoltaic models
Energy Convers. Manage.
(2019) - et al.
Parameters identification of photovoltaic cells and modules using diversification-enriched Harris hawks optimization with chaotic drifts
J. Cleaner Prod.
(2020) - et al.
Parameters identification of solar cell models using generalized oppositional teaching learning based optimization
Energy
(2016) - et al.
Recent advances in differential evolution – an updated survey
Swarm Evol. Comput.
(2016) - et al.
Simulated Annealing algorithm for photovoltaic parameters identification
Sol. Energy
(2012) - et al.
Parameter estimation of photovoltaic system using imperialist competitive algorithm
Renewable Energy
(2017) - et al.
Parameter extraction of solar cell models using improved shuffled complex evolution algorithm
Energy Convers. Manage.
(2018) - et al.
A multi-role based differential evolution
Swarm Evol. Comput.
(2019)