APPLICATION OF RADIAL BASIC FUNCTION NETWORK BASED ON PRINCIPAL COMPONENT ANALYSIS IN LOAD FORECASTING
ZHAO Jie-hui,GE Shao-yun,LIU Zi-fa (School of Electrical Engineering andAutomation,Tianjin University,Tianjin 300072,China)
When radial basic function (RBF) is applied to power load forecasting, if the input space is heavily self- correlated and the input numbers are too many, in that case too much centres of the neurons will be overlapped, finally the accuracy of load forecasting by RBF network will be descendent. To solve this problem the original input space is reconstructed by principal component analysis and the structure of the network is determined according to the contributions from the principal components respectively, thus, the above mentioned problem is effectively solved. The effectiveness of the proposed algorithm is verified by the practical data of a certain provincial power network.