CN-122016449-A - Parameter probability inversion method and system based on SPH water-soil coupling erosion

CN122016449ACN 122016449 ACN122016449 ACN 122016449ACN-122016449-A

Abstract

The invention belongs to the technical field of geotechnical engineering numerical simulation and parameter identification, and discloses a parameter probability inversion method and system based on SPH water-soil coupling erosion, wherein the method constructs a probability inversion frame combining smooth particle fluid dynamics (SPH) and Bayes-Markov Chain Monte Carlo (MCMC) method, simulates water-soil interface large deformation and dynamic evolution in the water-sand two-phase flow erosion process through the SPH method, adopts Bayes theory to fuse priori knowledge and observation data, the posterior probability distribution estimation of key soil parameters (such as initial internal friction angle and erosion rate) is realized by using the MCMC algorithm, the probability prediction from the parameter uncertainty quantization to the erosion model is realized, the reliability of erosion numerical simulation is improved, the dependence on experience parameters is reduced, and the technical problem that the complex water-soil coupling process simulation and the parameter uncertainty quantitative inversion are difficult to be compatible in the existing method is effectively solved.

Inventors

JIANG HUI
FENG YENAN
WANG JINTING
DU XIULI

Assignees

北京工业大学
清华大学

Dates

Publication Date: 20260512
Application Date: 20260131

Claims (10)

1. A parametric probability inversion method based on SPH water-soil coupled erosion, the method comprising: S1, establishing a water-sand two-phase flow erosion numerical model based on a unified SPH frame; s2, constructing a Bayes-MCMC parameter inversion frame; S3, constructing a likelihood function based on physical test observation data, and quantifying the consistency between a predicted result of the water-sand two-phase flow erosion numerical model and the observation data to provide data constraint for Bayesian inference; S3, performing MCMC iterative sampling under a Bayes-MCMC parameter inversion frame; S4, performing convergence diagnosis on the MCMC iterative sampling result, and performing posterior distribution analysis on the converged sample to obtain posterior statistical characteristics of parameters; s5, carrying out erosion probability prediction based on posterior statistical characteristics of the parameters.
2. The method of claim 1, wherein the method of modeling the water-sand two-phase flow washout erosion numerical model based on the unified SPH framework comprises: the water-soil interface dynamic coupling is realized by adopting a uniform SPH method to disperse water-soil two-phase medium and through the normal contact force and tangential viscous force among particles, and a linear time-varying strength model is introduced to describe the strength attenuation of the soil body in the scouring process, wherein an evolution formula is as follows: ; In the formula, For the initial internal friction angle, k is a coefficient representing the decay rate, t 0 is reduced to the moment of initiation of the water flow action, Is the moment of action of the water flow, Is that Internal friction angle at time.
3. The method of claim 1, wherein the method of constructing a bayesian-MCMC parametric inversion framework comprises: Embedding a water-sand two-phase flow scouring erosion numerical model into a Bayes inference framework, and realizing updating of parameter posterior distribution by MCMC sampling, wherein a core Bayes formula is as follows: ; In the formula, For the posterior distribution of the samples, For the parameter vector to be inverted, For the a priori distribution, For the normalization constant(s), For the purpose of experimental observation data, And y is observation data, and is a likelihood function.
4. The method of claim 1, wherein MCMC sampling employs a Metropolis-Hastings algorithm and automated processes of parameter automatic transfer, model invocation and result extraction are implemented by developing a dedicated interface between Python and LS-DYNA solver.
5. The method of claim 1, wherein the convergence diagnosis is evaluated by examining the stationarity of the parameter trajectory, and wherein the posterior distribution analysis comprises calculating a posterior mean, standard deviation, and plotting a posterior probability density function.
6. A parameter probability inversion system based on SPH water-soil coupling erosion, which is used for realizing the method of any one of claims 1-5, and is characterized by comprising a first construction module, a second construction module, a quantization module, an iterative sampling module, a diagnosis and analysis module and a prediction module; the first construction module is used for establishing a water-sand two-phase flow scouring erosion numerical model based on a unified SPH frame; The second construction module is used for constructing a Bayesian-MCMC parameter inversion framework; The quantization module is used for constructing a likelihood function based on physical test observation data, quantizing the consistency between the prediction result of the water-sand two-phase flow erosion numerical model and the observation data, and providing data constraint for Bayesian inference; The iterative sampling module is used for executing MCMC iterative sampling under the Bayesian-MCMC parameter inversion frame; The diagnosis and analysis module is used for carrying out convergence diagnosis on the MCMC iterative sampling result, carrying out posterior distribution analysis on the converged sample and obtaining posterior statistical characteristics of parameters; And the prediction module is used for predicting the erosion probability based on posterior statistical characteristics of the parameters.
7. The system of claim 6, wherein the process of modeling the water-sand two-phase flow washout erosion numerical model based on the unified SPH framework comprises: the water-soil interface dynamic coupling is realized by adopting a uniform SPH method to disperse water-soil two-phase medium and through the normal contact force and tangential viscous force among particles, and a linear time-varying strength model is introduced to describe the strength attenuation of the soil body in the scouring process, wherein an evolution formula is as follows: ; In the formula, For the initial internal friction angle, k is a coefficient representing the decay rate, t 0 is reduced to the moment of initiation of the water flow action, Is the moment of action of the water flow, Is that Internal friction angle at time.
8. The system of claim 6, wherein the process of constructing a bayesian-MCMC parameter inversion framework comprises: Embedding a water-sand two-phase flow scouring erosion numerical model into a Bayes inference framework, and realizing updating of parameter posterior distribution by MCMC sampling, wherein a core Bayes formula is as follows: ; In the formula, For the posterior distribution of the samples, For the parameter vector to be inverted, For the a priori distribution, For the normalization constant(s), For the purpose of experimental observation data, And y is observation data, and is a likelihood function.
9. The system of claim 6, wherein MCMC sampling employs a Metropolis-Hastings algorithm and implements automated processes of parameter delivery, model invocation, and result extraction by developing a dedicated interface between Python and LS-DYNA solver.
10. The system of claim 6, wherein the convergence diagnosis is evaluated by examining the stationarity of the parameter trajectory, and wherein the posterior distribution analysis includes calculating a posterior mean, standard deviation, and plotting a posterior probability density function.

