调度时用水量预测的分时段混沌建模方法

针对传统调度时用水量需求预测方法在预测结果精度及稳定性方面存在的困难,鉴于分时段时用水量序列的最大预测时间尺度远大于连续时序列的混沌理论分析结果,利用混沌相空间近邻预测建模理论,提出了时用水量向前24 h预测的分时段混沌建模方法.该方法经实例验证,向前24 h预测结果稳定性较好,误差较小(平均相对误差小于2%,最大相对误差小于5%),可满足供水优化调度的需要.

Divided-period chaotic forecasting model of hourly water consumption for optimal operation

In allusion to the difficulties existing in classical methods for forecasting operational demand, a divided-period chaotic forecasting model of hourly water consumption 24()h in advance was developed based on the theory of phase space reconstruction. The chaotic analysis shows that the maximum predictable time scale of divided-period water consumption time series is much larger than that of continuous time series, the hourly water consumption series discussed were the series of 24()h intervals but not that of the usual one-hour interval. Testing by examples shows that the prediction results are stable and the prediction (errors) are small(average absolute value of relative error less than 2% and the maximum absolute value of relative error less than 5%), which can satisfy the demands of optimal operation of water distribution (system.)