改进BP神经网络在郑州市需水量预测中的应用

摘要

It is attracting more and more attention to accurately forecast the water demand,for the gap between water supply and demand is enlarging sharply.Accurately water demand forecasting,which "The Three Red Lines" refers as a significant principle,is often used to strengthen the supervision and management of water resources.Based on BP neural network model,the improved BP neural network with self-adaption and gradient descend can be gained if learning rate of BP neural network can adjust automatically.Moreover,the socioeconomic water demand of Zhengzhou city in 2012 and 2015 are predicted will be 14.41× 108 ma and 14.84× 108 ma,using the new method of the impoved BP neural network model.Compared with BP neural network model and the principal component method,the improved BP neural network is more precise,time saving and can automatically adjust learning rate with the iterative error; it can be applied to the process of water demand predication of Zhengzhou city as well.%水资源供需矛盾日益突出,需水量预测已成为广泛关注的焦点.需水量预测可以为“三条红线”的实施提供依据,以强化水资源管理和节水监督管理,缓解水资源供需矛盾.基于BP神经网络模型,采用自适应调整的算法,改进了BP神经网络模型中学习率的求解方法,并将其应用到郑州市经济社会需水量预测中,预测了2012年和2015年经济社会需水量,分别为14.41亿m3和14.84亿m3；通过与BP神经网络模型、主成分回归分析结果对比,发现改进后的BP神经网络模型根据迭代误差自动调整学习率,求解速度和计算结果精度明显提高,适用于郑州市需水量预测.