基于进化神经元算法的堡镇隧道软弱围岩施工弹塑性智能位移反分析

结合宜万铁路堡镇隧道的施工,将BP神经网络和遗传算法引入特长隧道软岩段的施工位移反分析,采用遗传算法自动搜索BP神经网络训练效果最优的参数,建立起反映围岩变形与岩体物理力学参数及初始地应力之间高度非线性、不确定的GA-BP智能模型,然后采用遗传算法在岩体物理力学参数和初始地应力取值范围内,搜索BP神经网络预测围岩变形与实测围岩变形最接近的参数组合,取得反演获得的岩体物理力学参数和初始地应力.从堡镇隧道应用结果来看,这种进化神经元算法反演结果可以满足隧道施工的需要,并为类似工程提供了借鉴.

Elasto-Plastic Intelligence Displacement Back Analysis of Soft Rock Construction in Baozhen Tunnel Based on Evolutionary-Neuron Algorithm

Combined with construction of Baozhen tunnel in railway from Yichang to Wanzhou,ming the BP neural network and genetic algorithm were introduced to the displacement back analysis field of soft rock segment of this tunnel construction in this paper.The genetic algorithm was used to search the optimal BP model parameters during the neural network training procedure formed the GA-BP algorithm,therefore the nonlinear and uncertainty mapping relation among the mechanical parameters,initial ground stress of surrounding rock and deformation has been built.Then,GA was used to search the appropriate parameters of physical mechanics and initial ground stress which can make the calculation displacement by BP algorithm coincide with the measured displacement.This was the mechanical parameters and initial ground stress of surrounding rock.It can be concluded that this evolutionary-neuron algorithm can meet the need of tunnel construction from its application in Baozhen tunnel and offered reference for similar engineering.