Study On Photovoltaic Power Generation Prediction In Smart Grid

Solar power as a kind of clean and renewable energy,the world begin to pay attention to its development and use.At present, large-scale photovoltaic power generation to access power grid is one of the effective ways of using solar energy,Photovoltaic power generation has obvious intermittent and volatility because it is affected by the intensity of solar radiation and weather, it is harmful to power system security, stability and economic operation.With the increasing of solar power capacity, its influence to power grid is also growing, it is beneficial to scheduling department to adjust the scheduling plan and reduce the adverse effects of time variability by making a high precision, high stability and effective forecasting for the photovoltaic power. It can also reduce system operating costs and spinning reserve power to provide a scientific basis for economic dispatch.After reading a large number of references, in this paper, photovoltaic power generation systems introduced in detail, mainly on the grid-connected photovoltaic power generation system. Then describes the development and characteristics of the smart grid, and analyzes the practical application of photovoltaic power generation in the smart grid.This paper mainly studies the artificial neural network prediction model, at first,describes the basic structure and main features of the model,then analyzes the structure of BP neural network model and analyzes the transfer process of information in the BP neural network. With the goal of improving the prediction accuracy of photovoltaic power, this paper propose ridgelet neural network prediction model and deep neural network prediction model, then their structure and learning algorithm introduced in detail.In order to verify the practicability and feasibility of the model used in this paper, select the historical generation data and meteorological data of U.S. Ashland photovoltaic power plant as the sample data. First, use the data to analyze the main factors affecting the photovoltaic power generation, to provide theoretical reference for choosing input information for prediction model. Then take the sample data to train and test the prediction model, the simulation results show that, comparing with BP neural network model, the ridgelet neural network prediction model and the deep neural network prediction model have a higher forecasting accuracy.