论文部分内容阅读
针对BP网络收敛速度慢、网络结构难确定、过拟合等缺点,提出将极限学习机(ELM)应用于区域地应力场反演计算中。利用FLAC3D软件进行正分析平衡计算,以实测点处应力值为网络输入,边界条件和岩石力学参数为网络输出,建立ELM反演模型。将该模型应用于某水电站地下厂房区地应力场反演分析中,计算结果与实测应力值拟合较好,说明该模型简单、实用,具有良好的泛化性和反演精度,可满足工程设计要求。研究成果为初始地应力场的反演计算提供了一种新思路。
Aiming at the disadvantages of slow convergence of BP network, difficult network structure and over fitting, this paper proposes to apply Extreme Learning Machine (ELM) to the calculation of regional geostress field inversion. FLAC3D software is used for positive analysis and balance calculation, and the stress value at the measured point is the network input, the boundary conditions and the rock mechanics parameters are network output, and the ELM inversion model is established. The model is applied to the in-situ stress field inversion analysis of the underground powerhouse of a hydropower station. The calculated results fit well with the measured stress values, which shows that the model is simple, practical, has good generalization and inversion accuracy, Design requirements. The research results provide a new idea for the initial calculation of in-situ stress field.