小波及混沌学习神经网络在短期电力负荷预测中的应用

该文提出了采用小波和神经网络混合模型进行电力系统短期负荷预测方法。首先基于小波多分辨率分析方法将负荷序列分解成具有不同频率特征的序列。然后,根据分解后的各个分量的特点构造不同的神经网络模型对各分量分别进行预测。神经网络算法采用混沌学习算法,与传统BP算法相比,该算法利用混沌轨道的游动性使系统能够跳出局域极值的束缚而寻求全局最优点,这样克服了BP学习算法所存在的本质问题,可以加快网络学习速度和提高学习精度。最后对各分量预测信号进行重构得到最终预测结果。在构建网络模型时,该文考虑了气候因素的影响,并把它作为网络的一组输入点。实验结果表明基于这一方法的负荷预测系统具有较好的精度及稳定性。

Short-term Load Forecast Using Wavelet Transform and Chaotic Learning Neural Network

In this paper,a modified method which combining the wavelet transform and neural networks for short-term load forecast is presented.Firstly,based on wavelet multi-resolution analysis method,the load serials are decomposed to different sub-serials which show the different frequency characteristics of the load.Then an artificial neural network is constructed for each sub-serial according to its characteristics.This paper gives a chaotic learning algorithm to train connection weights of multi -layer feed forward neural network(BP),which has similar ability with random search approach so that the walking of chaotic track can escape from the local extremum and reach the global optimum.Compared with the BP algorithm,this algorithm can quicken the learning speed of the network and improve the predicting precision.Lastly,the forecasting load is obtained by the reconstruction of the learning results.The paper also considers the influence of climate for the short-term load and makes it as one of the inputs for the BP.Experimental results show that the short-term load forecast system based on this method has high precision and high learning rate.