基于粗集和改进BP网络的城镇日用水量预测模型

Forecasting model of urban daily water demand based on rough sets and improved back-propagation network

针对城镇日用水量受某些影响因素冗余性、非定量性、非线性的影响以及这些影响在预测模型中很难体现等问题,分析了影响城镇日用水量的因素,利用粗集知识约简方法去除冗余,选择影响城镇日用水量的主要因素,结合改进的BP网络建立城镇日用水量预测模型,并将该模型的预测效果与未采用粗集方法去除因素冗余的模型预测效果进行比较,结果显示该模型的预测精度更高、所需时间更短、更加适用于影响因素较多的城镇年、月用水量的预测。

In view of the fact that urban daily water demand is subject to certain factors that are redundant, non-quantitative, nonlinear and not easily incorporated into a forecasting model, these factors are deleted with the rough sets theory and a forecasting model of urban daily water demand is established based on improved back-propagation （BP） network analysis. The results of an example simulation show that, after the redundant factors have been discarded with the rough sets theory, the prediction accuracy is higher and the forecasting time is shorter. Hence, the forecasting model based on rough sets and improved BP network analysis is valid and practical. It is especially appropriated for forecasting urban annual and monthly water demand, which is affected by many influencing factors.