Intelligent back analysis of rock mass creep parameters for large underground caverns under high in-situ stress based on incremental displacement

The creep mechanical parameters inaccuracy of surrounding rock under high in-situ stress have become the “bottleneck” problem on the theoretical analysis and numerical calculation,therefore,determining the reliable rock mass creep parameters is urgently needed to be solved in practical engineering. An intelligent back analysis method of rock mass creep parameters based on the displacement increment sensitivity analysis for large underground caverns un- der high in-situ stress is put forward in this paper. This method is to use the fractional order calculus creep constitutive model which can describe the failure characteristics of surrounding rock creep over time under the high ground stress.Taking the displacements of arch roof,arch shoulder and sidewall as characteristic indices,five creep parameters (in- stantaneous shear modulus,viscosity coefficient,viscoelastic shear modulus,Viscoelastic coefficient,β) under inversion are confirmed by sensitivity analysis. Displacement increment data were extracted following the principle of “great val- ue” and “sensitivity”. Fitness function of displacement increment is established by determining the inversion creep pa- rameters inputting the eight incremental displacement monitoring information from the main power house and main transformer chamber. The construction of different level combination of learning samples and training samples of each parameter by using uniform design method,parameters value is confirmed by adopting genetic algorithm-neural network searching in the global space. Finally the comparative analysis on displacement increment measured values and calcu- lated values is conducted through grey correlation analysis method and posteriori difference method. The results from underground powerhouse of Jinping II hydropower station show that the surrounding rock mass creep parameters are ac- curate,validated and rational. It also provides a new method for parameters determination during the long-term stability evaluation of large cavern under high in-situ stress.