Displacement Back Analysis Of Tunnel Surrounding Rock Based On Improved Limit Learning Machine

At present,the tunnel information construction technology is widely praised,which is essential is how to accurately determine the mechanical parameters of the surrounding rock.This paper is aimed at the shortcomings in intelligent back analysis of surrounding rock displacement of the tunnel,that the traditional learning training algorithm is slow,and so easy to fall into local minimum,the Particle swarm optirmization(PSO)is used to improve the extreme learning machine(ELM)to obtain PSO-ELM algorithm.Based on the PSO-ELM algorithm,the back-analysis model of the surrounding rock displacement of the tunnel is established with the geological condition,geological prospecting data,design information and industry standard of Yingpan Mountain tunnel.Based on the uniform experimental method and orthogonal experiment method,combined with the geological condition,geological prospecting data,design data and industry standard of Yingpan Mountain tunnel,the numerical test samples of the mechanical parameters of tunnel rock are designed The FLAC3D was used to simulate the tunnel excavation and support process to calculate the displacement of the surrounding rock corresponding to the mechanical parameters of each tunnel.-ELM algorithm for learning samples and test samples.The PSO-ELM algorithm with strong regression performance is established by using the particle swarm optimization algorithm to optimize the value and threshold value by using the global optimization ability of particle swarm optimization algorithm.PSO-ELM algorithm and BP neural network are implemented with MATLAB as the platform.PSO-ELM algorithm and BP neural network are trained by learning samples.The back-analysis model of tunnel rock displacement based on PSO-ELM algorithm and BP neural network is established respectively The feasibility and accuracy of the PSO-ELM algorithm in the back analysis of the tunnel rock are analyzed by comparing the BP neural network with the inverse solution and the direct displacement analysis of the PSO-ELM algorithm.Select 5 Yingpanshan Tunnel V Grade Rock sections,each section of the dome subsidence monitoring value and the value of the input level of convergence monitoring PSO-ELM tunnel surrounding rock displacement inverse analysis model,to calculate the corresponding tunnel rock mechanical parameters,the into the tunnel wall rock mechanical parameters FLAC3D numerical calculation,the sedimentation value is determined vault and horizontal convergence value of each section,and compared with the monitored values,results showed that PSO-ELM surrounding rock displacement inverse analysis model built herein is reliable,and it has some reference value for the value of the surrounding rock parameters in the information construction of the tunnel...