摘要
采用动量法、自适应学习率和异变换函数对BP算法进行改进 ,并通过模糊神经网络对中长期负荷进行预测 ,还针对BP算法中隐层节点难以确定的不足 ,大胆采用预测误差曲面方法 ,使隐层节点个数选择具有科学的依据 算法计算表明 。
An improved BP algorithm using momentum, adaptive learning rate and different transfer function is described.Fuzzy ANN is used to forecast long and medium term load. In order to eliminate the difficulty of selection of hidden layer nodes, load forecasting error surface is used so that the selecti on of hidden layer nodes has scientific foundation. Simulation result demonstrat es that the fuzzy ANN improves the forecast precision.
出处
《三峡大学学报（自然科学版）》
CAS
2001年第5期419-423,共5页
Journal of China Three Gorges University：Natural Sciences
