基于灰色系统与线性回归方法的水质预测

摘要

To develop an efifcient method, this paper selects the “Least Square Fit” method to ift the initial data, and selects the corresponding optimal prediction method by the value of R2. To verify the feasibility of the method, both approaches have been tested by the data from a long-term groundwater monitoring points in Shijingshan District, Beijing. By applying the grey system method, the relative error for total hardness is 2.62%, which is 5.65% lower than the result of linear regression method. The relative error for TDS is 2.33% by grey system method, which is 0.61% lower than the result of linear regression method. The results are similar to that by the value of R2. Namely, it is a rational and efifcient technical method to select the suitable approach for groundwater quality prediction .%以北京市石景山区某地地下水监测点多年监测数据为例，进行适宜性预测方法的验证。其中，总硬度灰色预测值的平均相对误差为2.62%比线性预测低5.65个百分点；溶解性总固体线性预测值的平均相对误差为2.33%比灰色预测低0.61个百分点，与通过拟合度R2值大小选取的最优预测方法一致，表明通过拟合度R2值大小来选取合理的预测方法是一种便捷、合理的技术手段。