摘要
根据城市用水量的影响因素及特点，利用统计预测理论，建立了日用水量的实用动态模型。通过逐步回归分析方法剔除次要影响因素，并采用递推最小二乘（ＲＬＳ）方法动态预测回归残差项。经沈阳市实例验证，该法预测误差小，可满足供水系统调度的实际需要。
In accordance with the effective factors and characteristics,a practical dynamic algorithm was advanced for daily water municipal consumption based on statisic and predictive theories.The sequential regress analysis was adopted to screen off secondary factors,and the recursive least square technique (RLS) was used to estimate innovation coefficients of the time-varying-parameter autoregressive model of order 3 dynamically.By the experiments at the Shenyang city,it was proved that the forecasting errors of the algorithm is little and can meet the practical requirement of water-supply dispatch system.
出处
《哈尔滨建筑大学学报》
EI
1998年第3期39-44,共6页
Journal of Harbin University of Civil Engineering and Architecture
基金
国家自然科学基金
关键词
城市用水
预测
逐步回归分析
动态模型
用水量
municipal water
forecasting
sequential regress analysis
recursive lesat squares algorithm(RLS)
practical dynamic algorithm
