Study On Land Subsidence And Predictive Model In The Unsymmetrical Loading Section Of Gaoligou Tunnel

Nowadays the excavation of mining tunnels is becoming more and more frequent.The engineering problems caused by the excavation of unsymmetrical loading tunnels are common.Among them,the unsymmetrical loading tunnel caused by topographic factor,which is often shallow buried and close to the tunnel entrance.This kind of tunnel not only subject to unsymmetrical pressure,and the surrounding rock near the earth's surface,which is relatively loose and broken and the strength is low after long-term weathering and other factors,and it is difficult to form self-stabilization,so the probability of engineering problems and even accidents caused by improper tunneling construction is a little large.And the land subsidence is an important index to judge the stability of surrounding rock when tunneling,which is easy to monitor in real time.However,due to the complex shape of chamber convergence,the land subsidence have obvious asymmetry and it is difficult to calculate directly by theoretical method.Therefore,it has certain theoretical guiding significance and engineering safety value to study the land subsidence law and its related researches of the unsymmetrical loading tunnel with shallow buried depth.Based on the engineering background in the unsymmetrical loading section of Gaoligou tunnel from Jingyu to Tonghua section of Heda Expressway,the land subsidence caused by tunnel excavation was studied through a series of methods mainly including field monitoring,finite element numerical simulation,first slope method,quadratic response surface-structure reliability method,grey correlation theory and wavelet neural network prediction model.(1)The buried depth and unsymmetrical pressure properties in the unsymmetrical loading section of the left and right line of Gaoligou tunnel were analyzed,and the main influencing factors of land subsidence caused by its excavation were further deduced and analyzed by using the existing formula of stochastic medium theory.According to the engineering background,the finite element numerical calculation model of the right-line section was established,combined with the actual observation results of land subsidence,the rationality of the numerical calculation model was verified,and the characteristics of land subsidence in the excavation of unsymmetrical loading tunnel were analyzed.(2)This paper analyzed the influence of the degree of unsymmetrical pressure on the deformation and settlement of surrounding rock caused by tunneling.On this basis,the influence and mechanization of topographic slope,surrounding rock grade and tunneling diameter on land subsidence caused by excavation of unsymmetrical loading tunnel were analyzed respectively.And the influence and mechanization of topographic slope,surrounding rock grade and tunneling diameter on the relation of vault settlement and maximum land subsidence were analyzed.(3)For symmetrical loading and unsymmetrical loading tunnels caused by topographic factor,the sensitivity of the basic mechanical parameters(including elastic modulus,unit weight,angle of internal friction,Poisson's ratio,cohesion,and tensile strength)of surrounding rock to land subsidence caused by tunneling was calculated and analyzed.The sensitivity and grey correlation of the influencing factors(topographic slope,surrounding rock grade and tunneling diameter)for land subsidence were analyzed.(4)For the excavation of unsymmetrical loading tunnel,considering the complex shape of chamber convergence and the obvious asymmetry of deformation settlement of surrounding rock,this paper used neural network algorithm to predict land subsidence.It introduced the idea of variation in the genetic algorithm and puts forward a Particle Swarm Optimization with second-order oscillation and adaptive perturbation aiming at the deficiency of the Second-order Oscillating Particle Swarm Optimization.And by searching for the advantages of two kinds of multi-pole functions,the rationality of the improved algorithm was verified.On this basis,the prediction model of land subsidence in the excavation of unsymmetrical loading tunnel was established by the Improving Particle Swarm Optimization-Wavelet Neural Network.And two types of tunnels consist of overlying surrounding rock with single strata and multi-layers lithology in the unsymmetrical loading sections of the left and right line of Gaoligou tunnel were taken as the engineering background,and the land subsidence under different working conditions was predicted respectively,and the results showed that the algorithm of the Improved Particle Swarm Optimitation-Wavelet Neural Network was better than the Standard Wavelet Neural Network.