In this study, a non-mechanism model for the output power of photovoltaic(PV)plant considering weather factors such as irradiance, etc. was established based on a two-layer feed-forward neural network. Firstly, a neural network-based model of output power of PV power plant was established by using weather factors as the inputs. Secondly, the combination of input features for the neural network model was selected. The impacts of different weather factors combinations to the model accuracy were compared to select the input combination of the power model. Then, the training algorithm, the number of hidden layer neurons, and the training times, were changed in the neural network to compare the estimation accuracy and simulation time, and thus the optimal network configuration and parameters of the power model were determined. Finally, the optimized power model of the PV power plant was validated based on the actual measured data. The result shows that the proposed power model has high accuracy.