免疫前馈神经网络在电力负荷预报中的应用

Application of Immune Feedforward Neural Network in Short-term Load Forecasting

提出了用免疫算法优化前馈神经网络,将网络结构、激活函数和训练方法等编码作为个体,进行免疫操作,得到最优或次优解。克服了对网络结构、激活函数和训练方法的确定没有可循规则的问题,应用于电力系统负荷预报,取得了比由经验确定的前馈神经网络更好的效果。

A method is presented to optimize the feedforward network by Immune Algorithm （IA）, in which the network structure, activation function and training method are encoded as an individual, in the purpose of optimum solution. It has availably solved problem that there isn＇t a guided rules to specify the network structure, activation function and training method. An application of immune feedfordward neural network in short-term load forecasting is given and the experiment shows that the performance of optimized network is better than that of experiential network and identifies validity of the method.