基于改进径向基函数网络的电力系统短期负荷预测

电力系统短期负荷预测是电力生产部门的重要工作之一，本文利用径向基函数网络（RBF）进行负荷预测，针对RBF在负荷预测中隐含层节点数难求问题，提出了一种改进的最近邻聚类学习算法即可解决该难点，又可提高RBF神经网络收敛速度和负荷预测精度．根据某地区电网的实例进行研究，结果发现本文算法比改进前的算法预测的最小、最大相对误差分别减小0．14和1．12，证明了改进后算法有效性和可行性，为电力系统负荷预测提供了一种新途径．

Power System Short Term Load Forecasting Based on ImprovedRadial Basis Function Network

Power system Short term load forecasting is one important work of the electricity production sector.In this paper,radial basis function network (RBF) is used in load forecasting.Load forecasting for the RBF in the hidden layer nodes is hard to find.An improved nearest neighbor clustering algorithm is proposed to solve the difficulties and improve RBF neural network convergence speed and load forecasting accuracy.According to the instance of a regional power grid study,we found that the minimum,maximum relative error were reduced by 0.14 and 1.12,if we used the improved algorithm to predict.Case study results prove its effectiveness and feasibility.It provides a new way for the power system load forecasting.