基于RBF神经网络的埋藏式月牙肋钢岔管围岩承载率研究

Research on surrounding rock bearing rate of underground crescentrib reinforced steel bifurcation based on RBF neural network

根据西龙池抽水蓄能电站的工程资料,建立埋藏式月牙肋钢岔管有限元模型,利用接触单元模拟钢岔管和围岩的联合承载。通过构造一系列围岩参数依次进行联合承载有限元计算,得到测点的计算应力。依据测点的计算应力和实测应力等数据,进行神经网络反演分析,根据反演结果的误差大小确定围岩参数。研究表明,通过神经网络反演得到的围岩参数较为合理,符合工程经验,且围岩承载规律性较好,证明利用神经网络法对钢岔管主要设计参数进行反演分析是可行的。

Based on the design data of Xilongchi Pumped Storage Power Station,the finite element model of underground crescent-rib reinforced steel bifurcation is established,and the jointly bearing mechanism of steel bifurcation and surrounding rock is simulated by contact element.Then,a series of parameters of surrounding rock are constructed to carry out finite element calculation of jointly load-bearing in turn,and the calculated stress of the measured points is obtained.According to the calculated stress and measured stress of the measuring point,the neural network inversion analysis is carried out,and the actual surrounding rock parameters are determined according to the error of the inversion result.The research shows that the parameters of surrounding rock obtained by neural network inversion are reasonable,in line with engineering experience,and the surrounding rock bearing regularity is good,which proves that the inversion analysis of the main design parameters of underground crescent-rib reinforced steel bifurcation by neural network method is feasible.