Dynamic diversity capture differential evolution to accurately design complex nonlinear photovoltaic system: A heuristic case

Abstract

In the context of renewable energy transformation, the application of solar photovoltaic (PV) technology must rely on efficient and reliable parameter identification methods to ensure the long-term stable operation of the system. Due to PV models are highly nonlinear and it is difficult to estimate parameters, this paper proposes an improved dynamic diversity capture L-SHADE with parameters matrix pre-decomposition approach (DcL-SHADED) to estimate the unknown parameters of PV models. In DcL-SHADED, first, the parameters that need to be estimated are pre-decomposed into parameters of different properties by means of a decomposition method. Secondly, we propose a dynamic population diversity capture mechanism to determine the changing trend of population diversity of different generations. Furthermore, an optimal individual-guided evolution strategy is proposed to improve individual local development capabilities. Therefore, individuals are encouraged to adaptively choose one of two appropriate evolutionary strategies based on the changing trend of population diversity. Finally, DcL-SHADED further identifies parameters with nonlinear characteristics, followed by the construction of matrix equations to reassess the linear unknown parameters. By testing seven PV models with varying degrees of nonlinearity, it is evident that DcL-SHADED outperforms well-known comparative algorithms. This confirms the strong competitiveness of DcL-SHADED.