论文部分内容阅读
分析了一种结合小波变换、粗糙集算法和支持向量回归机的滑坡位移预测方法.该方法依据白家包滑坡位移监测数据,利用小波变换将典型监测点的累计位移分解为趋势项位移和周期项位移,通过曲线拟合获得趋势项位移预测函数,使用粗糙集算法筛选滑坡位移影响因子集,将挑选出来的因子集作为支持向量回归机的输入因子集,从而得到滑坡位移预测结果.应用结果显示:预测平均绝对误差及相对误差分别为3.56cm和5.5%,均方差为1.58cm,相关系数为0.97,表明该预测方法的预测结果能够很好地体现滑坡位移的发展变化趋势,具有较强的预测能力,是一种精确、有效、实用的滑坡位移预测方法.
A method of landslide displacement prediction combining with wavelet transform, rough set algorithm and support vector regression is analyzed. According to the displacement monitoring data of Baijiaba landslide, the cumulative displacement of typical monitoring points is decomposed into the displacement and period Term displacement prediction function is obtained by curve fitting, and the set of influence factors of landslide displacement is screened by rough set algorithm, and the selected set of factors is used as input factor set of support vector regression machine to obtain landslide displacement prediction results. The results show that the predicted average absolute error and relative error are 3.56cm and 5.5% respectively, the mean squared error is 1.58cm and the correlation coefficient is 0.97, which shows that the forecasting result can well reflect the trend of development and displacement of landslide displacement, Is an accurate, effective and practical prediction method of landslide displacement.