基于PCA的BP神经网络在城市供水预测中的应用

URBAN WATER SUPPLY FORECAST MODEL OF BP MEUTRAL NETWORKS BASED ON PCA

可靠的城市供水量预测,是区域水资源科学管理与优化配置的重要基础。为客观预测城市供水量变化趋势,本文采用主成分分析法(PCA)进行主要影响因子分析,根据确定的主要影响因子构造BP神经网络的输入样本,从而进行不同水平年的供水预测。以抚顺市供水量为工程背景构筑预测模型,模拟检验精度较高,可用于抚顺市的供水预测,并为抚顺市水资源的优化配置提供了科学依据。

The reliable forecast on urban water supply is the important foundation for the scientific management and configuration of water resources. In order to objectively reflect the trend of urban water supply, the method of Principle Component Analysis （PCA） was adopted which was used to analysis the principal impact factors, ac- cording to the input samples of BP neural network based on the determined main impact factors, the different lev- els of water supply will be forecasted. Fushun City as a model, the forecasted results with the higher accuracy can be attained. And this conclusion can forecast the urban water supply and also provide a scientific basis for the optimal allocation of water resources in Fushun City.