Research On Photovoltaic Power Forecast System Based On Combination Mode

As the fossil energy depletion, the use of new energy is imminent. Solar energy as a clean new energy, has been promoted in many areas. Photovoltaic power generat ion is a typical application of solar energy resources. Solar power generation and po wer plant have close relation with the geographical environment and meteorological conditions, and photovoltaic power generation has certain volatility and instability, w hich brings a certain pressure to the large-scale grid, so the reasonable forecast of ph otovoltaic (pv) power output will be a theoretical basis for the grid and reasonable sc heduling. In this paper, a combined forecasting model is proposed to predict the futu re photovoltaic power output.At present, the research on photovoltaic power prediction mainly has two directions. One is through the establishment of local solar radiation prediction model, through the meteorological data to forecast the irradiance, and then through fitting and prediction power data are obtained. Another, it is through the statistical analysis of the history of the photovoltaic power station operation data to forecast the power data directly, this method is simple and need not to establish the forecast model of complex. From the perspective of the algorithm, neural network, wavelet analysis and support vector machine (SVM) is a training method is generally used to forecast model, and single prediction algorithm has some limitations, is not suitable for applications in a variety of scenarios, so this paper will put forward a model based on combination of photovoltaic power prediction system, improve the accuracy of the prediction, reduce the photovoltaic effect on the grid.