Short-term Load Forecasting In Power System Based On PCA-RBF Neural Network

The electricpower industry is an important pillar industry in China,which is closely related to the people’s life.With the development of the power market,the structure and operation mode of the power market have changed dramatically.Therefore,how to ensure the safety and efficiency of the power system becomes an important issue.And more accurate prediction on power load can provide powerful data support for the power system,which can make generating unitsoperate reasonably,reduce energy consumption and economic spending,and improve the efficiency of power generation at the same time.Nowadays,with a substantial increase in power market demand,the power load prediction are becoming increasingly important,especially the more precise prediction.The neural network has a strong approximation effect and non-linear processing ability.It also has a good processing power for multidimensional signal input.At the same time,it has the characteristics of multiple output.Its self-organization and self-adaptability can deal with non-linear complex system.Principal component analysis(PCA)is a commonly used method in data processing.It mainly transforms high-dimensional variables into low-dimensional independent variables,and maintains all the information of the original data at the same time.There are many factors that affect the accuracy of power load prediction.When the neural network is predicted,the more comprehensive the impact factor is,the more accurate the prediction is,but the input of high dimension will reduce the effectiveness of neural network training.Therefore,this paper uses principal component analysis and neural network to predict the power load.There exist shortcomings of RBF neural network in the short-term power load forecasting,which can be improved by combining PCA with RBF neural network.PCA is used toreduce the network input dimension,eliminate the effect of redundancy and collinearity,and construct PCA-RBF neural network model when extracting effective information.The simulation results show that PCA-RBF neural network can improve the efficiency and maintain the high prediction accuracy as well,which proves the effectiveness of the method.