采用小波-差分指数平滑方法对三峡库区八字门滑坡地表位移进行了预测。小波变换以其多分辨率分析的特性,广泛应用于信号分析,指数平滑方法是在移动平均法基础上发展起来的一种时间序列分析预测法,在预测分析中应用广泛。针对三峡库区八字门滑坡的GPS监测水平累积位移,采用小波变换对时间序列进行去噪预处理,并分别建立一阶差分和二阶差分指数平滑模型对其进行预测,预测得到的平均绝对误差百分比仅为1.5%和1.1%,取得了非常好的效果,该模型可以应用于实际工程。
The wavelet- difference index smoothing method was used to predict cumulative displacement of the Bazimen landslide located in Three Gorges Reservoir area. Wavelet transform is widely used in signal analysis because of its multi resolution analysis,while the index smoothing method is a kind of time series analysis method,which is developed based on the moving average method and commonly applied in prediction analysis. Aiming at the accumulative displacement data of the landslide,the time series is denoised by the wavelet transform in pretreatment,and the first difference- index smoothing model and second difference- index smoothing model are established to predict the accumulative displacement of the landslide. The results show that the MAPEs( Mean Absolute Percent Error) are 1. 5% and 1. 1%,demonstrating that this model can be applied to practical engineering.