Parameters identification of photovoltaic models using an improved JAYA optimization algorithm

Highlights

• IJAYA algorithm is proposed to identify the PV model parameters efficiently.

• A self-adaptive weight is introduced to purposefully adjust the search process.

• Experience-based learning strategy is developed to enhance the population diversity.

• Chaotic learning method is proposed to refine the quality of the best solution.

• IJAYA features the superior performance in identifying parameters of PV models.

Abstract

Parameters identification of photovoltaic (PV) models based on measured current-voltage characteristic curves is significant for the simulation, evaluation, and control of PV systems. To accurately and reliably identify the parameters of different PV models, an improved JAYA (IJAYA) optimization algorithm is proposed in the paper. In IJAYA, a self-adaptive weight is introduced to adjust the tendency of approaching the best solution and avoiding the worst solution at different search stages, which enables the algorithm to approach the promising area at the early stage and implement the local search at the later stage. Furthermore, an experience-based learning strategy is developed and employed randomly to maintain the population diversity and enhance the exploration ability. A chaotic elite learning method is proposed to refine the quality of the best solution in each generation. The proposed IJAYA is used to solve the parameters identification problems of different PV models, i.e., single diode, double diode, and PV module. Comprehensive experiment results and analyses indicate that IJAYA can obtain a highly competitive performance compared with other state-of-the-state algorithms, especially in terms of accuracy and reliability.

Introduction

To tackle the issues of climate change, global warming, and depletion of classical fossil fuels, increasing attention has been focused on the utilization of renewable energy sources. Solar energy can be generally presented as a promising alternative of inexhaustible and clean sources [1]. Solar energy is converted into electrical energy through photovoltaic (PV) systems such as solar cell. PV systems usually operate in harsh outdoor environment and their PV arrays are easy to be deteriorated, which greatly affect the solar energy utilization efficiency [2]. Hence, in order to control and optimize PV systems, it is vital to evaluate the actual behavior of PV arrays in operation using accurate model based on measured current-voltage data. There are several mathematical models that successfully describe the performance and nonlinear behavior of PV systems. The most common and widely adopted models are the single diode model and double diode model [3]. The accuracy of PV models mainly depends on their model parameters. However, these parameters usually are unavailable and change due to aging, faults, and volatile operating conditions. Hence, the accurate identification for parameters is indispensable to the simulation, evaluation, and control of PV systems, and various parameter identification methods have been developed over recent years [4], [5].

Some attempts have been devoted to using deterministic techniques for parameter identification based on minimization of a suitably chosen function [6], [7], [8]. However, deterministic techniques impose various model restrictions such as differentiability and convexity in order to be correctly applied. Besides, since the parameter identification of PV models is a non-linear and multi-modal problem, leading to high probability of falling in local optimal when employing deterministic techniques.

As a promising alternative to deterministic techniques, heuristic methods inspired by various natural phenomenon have been widely used to identify parameters of PV models. They impose no restrictions on the problem characteristic, thus can be easily implemented for various real-world problems. In [9], a penalty based differential evolution (P-DE) was proposed for estimating the parameters of solar PV modules at different environmental conditions. In [10], an improved adaptive DE (IADE) based parameter estimation method was developed by introducing the new formulas for scaling factor and crossover rate. In [11], artificial bee swarm optimization (ABSO) was used to identify the solar cell parameters. In [12], bacterial foraging algorithm was proposed to model the solar PV characteristics accurately. In [13], a biogeography-based optimization with mutation strategies (BBO-M) was developed by incorporating the mutation of DE and chaos theory into the BBO structure. BBO-M was first tested on benchmark functions, and then applied to the model parameter estimation of solar cell. In [14], an improved and simplified teaching-learning-based optimization (STLBO) with an elite strategy and a local search was designed for identifying the parameters of proton exchange membrane fuel and solar cells. In [15], TLBO was implemented by developing an interactive numerical simulation and then applied to the reported current-voltage data of different types of solar cells. In [16], a mutative-scale parallel chaos optimization algorithm (MPCOA) employing crossover and merging operation was developed for solving the designed parameter estimation problem. In [17], artificial bee colony (ABC) was utilized to extract the parameters of solar cells accurately. In [18], bird mating optimizer (BMO) was simplified and then employed to estimate the parameters of module model at different operation conditions. In [1], a DE with integrated mutation per generation (DEIM) was developed to identify the unknown parameters of double diode PV module model. In [19], the performance of six bio-inspired optimization algorithms were compared on the parameters identification of single diode model. In [20], month flame optimizer (MFO) was developed for the parameters estimation of three diode model. In [21], a generalized oppositional TLBO (GOTLBO) was proposed by introducing the generalized opposition-based learning into the initial step and generation jumping, and then used to extract the parameters of solar cell models. In [22], five different versions of the bacterial foraging algorithm (BFA) were developed to extract the parameters of PV module from nameplate data. In [23], a time varying acceleration coefficients particle swarm optimization (TVACPSO) was developed for estimating parameters of PV cells and modules. Although these attempts have achieved satisfied results, the performance of aforementioned algorithms are affected by their algorithm-specific or introduced parameters. It is difficult for users to set the appropriate parameters for a specific or new optimization problem, and the inappropriate tuning of parameters either increase the computational burden or achieve the local optimal solution.

