Short-term Load Forecasting Of Microgrid Based On Improved BP Neural Network

Short-term load forecasting of microgrid is the guarantee of safe,energy-saving and efficient operation of microgrid.It is helpful for optimal dispatching and energy management of micro-grid.Microgrid loads with strong randomness are likely to be affected by external factors and produce large fluctuations,which often results in unsatisfactory forecasting results.Therefore,this paper focuses on how to improve the accuracy of short-term load forecasting in microgrid.Facing the phenomenon of the abnormal data appearing in the historical data of microgrid prediction,the collected data are identified,complemented and corrected,and the related data are normalized.The characteristics of micro-grid load are analyzed,and the variation law of microgrid load in the time dimension of daily cycle and weekly cycle is emphatically studied.Aiming at the problem of connection and implicit information between data,it is proposed to analyze it by using data mining technology.The correlation analysis method is used to study the data,and the dependence and correlation degree between each influencing factor and load are determined according to the characteristic curve of the microgrid load.Principal Component Analysis(PCA)is used to reduce the dimensionality of historical data due to many factors,remove redundant information between data and get less dimension processing data,and use it as input data to improve the model,so as to improve the prediction effect of the prediction model.There is a strong correlation between the load of the microgrid selected in this paper and the external influence factors.Considering that BP neural network has good approximation effect and adaptive ability,firstly,BP neural network prediction model considering multiple influencing factors is established.Secondly,aiming at the problem that BP neural network has poor prediction effect,slow convergence speed and easy to fall into local optimal solution,a method combining mind evolution algorithm and BP neural network is proposed to improve the deficiency of BP neural network algorithm by optimizing the weights and thresholds of the network,and to achieve the purpose of improving the predictive effect.Finally,a model of short-term load forecasting for microgrid with PCA-MEA-BP neural network is established based on principal component analysis,Through MATLAB simulation,it is verified that the BP neural network prediction model improved in this paper is better than the traditional BP and GA-BP neural network in predicting the short-term load of the microgrid.The predicted load value is more consistent with the actual value.