Abstract: To study the influence of the randomness and fuzziness of geotechnical parameters on the slope stability, this paper presents a case history of a slate slope of an expressway in Hunan Province. The generalized reliability analysis is performed by using the membership function and the Rosenblueth method. The method is compared with the traditional method with a fixed safety factor. The membership function and cut-set numbers influence on reliability are studied. Suggestions for the reasonable analysis are proposed. The results show that the mean value of safety factors at different cut-set levels cannot reflect the fuzziness of geotechnical parameters, while the reliability index has a good correlation with the fuzziness of parameters. With the increase of the fuzzy cut-set level, the reliability index increases. For this slope, the failure probability obtained by the fuzzy reliability method is 0.889% < 5%, which means a stable state and is more consistent with the actual situation compared with the fixed safety factor method. The parameter membership function form and the cut-set number have a great influence on the generalized reliability index of slope. When using fuzzy point estimation method, the trapezoidal or triangular membership distribution should be selected first. The number of cut-set is suggested to be nine. The generalized reliability analysis of the slope considers the fuzzy randomness of the shear strength parameters, avoiding the problem of the parameter accuracy of the fixed safety factor method. Therefore, the thinking of slope stability judgment is more scientific. The study results provide a reasonable reference for the analysis of similar slope projects.

Key words: Slope, Slate, Fuzzy Random characteristics, Generalized Reliability, Membership Function, Cut-set