基于BP人工神经网络的引水系统平硐围岩参数反演分析

Parameter Inversion Analysis of Adit Surrounding Rock Mass of Water Diversion System Based on BP Neural Network

鉴于影响围岩稳定性的一个重要因素是岩体的开挖、卸荷过程引起围岩质量劣化和变形,基于卸荷岩体力学理论及方法,利用BP人工神经网络对丹巴电站平硐围岩参数进行了反演,有效降低了参数选取过程中的主观因素,并将反演所得参数代入三维模型进行数值模拟计算。结果表明,计算所得的测线收敛值变化趋势与监测数据大致相同,可见利用BP人工神经网络能获得反映岩体真实性能的参数值。

The excavation and unloading process of rock mass, which leads to quality degradation and deformation, is one of the important factors that influencing the stability of surrounding rock. On the basis of unloading rock mass mechanics theory and method, combining with the BP neural network,an inversion analysis is made on the adit surrounding rock parameters of water diversion system in Danba power station, which can effectively reduce the subjective factors in parameter selection. The inversed parameter is used to make a calculation with three-dimensional model. The results show that the variation trend of calculated convergence values of survey lines is closed to the monitoring data. Therefore, the parameters reflecting the real performance of rock mass can be obtained by the BP neural network.