RBFNN Method of Calculating Energy Losses of Power Distribution Systems
配电网线损计算径向基函数神经网络方法

A Radial basis function neural networks(RBFNN) method of calculating energy losses in distribution systems is proposed. RBFNN method, due to its strong regression ability, is able to map complex non-linear relation between energy losses and feature parameter in distribution systems, and memorize the rule of energy losses varying with distribution net structure and operation parameters. LBG clustering algorithm and a clustering criterion are used to determine optimal number of hidden nodes of RBFNN, and therefore the use efficiency of the RBFNN is improved. Simulation has verified the validity and practicability of the proposed method.