Short term load forecast using fuzzy logic and wavelet transform integrated generalized neural network
Short term load forecast using fuzzy logic and wavelet transform integrated generalized neural network
复制标题
DOI：
10.1016/j.ijepes.2014.11.027
复制
发表时间：
2015-05
影响因子：
5.2
通讯作者：
D. Chaturvedi;A. Sinha;O. Malik
中科院分区：
文献类型：
--
作者：
D. Chaturvedi;A. Sinha;O. Malik
人工神经网络(ANN)在电力负荷预测中的应用已有文献报道。人工神经网络有一些固有的缺陷和局限性，如难以确定神经网络的结构、神经元类型的选择、训练时间长、容易陷入局部极小值等，为了克服这些缺陷，人们提出了一种广义神经网络。提出了一种将小波变换、自适应遗传算法和模糊系统与广义神经网络相结合的算法，并将其应用于短期周内日电力负荷预测问题。在预测误差的基础上，将该算法与其他GNN变种的性能进行了比较。
Application of Artificial Neural Networks (ANNs) for electrical load forecasting has been proposed in the literature. ANNs have some inherent drawbacks and limitations, such as difficulty in deciding the structure of ANN, selection of type of neuron, large training time, sticking to local minima, etc. To overcome the drawbacks of ANN, a Generalized Neural Network (GNN) has been proposed in the past. An algorithm that integrates wavelet transform, adaptive genetic algorithm and fuzzy system with GNN is described and applied to the short term week day electrical load forecasting problem. Performance of the proposed algorithm is compared with other GNN variants on the basis of prediction error.