Abstract: Based on saddlepoint approximation method and sensitivity analysis method, reliability and reliability sensitivity of machine components with random parameters are studied. On the premise that the probability distribution of basic random parameters is known, probability density function and probability distribution function of stochastic structure response under the action of external loads are obtained by using saddlepoint approximation technique. In the calculation of practical example, numerical analysis of the bolt and automotive front axle is carried out, the result obtained by using the saddlepoint approximation method is compared with that of Monte-Carlo method, which shows that the former has the advantages of higher calculation accuracy and speed, and better tail distribution. The sensitivity analysis method and probability density function are employed to discuss the variation regularities of reliability sensitivity, and the effect of design parameters on reliability of the bolt and the automobile front axle is analyzed. This method provides the theoretical basis for reliability design of machine components.

Key words: Random parameter, Reliability sensitivity, Saddlepoint approximation