Reliability of underground caverns based on genetic algorithm and support vector machine

The genetic algorithm (GA) and support vector machine (SVM) are applied to analyze the reliability of underground caverns. The explicit form of performance function is established by use of the relative displacement values and relative displacement limit values of surrounding rock. The learning samples of the relative displacement values are built by numerical simulation; then the relative displacement values are predicted by the support vector machine that is optimized by the genetic algorithm. An example in Jinping Hydropower Station is given for illustrating the application of the proposed approach. The new method is proved effective in the reliability analysis of underground caverns.