基于加权马尔可夫链的地下水动态预测

A Prediction Model of Regional Groundwater Based on Weighted Markov Chain

为了降低地下水变化中的季节性影响,对实测样本序列进行了分月标准化处理,以此地下水标准化指标为基础,考虑不同滞时的地下水埋深为相依随机变量的特点,用归一化的各阶自相关系数计算权重,应用加权马尔可夫链构建了地下水动态预测模型。以赤峰市中心城区2000年至2013年逐月地下水埋深实测序列为基础,实现了赤峰市中心城区预见期为1个月的地下水动态预测。结果表明:通过对原始时间序列进行分月标准化处理,有效降低了地下水变化过程中季节性的影响,使得以月为时间尺度的地下水预测成为可能;该模型依据不同滞时变量的相依关系计算权重,充分利用了原时间序列的信息;建模过程简单,所需资料较少,拟合精度较高,对于地下水预测具有较强的适用性。

In order to diminish the seasonal effect on the groundwater variation,the standardized index was proposed by standardizing the original sample series related to each month. Based on the index,standardized autocorrelation coefficients were considered as weights due to the dependency relationships among the groundwater variables with different leading time. The weighted Markov chain was used to building a dynamic prediction model of regional groundwater. Then this prediction model was applied to the central urban area of Chifeng City with the monthly groundwater depth data during the period from 2000 to 2013. The results show that the forecast with 1 month leading time can be achieved by using the model constructed in this paper. Moreover,it is proved that the model with high accuracy is simple and feasible to build. Therefore it is applicable in groundwater forecast.