Research On Resistivity Imaging Technology Of Landslide Geological Disaster Monitoring Based On IMM-UKF

Landslide is one of the most serious geological disasters in China in terms of casualties,economic losses and total amount.It has the characteristics of complexity,suddenness and openness.Compared with other natural disasters,the means of monitoring and early warning are more complex.The existing landslide geological disaster monitoring methods have different advantages and disadvantages.GNSS,In SAR and UAV close range imaging technology are the main methods to monitor the shallow displacement of landslide.However,the size,speed and direction of the shallow displacement of landslide are not completely consistent with the deep deformation.The monitoring model based on the law of shallow displacement of landslide can not accurately reflect the deformation characteristics of deep rock and soil.The time series model based on borehole deep displacement monitoring technology,combined with hydrogeology,tectonic stress analysis and other measured data can better reflect the deep deformation characteristics and internal geological structure of landslide,and identify the position and deformation trend of slope sliding surface.However,due to the limitation of borehole location,quantity and cost,borehole deep displacement monitoring can not realize the spatial distribution of landslide Complete and continuous displacement data acquisition on the platform.Resistivity imaging technology has a good application in judging the deep deformation law of landslide by building an effective geoelectric model and analyzing the internal electrical characteristics of landslide.However,due to the difference of sensitivity of resistivity imaging device,the inversion imaging of data observed by a single device can not reflect the real structure of landslide.Because of the nonlinearity of landslide geological disasters,the traditional resistivity imaging technology relies heavily on the current time series measurement data,which is difficult to make timely and effective early warning in the early development of landslide geological disasters.Therefore,how to integrate the sensitivity characteristics of different resistivity imaging devices,effectively predict the lead time series resistivity data,and build an efficient landslide geological disaster prediction model is an urgent problem to be solved.The difference of geological structure and electrical characteristics of landslide,as well as the mutual coupling effect between internal and external uncertain dynamic factors,make the deep displacement of landslide present nonlinear and uncertain changes with time.Aiming at the problem of error caused by linearization of nonlinear model,this paper uses unscented Kalman filter to filter the resistivity imaging data,and uses unscented transformation to determine the resistivity probability density,so as to improve the prediction speed and ensure the data accuracy.On the premise of overcoming the error caused by the linearization of nonlinear model,the monitoring and prediction of the electrical characteristics of the whole deformation process in the landslide geological body are realized.There are great differences in inversion depth,boundary loss and resolution of different geological structures for various devices of resistivity imaging technology.In this paper,based on the interacting multiple model filtering algorithm,Markov transition probability is used to control the fusion of different resistivity imaging device models.On the premise of maintaining the original inversion imaging ability,the inversion data set is effectively compressed,and the data error and prior information are balanced to achieve fast,accurate and stable inversion.The problem of sensitivity matching between resistivity imaging devices is solved.In this paper,the resistivity imaging technology based on interactive multiple model unscented Kalman filter algorithm can effectively improve the accuracy of landslide geological disaster monitoring and prediction,and realize the fast and effective prediction imaging of landslide geological mass internal structure.