Research On Photovoltaic Power And Load Forecasting Method Of Microgrid Based On Improved Neural Network

In order to solve the worsening problems of environmental pollution and energy shortage,many countries are vigorously developing distributed power generation and constructing microgrid based on distributed power generation.Photovoltaic power generation has gradually become the main development direction of distributed power generation due to its advantages of convenient access and environmental protection.However,the photovoltaic output and load of microgrid are easily affected by objective and uncontrollable factors such as various weather and external environment,and will have greater volatility and randomness.Therefore,it is of great significance to obtain accurate photovoltaic output and load data of microgrid for realizing economic and stable operation of microgrid.The main research work of this paper on photovoltaic output and load forecasting of microgrid includes:(1)The structure and mathematical principle of radial basis function neural network(RBF neural network)and a multi-layer feedforward neural network(BP neural network)with back propagation of errors are deeply studied.The advantages and disadvantages of RBF neural network and BP neural network are compared.The method of applying BP neural network to photovoltaic output and load forecasting of microgrid is explored.(2)Because there are many factors that affect the photovoltaic output and load of microgrid,it is necessary to simplify the data.In this paper,the data preprocessing method and grey correlation theory are studied in depth,and the grey correlation theory is used to analyze the influence of various meteorological conditions and date types on photovoltaic output and load,so as to obtain similar days with higher similarity,so as to achieve the purpose of simplifying the input of neural network and improving the prediction accuracy.(3)This paper studies the power generation principle of photovoltaic microgrid,introduces the structure and components of AC and DC microgrid,and analyzes the output power characteristics of photovoltaic microgrid and the influence of main meteorological factors on the power generation of photovoltaic microgrid.At the same time,the related influencing factors of microgrid load,such as temperature,date type and weather type,are analyzed.The selection method of similar days for photovoltaic output prediction and load prediction of microgrid and the input vector composition and network structure of neural network are determined.(4)The mind evolution algorithm is deeply studied,mainly including the basic concept,algorithm steps and mathematical principle of mind evolution algorithm,and the process of improving BP neural network by mind evolution algorithm is introduced.Finally,combined with practical cases,three prediction methods based on neural network are verified.The results show that,compared with the traditional BP neural network and radial basis function neural network,the BP neural network prediction method(MEA-BP)optimized by mind evolution algorithm has higher prediction accuracy in short-term photovoltaic output and load prediction of microgrid.