摘要
介绍了BP(误差反向传播算法)和GA(遗传算法)及GA-BP3种神经网络,并以此分别对城市用水量进行预测。实验结果表明,基于GA-BP算法的神经网络方法应用于城市用水量的预测问题,能采用遗传学习算法优化BP神经网络模型的初始权重,即先利用遗传学习算法进行全局训练,再用BP算法进行精确训练,使网络收敛速度加快和避免局部极小。GA-BP神经网络在收敛速度和预测精度等方面均优于BP和GA网络,从而为未来短期城市用水量负荷的准确预测提供了新的思路与方法。
This article introduced three NN such as BP, GA and GA-BP, and it also forecasted the city water consumption by them. The result indicates that the city water consumption forecast problem which based on GA-BP NN could optimize BP. First it carry through whole education with GA, and then use BP to carry through accurate training. With that the net convergence velocity becomes faster. GA-BP NN is better than BP and GA networks in constringency speed and forecast precision, thereby, it took a new mentality and method for accurate forecast of urban water consumption burthen in future.
出处
《水科学与工程技术》
2007年第3期1-3,共3页
Water Sciences and Engineering Technology
