Abstract: In order to address deficiencies of linear approximation of basic events' multiplication results in traditional trapezoidal fuzzy fault tree prediction algorithm, an improved algorithm was proposed based on both cut set theorem and representation theorem of fuzzy mathematics. Then, with fire accidents in a coal mine as an example, accident consequence was simulated in five cases, and furthermore, reliability and adaptability of the new algorithm was verified according to both traditional and improved trapezoidal fuzzy fault tree prediction methods. Finally, approximate and accurate membership degree graphs were drawn, and approximate and accurate fuzzy importance degree of electrical equipment explosion events were calculated. The results show that the new algorithm features good adaptability, and its application in five cases can all improve prediction accuracy. The bigger blindness and core of basic events' trapezoidal fuzzy number and the longer data sequence are, the more significant improvement new algorithm will achieve.

Key words: trapezoidal fuzzy fault tree, prediction alogorithm, membership degree, cut set theorem, coal mine, fire accident