Abstract: The fatigue life of the parts were estimated by using the Goodman linear correction method and the linear cumulative damage rule， and the working load was statistically analyzed by the rain flow count method， considering the impact of the load change and the strength degradation on the reliability of mechanical parts. Then， the gradient reliability mathematical model was set up with maximum load in equal time to replace the variable load by using Gamma random process simulating strength degradation. Finally， the expression of reliability sensitivity was derived by stochastic perturbation theory and four rank moment method. The torque shaft of coal mining machine was taken as an example， the fatigue life of the shaft under actual load was estimated and the each parameter variables′ variation rule of dynamic reliability sensitivity during the fatigue life was given. Furthermore， the influence of the parameter variables′ variation on torque shaft reliability was analyzed， and the Monte Carlo simulation method was used to verify the results. All analysis results show that the dynamic reliability analysis process adopted conforms actual working conditions， and can provide a method for dynamic reliability analysis of mechanical parts in real situations.

Key words: mechanical parts, fatigue life, Gamma process, fourth-order moment, reliability sensitivity