基于主成分分析L-M神经网络高峰负荷预测研究

在论述电网高峰负荷预测重要性和特点的基础上,将高峰负荷样本,按节气工作日和节假日样本进行聚类,从输入空间入手,采用主成分分析法,减少输入向量的维数,并保留原来输入向量的有用信息,再利用L-M优化算法的多层神经网络预测模型对辽宁电网高峰负荷进行了模拟预测,预测精度令人满意。

Peak load forecasting based on neural network with principal component analysis

This paper dissertates the nessecity and peculiarities of the peak load forecasting. According to the records of workday and holidays, peak load samples are clustered. The principal component analysis approach can reduce the vectorial dimension in the inputting space and keep the useful information of input vector. Then using the predicting model of L-M optimizing algorithm with multilevel neural network simulates Liaoning peak load forecasting and the predicting precision is satisfactory.