Main Factor Hidden Model Of Power Station Photovoltaic Power Prediction Method

Photovoltaic power generation is an important part of the power system. The proportion of solar photovoltaic power generation in the power system power supply has increased due to its advantages such as renewable, no environmental pollution and huge development scale. However, the inherent volatility and periodic intermittent which exist on solar photovoltaic power will serious affect the safety, stability and economy on power system. To ensure the safe and reliable operation of the electric power system, the changes of the solar photovoltaic power generation must be acquired accurately. Therefore, it is of great theoretical and practical meaning to study the accurate prediction method of solar photovoltaic power generation.Focusing on the physical relationship of PV power station, in-depth analysis of the influence way and characteristics of the various factors on the photovoltaic power station. Based on the influence way, these factors will be divided into two forms of direct factors and indirect factors. Solar radiation intensity on photovoltaic panel is the principal directly influence factor on output power of photovoltaic power station. And this factor is determined by the intensity of solar radiation on the atmospheric boundary and other indirect factors. Considering the changes of in solar radiation intensity on the atmospheric boundary the same time everyday within one month is small, the radiation intensity can be regarded as a constant. Thus, the intensity of solar radiation on the atmospheric boundary can be faded reasonable, and the main cause of the hidden model of indirect causation photovoltaic power plants power is built.A power prediction model of stack auto encoders(SAE) based on deep learning is proposed. The model consists of a number of different automatic encoder(AE) sub-models. Input layer of AE sub-model is indirect factors of the main cause of the hidden model photovoltaic plant power. Output AE sub- model layer corresponding to the output power of the photovoltaic plant. Training for each layer of AE sub-model, at the end of the training, the hidden layer is used as the input of the next layer of AE sub model, then train the next layer of AE sub model, layer by layer training until the output layer, so as to achieve the indirect impact factor to the photovoltaic power plant output mapping.A power prediction method based on radial basis function neural network(RBFNN) for photovoltaic power station is proposed. The main RBFNN of the input layer and the output layer corresponding to the hidden power of photovoltaic power plants and photovoltaic power output of the indirect factors. Firstly, the correlation strength of samples is selected by using the Mahalanobis distance. Then using these samples to train RBFNN, mapping the indirect factors to the output power of the photovoltaic power station.The simulation program is compiled and debugged based on the MATLAB platform. The two kinds of photovoltaic power plant power prediction method based on the SAE model and RBFNN model is simulated and analysed,respectively. And compared with the typical traditional forecasting methods. The results show that the main cause of the hidden model of indirect causation photovoltaic power plants power, based on the SAE model and two kinds of photovoltaic power station power based on RBFNN prediction method is feasible and effective...