JAYA algorithm is a new yet powerful heuristic method proposed by Rao for constrained and unconstrained optimization problems [24]. It does not require any algorithm-specific parameter except two common parameters namely the population size and the number of generation. Different from JAYA, many other algorithms require the algorithm-specific parameters in addition to common parameters. For example, DE requires the scaling factor and crossover probability, and PSO needs the inertia weight and acceleration coefficients. Hence, a significant benefit of JAYA algorithm can be achieved in terms of omitting the difficulty of adjusting parameters and decreasing the time necessary for conducting optimization process. Although TLBO algorithm is also free from algorithm-specific parameters, it requires two phases (i.e. teacher phase and learner phase) per generation, leading to two function evaluations (FE) for each individual in each generation. Thus, the computation cost of TLBO in a single generation is larger than that of an algorithm with one FE per generation. Unlike TLBO, JAYA algorithm needs only one phase, thus making it less computation time and implementation complexity. JAYA has been improved and widely applied to various real-world optimization problems such as thermal devices [25], two-area interconnected linear power system [26], modern machining processes [27], optimum power flow problem [28], heat exchangers [29], [30], [31], coefficients optimization of proportional plus integral controller [32], constrained mechanical design optimization [33], machining performance optimization during the tuning operation of CFRP composites [34], dimensional optimization of a micro-channel heat sink [35], and other problems [36], [37]. However, as a new algorithm, JAYA has some disadvantages. The first is that there is only guidance as approach to the best solution and avoid the worst solution, although the convergence rate is accelerated, the population diversity may not be maintained efficiently, leads to local optimal solution. The second is that no strategy is used to improve the best solution during each generation, may result in the poor quality of final solution. Besides, to the best of our knowledge no attempts to employ JAYA in solving the parameter identification problems of PV models have been reported in the literature.

In this paper, an improved JAYA (IJAYA) algorithm is proposed to identify the parameters of PV models accurately and reliably. In IJAYA, a self-adaptive weight determined by the best and worst function values is introduced to adjust the tendency of approaching the best solution and avoiding the worst solution. This weight assists the algorithm to approach the potential area at the early stage and implement the local search at the later stage. In addition, a learning strategy based on the experience of other individuals is developed and used randomly to enhance the population diversity efficiently. A chaotic learning is employed to improve the quality of the best solution in each generation. In order to verify the effectiveness of the proposed IJAYA algorithm, it is compared with other well-established algorithms on parameters identification problems of different PV models, i.e., single diode, double diode, and PV module. Experimental results and analyses demonstrate that IJAYA exhibits superior performance in terms of accuracy and reliability. Thus, IJAYA can be an effective alternative for other complex optimization problems of PV systems.

The main contributions of this study are as follows:

• IJAYA algorithm is proposed for the parameters identification of PV models. In IJAYA, a self-adaptive weight is introduced to purposefully adjust the tendency of approaching the best solution and avoiding the worst solution at different search stages.

• An experience-based learning method is designed and implemented randomly to improve the population diversity efficiently.

• A chaotic elite learning strategy is proposed to refine the quality of the best solution in each generation.

• The effectiveness of IJAYA is demonstrated through comprehensive experiments and comparisons on parameters identification problems of different PV models.

The rest of this paper is organized as follows. The problem formulation of PV models is given in Section 2. Basic JAYA algorithm is introduced in Section 3. The proposed IJAYA algorithm is presented in Section 4. The experimental results on different PV models are shown and analyzed in Section 5. Finally, the conclusions are given in Section 6.