Description

Parameter probability inversion method and system based on SPH water-soil coupling erosion Technical Field The invention belongs to the technical field of geotechnical engineering numerical simulation and parameter identification, and particularly relates to a parameter probability inversion method and system based on SPH water-soil coupling erosion. Background Water and sand erosion is a key physical process in geological disasters such as river evolution, coastal erosion, dam break and the like. The accurate prediction of water and sand erosion behavior is highly dependent on the rationality of the values of key mechanical parameters (such as initial internal friction angle and erosion rate) of soil. However, these parameters are affected by factors such as soil space variability, test cost, measurement errors, and the like, and there is significant uncertainty. Most of traditional erosion models are based on empirical formulas or semi-empirical formulas, and although the models can reflect the basic rule of erosion to a certain extent, the physical mechanism in the erosion process is difficult to accurately describe, and uncertainty of parameters cannot be effectively quantified. The existing scour erosion research based on SPH mainly focuses on model framework construction, algorithm efficiency optimization and physical process reproduction, but for determining key parameters in a model, the determination still usually depends on limited physical test data or engineering experience values. The method for relying on the experience parameters has obvious limitations that the cost of acquiring the parameters of the specific working conditions through experiments is high and the representativeness is limited, experience values often have subjectivity, inherent space variability and uncertainty of the parameters are difficult to objectively reflect, and the existing method cannot consider the parameter uncertainty in a model prediction framework and cannot quantify the reliability of the critical prediction index of erosion. Disclosure of Invention In order to solve the problems of high uncertainty of soil parameters, dependence on experience values, difficulty in quantification and prediction reliability and the like in the existing erosion model, the invention provides a parameter probability inversion method and a system based on SPH water-soil coupling erosion, the method constructs a probability inversion framework combining smooth particle fluid dynamics (SPH) and a Bayesian-Markov Chain Monte Carlo (MCMC) method, the SPH method is used for simulating large deformation and dynamic evolution of a water-soil interface in the erosion process of water-sand two-phase flow, the Bayesian theory is adopted to fuse priori knowledge and observation data, and the MCMC algorithm is utilized to realize posterior probability distribution estimation of key soil parameters (such as initial internal friction angle and erosion rate), so that probability prediction from parameter uncertainty quantization to erosion model is realized, the reliability of erosion numerical simulation is improved, the dependence on experience parameters is reduced, and the technical problem that the complex water-soil coupling process simulation and the parameter uncertainty quantitative inversion are difficult to be compatible in the existing method is effectively solved. In order to achieve the above object, the present invention provides the following solutions: A parametric probability inversion method based on SPH water-soil coupled scour erosion, the method comprising: S1, establishing a water-sand two-phase flow erosion numerical model based on a unified SPH frame; s2, constructing a Bayes-MCMC parameter inversion frame; S3, constructing a likelihood function based on physical test observation data, and quantifying the consistency between a predicted result of the water-sand two-phase flow erosion numerical model and the observation data to provide data constraint for Bayesian inference; S3, performing MCMC iterative sampling under a Bayes-MCMC parameter inversion frame; S4, performing convergence diagnosis on the MCMC iterative sampling result, and performing posterior distribution analysis on the converged sample to obtain posterior statistical characteristics of parameters; s5, carrying out erosion probability prediction based on posterior statistical characteristics of the parameters. Preferably, the method for establishing the water-sand two-phase flow scouring erosion numerical model based on the unified SPH framework comprises the following steps: the water-soil interface dynamic coupling is realized by adopting a uniform SPH method to disperse water-soil two-phase medium and through the normal contact force and tangential viscous force among particles, and a linear time-varying strength model is introduced to describe the strength attenuation of the soil body in the scouring process, wherein an evolution formula is as follows: