基于差分法及神经网络的硐室围岩力学参数反分析

Back analysis of mechanics parameters of rocks surrounding openings on the basis of calculus of difference and neural network

结合现场实践,基于差分法和人工神经网络建立了新的围岩力学参数反分析方法。进行了硐室模型简化;针对不同力学参数,用FLAC差分程序进行求解,得出相应的神经网络分析样本;用人工神经网络对围岩力学参数进行了学习训练,并进行了网络结构及参数优化;利用现场实测位移值,对围岩力学参数进行了反分析,与现场实测值对比表明其满足误差精度要求,证实了反分析方法的正确性和可靠性。

Aiming at practice in site, the new back analysis method of mechanics parameters of surrounding rock was established by calculus of differences and artificial neural networks. The model of calculation of cave was simplified. Aiming at different parameters, using FLAC program to calculate, the related analytical samples of neural networks was given. Training was conduced by artificial neural networks, the training parameters was optimized. By the on-the-spot survey displacements, mechanics parameters of surrounding rock were forecasted and contrasted, the precision of calculated results were satisfied. It confirmed correctness and reliability of back analysis method.