Abstract: The biggest bottleneck in Slope Stability Evaluation System is the actual occurrence state of evaluation factor is discrete, and the interlayer is clear. For this, we took samples of results of rock mechanics laboratory test from about 100 groups as the total sample, and introduced statistical theory to put the membership probabilities of factor under different stable levels as the membership value in the level, which avoiding excessive subjective sense when measured value is selected to participate in the evaluation. There are many methods of determining the weights, but some problems are existing in those . On the basis of the theory of probability and statistics, and using the brainstorming method to summarize the opinions of different experts in mathematical statistics, the representative weight values are obtained, which can take emphasis of different experts into comprehensive consideration and achieve better results. The findings are applied in the fuzzy evaluation system to evaluate strength of the rock slope in the left bank of Rumei power station, and results are better able to reflect the slope stability.

Key words: slope engineering, probability and statistics, membership, weights, fuzzy evaluation