基于小波分析的地铁基坑沉降预测模型研究

Research on Prediction Model of Subway Foundation Pit Settlement Based on Wavelet Analysis

为了提高地铁基坑沉降预测精度,准确把握地铁基坑沉降变形趋势,本文在支持向量机(Support Vector Machine,SVM)模型基础上引入小波分析与遗传算法(Genetic Algorithm,GA),提出了一种基于小波分析的GASVM预测模型。首先利用小波分析良好的去噪性能对原始观测数据进行去噪处理;其次使用遗传算法优化的SVM模型对去噪后数据进行建模与预测。以某地铁沉降监测数据为例进行相关实验,结果表明,在使用小波分析去噪时,在分解层数为1、Sym4小波基函数、阈值准则为rigrsure、scal=sln时的去噪效果最好。同时,相较于单一的SVM模型与基于小波分析的SVM模型,本文提出的基于小波分析的GASVM模型的预测精度更高,且预测精度不会随着预测期数的增加有较为明显的降低,具有较高的实际应用价值。

In order to improve the accuracy of subway foundation pit settlement prediction and accurately grasp the settlement deformation trend of subway foundation pit,this paper introduces wavelet analysis and genetic algorithm(GA)based on support vector machine(SVM)model,and puts forward a GASVM prediction model based on wavelet analysis.Firstly,the original observation data are de-noised by using the good de-noising performance of wavelet analysis;Secondly,the SVM model optimized by genetic algorithm is used to model and predict the denoised data.Taking the settlement monitoring data of a subway as an example,the results show that when wavelet analysis is used for de-noising,the de-noising effect is the best when the number of decomposition layers is 1,Sym4 wavelet basis function,threshold criterion is rigrsure and scale=sln.At the same time,compared with single SVM model and SVM model based on wavelet analysis,the prediction accuracy of GASVM model based on wavelet analysis proposed in this paper is higher,and the prediction accuracy will not decrease significantly with the increase of prediction periods,so it has high practical application value.