论文部分内容阅读
由抽水试验数据应用计算机进行直线图解水文地质参数时,拟合直线的方法是基于最小二乘理论,但该方法易受异常值影响,拟合结果常常不理想,从而导致求解的参数值误差较大。相比最小二乘法,基于信息损失最小化的最频值算法对异常值不敏感,是一种更为稳健的算法。本文将最频值算法与直线图解法耦合,推导出了求解导水系数和贮水系数的新公式。最后,通过MATLAB环境下自行编制的程序,以一个抽水试验为例,对两种直线图解法进行了对比。结果表明,基于最频值算法的直线图解法求参结果更准确可靠。
The method of fitting a straight line is based on the least-squares theory when using a computer to conduct a straight-line graphic interpretation of hydrogeological parameters by using pumping test data. However, the method is susceptible to outliers and the fitting results are often not ideal, resulting in the error of the parameter values Big. Compared with the least square method, the most frequent value minimization algorithm based on information loss is not sensitive to outliers, which is a more robust algorithm. In this paper, the most frequent value algorithm is coupled with the linear graph method to derive a new formula to solve the water conductivity and water storage coefficient. Finally, a program of self-programming in MATLAB is taken as an example to compare two straight-line graphs. The results show that the linear method based on the most frequent algorithm is more accurate and reliable.