The accuracy of model and parameter identification in photovoltaic（PV）power generation system is of great significance for PV plant fault diagnosis, power prediction and stability evaluation of grid-connected operation. Based on single diode model, the mathematical model of PV array is established. An improved teaching and learning - based optimization algorithm based on the Tent chaotic maps is used to identify the model parameters.The measured data are compared with the simulation results of parameter identification. The accuracy, solving speed ability and stability of model are verified. The proposed model lay the foundation of modeling integrated energy system in the further research.