宁夏青铜峡灌区年退水量时间序列预测模型研究

摘要

In order to address the prediction problem of return water in Qingtongxia irrigation area,this paper ana-lyzed the properties of annual return water volume and built a prediction model by time series method.The research re-sults showed that the return water volume exhibited a correlation pattern between years.Its time series was a non-white noise and non-stationary series,and its first-order differential sequence was a non-white noise and stationary series.As a result,a time series prediction model of annual return water volume had been established with an average relative simula-tion error of 5 .66% and a relative prediction error up to 5%,demonstrating the high simulation and prediction accuracy of this model.Thereby,it could be used for the prediction of annual return water in irrigation area.%针对宁夏青铜峡灌区年退水量预测问题，采用时间序列方法分析了灌区年退水量特性，建立了预测模型，结果发现青铜峡灌区年退水量在年际间相互关联，其时间序列是一个非白噪声非平稳时间序列，一阶差分序列是一个非白噪声平稳序列，建立了 ARIMA 年退水量时间序列模型，模型模拟的平均相对误差为5．66％，预测的相对误差在5％以内，精度较高，可以用于灌区退水量的预测。