Abstract: In order to scientifically select the structural parameters of underground phosphate mine stopes，the response surface methodology is adopted to carry out multi-level comprehensive optimization of stope structural parameters from the perspectives of safety and economy. Four parameters including stope length and stope width are selected as experimental factors for an underground phosphate mine，and the theoretical levels of each experimental factor are fuzzily fitted according to actual production conditions. Taking stope roof displacement as the safety index and stope direct production cost as the economic index， a response surface methodology test scheme for multi-level safety and economy optimization of stope struc⁃ tural parameters was designed. The Design-Expert software is used to analyze the influences of stope struc⁃ tural parameters and their interaction factors on stope safety，and the key influencing factors of stope safety are obtained. The roof displacement values of 100 groups of stope structural parameter combinations are simulated and predicted，and 69 groups of stope structural parameter combinations with qualified safety were selected. The influences of stope structural parameters and their interaction factors on stope economy are analyzed，and the key influencing factors of stope economy were determined. Furthermore，3 groups of optimal economic combinations are further screened from the safety-qualified parameter combinations. Combining the roof displacement values of each combination，the stope structural parameter combination with stope length 70 m，stope width 16 m，stope height 4.5 m and pillar spacing 10 m is recommended as the comprehensive optimal scheme with good safety and economic benefits. Practical application proves that the multi-level optimization of stope structural parameters by response surface methodology is scientif⁃ic，effective and reliable，which can provide reference and guidance for safe production of mines.

Key words: phosphate mine, stope structural parameters, response surface methodology