Research On Data Analysis And Prediction Method Of Loess Landslide Deformation Monitoring Based On GNSS Technology

With the increasing human engineering activities in the loess area of China,natural disasters,mainly landslides,occur frequently.In order to reduce the losses caused by landslide disasters on the loess,it is necessary to carry out monitoring analysis and prediction and warning of the safety and stability of the landslide body.jobs.GNSS technology has become the most efficient technology in landslide monitoring.However,its signal is susceptible to multi-path effects and cannot accurately predict the deformation law of landslides.In the study of loess landslide deformation prediction,the relationship between the certainty and randomness of the landslide system is not considered.Therefore,this paper takes the loess landslide in Jingyangmiaodian,Shaanxi as an example,applies GNSS technology to the monitoring of loess landslide deformation,and designs corresponding monitoring technical schemes.At the same time,it analyzes and evaluates a variety of landslide deformation prediction models based on chaos theory for this type of loess.The applicability of the landslide,and using the measured data of Jingyangmiaodian landslide to evaluate the prediction accuracy of the model.The main research contents and results of this article are as follows:(1)In view of the shortcomings of traditional geodetic technology,this paper uses GNSS technology with high accuracy and can provide real-time positioning results to monitor Jingyang Miaodian landslide.In the design of the GNSS monitoring plan,the monitoring method combining static relative positioning and real-time dynamic monitoring is mainly used to obtain the deformation monitoring data of the loess landslide in Jingyang Miaodian.(2)In view of the shortcomings of the wavelet filtering method,this paper proposes a noise suppression method for deformation monitoring data based on S-transform.Simulation data and measured data of Jingyang Miaodian landslide deformation were used to verify the effectiveness of the method.The results show that,compared with the wavelet filtering method,the deformation data processed by the filtering method based on S-transform are superior in RMSE and SNR,and can accurately extract the deformation characteristics of the monitoring points,and provide reliable monitoring data for landslide deformation prediction To improve the accuracy of landslide deformation prediction.(3)In view of the existing research on the prediction method of loess landslide deformation without considering the relationship between the randomness and certainty of the landslide system,chaos theory is used to predict the deformation of loess landslide.The original GNSS landslide monitoring sequence and the time series after S-transform noise suppression are used to solve their respective phase space reconstruction parameters,and chaos identification is carried out.The results show that both the original GNSS sequence and the GNSS time sequence after S-transform noise suppression satisfy the chaotic characteristics.(4)Different chaotic time series prediction methods are used to predict the deformation data of the loess landslide in Jingyang Miaodian and the deformation data processed by S-transformation filtering,respectively,to verify that the filtered data has a better prediction effect.The results show that the prediction result of the time series after S-transform noise reduction is closer to the actual value,and MAE and MRE are better than the index evaluation values of the original time series.Among the various prediction methods,the prediction accuracy of the BP neural network prediction method is good.The average absolute error predicted by the original time series is 0.3993 mm,and the average relative error is 11.9%.The average absolute error of the time series prediction after S-transform noise suppression is 0.1416 mm,and the average relative error is 4.12%.The comparison shows that the noise in the original data has a greater impact on the prediction results,and the prediction effect is significantly improved after the noise suppression process.