基于人工神经网络的负荷预报模型

A Power Load Forecasting Model Based on Artificial Neural Network

针对基于ＢＰ（ＢａｃｋＰｒｏｐａｇａｔｉｏｎ）学习算法存在的问题，提出了变步幅最速下降和共轭梯度的混合算法来训练人工神经网络，并建立负荷预报人工神经网络模型．为了提高预测精度，对预报权值进行在线修正．实例证明，混合算法在全局收敛特性和收敛速度上要好于基本ＢＰ算法，所建立的预报模型能达到令人满意的精度．

In view of the existing problems of basic BP (Back Propagation) learning algorithm,a hybrid algorithm,which combines the conjugate gradient algorithm with the steepest decent gradient algorithm adopting variable step length is presented to train the neural network.A power load forecasting model based on artificial neural network is also established.In order to improve the forecasting precision,the forecasting weights are modified on line.The results of numerical simulations show that the hybrid algorithm is better than the basic BP algorithm as far as the global convergency and the convergent speed goes.The precision of the forecasting model presented is satisfactory.