Abstract: The analysis speed of commercial softwares cannot meet the timeliness requirements of information construction. In view of this, an entirely new idea is proposed in this paper. Firstly, 4 groups including 15 parameters which have significant impacts on the stability of tunnel are generalized and test plan is drawn up. Subsequently the relational database of numerical analysis results and the main influences is established, and the rapid analysis and evaluation system for underground chamber with double faults is built up with the help of BP neural network. This system is employed to analyze the stability of the powerhouse of Laxiwa hydropower station. The error of peripheral deformation is less than 20%, which meets the engineering requirements and shows a good reliability of this system.

Key words: information construction, chamber with double faults, BP neural network, rapid analysis system