| In order to meet the rapid development of the economy and society,the construction technology and construction standards of the transport infrastructure are now more demanding,and road tunnels are gradually developing into complex conditions with multiple holes,large spans and shallow burial.The conventional observational and empirical methods for optimising construction solutions and predicting the stability of complex tunnels are highly subjective and lack effective theoretical support,and the existing theoretical analysis methods have certain limitations in determining the stability of tunnels for specific complex working conditions and considering specific construction processes,making it difficult to meet engineering requirements.In this paper,a typical large-span four-hole small clearances road tunnel project in Fujian Province is used as the research vehicle to carry out the analysis of the mechanical characteristics of tunnel construction,optimise the construction plan of the tunnel,construct a spike mutation model to determine the stability of the surrounding rock,and finally establish a particle swarm optimisation-based LSTM model to determine the deformation trend and stability of the surrounding rock.The main findings are as follows.The main research contents and results are as follows.Research on the construction plan of four-hole tunnel cavity section based on FLAC 3D technology.In order to investigate the mechanical characteristics and stability of the cavern section during the excavation of the shallowly buried biased four-hole tunnel,a three-dimensional mechanical model was constructed using FLAC 3D technology to analyse the deformation characteristics of the shallowly buried biased four-hole tunnel under different excavation methods and excavation sequences,to reveal the deformation influence law of the surrounding rock considering the construction effect and to evaluate its rock stability.The results show that the right two holes are excavated first before the left two holes,and the combined excavation method of double sidewall method for vehicular tunnels and step method for pedestrian tunnels results in minimal settlement of the tunnel vault,horizontal convergence and bulging of the arch bottom,with good overall tunnel stability.The results of the calculations are basically consistent with the measured results,verifying the correctness of the numerical model and the rationality of the construction scheme,and providing practical guidance for the actual project.Stability analysis of multi-bore tunnels based on cusp mutation theory.In order to further investigate the stability of the multi-bore tunnel from a theoretical perspective,a theoretical model was constructed to extract the numerical simulation results of the four chambers under different working conditions,to analyse the characteristic values of the tunnel under various working conditions and to evaluate its stability.The results show that the solution of excavating the right two caverns first and the left two caverns later,using the double sidewall method for vehicular tunnels and the step method for pedestrian tunnels,can ensure that the eigenvalues of the cusp mutation based on vault settlement and horizontal convergence meet the requirements at the same time,and that all four tunnels are stable.The results of the study provide a theoretical basis for the evaluation of the stability of the multi-bore tunnel and the optimisation of the solution.Particle swarm optimisation based long and short-term memory network algorithm(PSO-LSTM)deformation prediction study.A particle swarm optimisation-based long and short-term memory network algorithm(PSO-LSTM)was developed to predict and evaluate the deformation of the surrounding rock in complex geological and terrain conditions,which makes monitoring difficult and the monitoring data may be lost.The results show that the optimised LSTM model has better prediction results than the traditional LSTM model.The analysis of the predicted deformation curves based on numerical simulations and actual measurements shows that the predicted curves of the PSO-LSTM model follow the same trend as the actual monitoring curves,with good prediction results.The model can provide theoretical support for safe tunnel construction and stability control. |