Photovoltaic (PV) system as a vital element in the utilize of solar energy, its optimization, control, and simulation are significant. The performance of the PV system is mainly influenced by its model parameters that are varying and unavailable, thus identifying these model parameters is always desired. However, accurate and robust parameters estimation of PV models brings great challenges to the existing methods, since the complicated characteristics when estimating the parameters. Hence, to efficiently provide accurate parameters for the PV model, this study develops a self-adaptive ensemble-based differential evolution algorithm. Three different mutation strategies with different properties are combined into two groups for updating each individual. Furthermore, in order to make the best of different mutation strategies, a self-adaptive scheme is suggested to equilibrate population diversity and convergence, by adjusting the proportion of the mutation strategies used in the population. To evaluate the performance of SEDE, it is used to obtain the parameters of three PV models and compared with other well-established algorithms. Systematic comparison results indicate that SEDE is capable of estimating the model parameters with higher efficiency.