Photovoltaic Power Prediction Method Based On Dendritic Neural Network And Its Application

New clean energy accounts for a higher proportion of total human energy consumption.Due to meteorological factors,solar energy is a new type of energy with large usage.It has intermittent and random characteristics.If it is not predicted,it will be connected to the large power grid.It will have an impact on the power system.In order to control power quality and improve system stability,accurate PV power prediction is particularly important.Neural network algorithms have been mathematically proven to be able to fit arbitrary continuous functions.In this paper,two PV power prediction methods are designed around the dendritic neuron model,and a PV power prediction system is developed:(1)PV power prediction method based on dendritic neuron modelA photovoltaic prediction method based on dendritic neuron model was designed.Combined with the actual dataset of PV power plant,the meteorological factors affecting photovoltaic power generation were analyzed and selected by covariance method.The influencing factors were normalized and then applied to the model.In this paper,the MAPE,MAE and RMSE methods are used to evaluate the effectiveness of the proposed prediction model,and the prediction curve,error curve and convergence curve are drawn.(2)PV power forecasting method based on wavelet transform-dendritic neuron modelAs an effective means in signal processing,the wavelet transform algorithm can make the time series exhibit more stable variance and less outliers.Based on this characteristic,this paper first uses wavelet before transmitting the data set to the neuron model.The transformation decomposes the data and then reconstructs the output data of the model,thereby combining the wavelet transform algorithm with the dendritic neuron model to construct a PV power prediction model.(2)PV power forecasting systemThe system is based on the B/S architecture,the front end is developed in Html,CSS and Javascript,using PHP scripts to interact with the back-end database,the database type is MySQL.The system integrates four modules of data acquisition,data storage,prediction algorithm and visualization interface.The prediction results are based on Highcharts framework and are displayed in various ways such as curves and graphs.