Determination of optimal solar cell parameters using improved particle swarm optimization algorithm

The escalating pace of global development has resulted in a persistent rise in the demand for resources, particularly energy. The combustion of petroleum and coal to satisfy energy requirements can result in the exhaustion of fossil fuel reserves, environmental contamination, climate alteration, and other associated issues. As a result, switching from fossil fuels towards renewable energy has been recognized as the recommended remedy for these issues. Renewable energy comprises wind, geothermal, biomass, and solar. Solar energy is a promising alternative energy source due to its purity, absence of pollution, sufficient reserves, and potential for long-term usage. These systems directly convert solar radiation into electricity, making solar cells one of the most appropriate renewable energy sources. The matter of accurate modeling of these cells is the major topic in this thesis. The parameters should be determined in such a manner that the outcomes of the model match the actual system's measured data. Since the characteristics of a solar cell (current curve regarding voltage) are extremely non-linear, it is impossible to recognize the undetermined parameters of the system using traditional classical methods. Therefore, intelligent algorithms need to be employed. Although many algorithms have been developed to find the optimal settings for photovoltaic cells, many of them share the same major flaw: they get stuck in local optimal points. Furthermore, some algorithms used to resolve the issue of identifying solar cell parameters have a sluggish rate of convergence and intermittent divergence. In order to prevent becoming stuck in the local optima, an algorithm that balances convergence with high accuracy must be given. The solar cell parameters were identified, and the average error power was minimized in this thesis using an improved PSO algorithm. For this investigation, we have taken into account solar cell models with 1- diode, 2- diodes, and 3- diodes. The outcomes of the objective function revealed that the three-diode model had the superior performance, followed by the two-diode and single-diode models with root mean square error values of 6.966263e-04, 7.28000e-04, and 7.7299e-04 respectively. We compared the simulation outcomes of various algorithms to the ones we got by identifying the points of the optimal solar cell model. Our findings indicate that compared to some algorithms, our algorithm produced more accurate results. These findings provide support that our technique for improving the performance of models of solar cells is productive. Furthermore, it has been observed that the estimated parameters obtained from the optimization procedure exhibit a high degree of consistency with actual results.