基于级联神经网络的短期负荷概率预测新方法

针对负荷历史数据的统计特征,提出一种基于RBPN网络和RBF网络的级联神经网络预测方法.本模型将历史负荷与其相对应的影响因素进行模式分类,由最大后验概率判别准则确定待预测日影响因素的模式,并利用其对应模式样本数据进行负荷预测.该算法减少了训练样本的数量,提高了预测精度,最后给出的算例证明该方法是合理有效的.

A Novel Method of Short Time Load Probability Forecasting Based on Cascaded Neural Network

To use the statistical character of past data on load , a Cascade Neural Network load forecasting method is proposed in this paper. In this modeling the influences of factors and past load on load forecasting are sorted. The method reduces the quantity of training set and improves the precision of forecasting , which makes use of the rule of maximum posteriori probability to confirm the pattern of forecasting day's factors and load. The experimental results show that the presented model is efficient and feasible.