A Parameter Intelligent Recognition Method and Numerical Simulation of Large Underground Caverns

Abstract

To gain the equivalent mechanical parameter of rock in underground engineering and to improve the precision of numerical simulation of excavation, a new multi-information intelligent recognition method of rock mechanical parameter is put forward. By coupling artificial neural network and genetic algorithm as a whole algorithm and by building an associated fitness function to absorb multi-information, a combinatorial intelligent optimization method is available for parameter recognition. The method realizes the global optimization efficiently and improves the uniqueness of searched parameter as far as possible. After recognizing the equivalent mechanical parameters of rock in Jinping II underground caverns by above method, the performance of surround rock in construction is calculated by 3D numerical simulation with the characters of 1168 thousand elements and 234 gridpoints. The practical application indicates that the intelligent method is reliable to analyze and predict the safety of underground engineering.