Dynamic Construction Method of Landslide Uncertain Model Based on Multi-source Heterogeneous Data Fusion

Abstract

The present invention provides a method for dynamically constructing an uncertain landslide model based on the fusion of multi-source heterogeneous data, including: joint use of geological survey, data collection, indoor mechanical test, CT scanning and numerical sample random reconstruction technology to obtain landslide rock and soil Spatial variation information of body mechanics parameters; on the basis of statistics on the spatial laws of rock and soil mechanics parameters, construct an initial uncertain landslide model; use non-invasive random finite element method to carry out numerical simulation of landslides under corresponding working conditions; construct different time The intelligent response surface model of point landslide output information and random variables, based on the intelligent response surface model and the reliability Bayesian inversion method of adaptive conditional sampling, uses the landslide direct information and monitoring indirect data to update the uncertain model in real time, thereby establishing a multi- A landslide uncertainty model for data fusion from heterogeneous sources. The beneficial effect of the invention is that a fine landslide geological model is established, which provides an effective means for subsequent mechanism analysis and stability evaluation of landslides.

Description

technical field [0001] The invention relates to the fields of landslide uncertainty analysis and stability evaluation, in particular to a method for dynamically constructing landslide uncertainty models based on multi-source heterogeneous data fusion. Background technique [0002] The accurate establishment of complex geological-mechanical models is an important basis and premise for carrying out landslide mechanism research and stability evaluation. Under multi-phase and multi-type internal and external dynamic geological actions, the material composition and slope structure of landslides have a large discontinuity. The complex slope structure and the uneven distribution of materials make the rock and soil parameters of different parts of the landslide have randomness, variability and structure. Obviously, considering the complexity of landslide evolution and the randomness, structure and time variability of rock and soil parameters, it is necessary to establish a geologica...

AI Technical Summary

Problems solved by technology

At present, the establishment of landslide average geological model only uses the previous direct information, and the data utilization rate of geotechnical parameters is low. In the process of establishing the common uncertain model, only direct information or indirect information is used less. The uncertainty of the model is large, and the amount of calculation is large during the model operation, and the real-time dynamic update of the landslide uncertainty model is not comprehensively used.

[0005] The input Bayesian method based on the posterior information can invert and update the probability distribution function of the prior information, and can better establish the relationship between the prior parameter model and various monitoring data. For inversion update, MH-based Markov Chain Monte Carlo (MCMC) is often used to obtain samples that obey the posterior probability distribution, but it is mostly suitable for low-dimensional Bayesian update problem, and the proposed function distribution needs to be given

The prior parameter model can be a simple random variable model or a random field model, and the random field model has high-dimensional variables. The traditional MCMC method cannot effectively solve this problem.

The spatial variation model is combined with landslide engineering through the stochastic finite element method. However, it is time-consuming and computationally intensive to obtain the posterior data under long-period conditions through the finite element numerical model, which limits the update efficiency of the uncertain model. And for the timeliness of monitoring information, it is necessary to continuously update the uncertainty model

Images