Reliability sensitivity analysis based on system failure evolution

System failures of the complex electromechanical system have strong coupling, integration and cross. Reliability design performed at the system level will get a better profit than at the bottom under the same conditions. For improving the system reliability and reducing the costs, taking advantage of agent-oriented modelling theory, a novel system reliability analysis model called system failure evolution model is proposed, and a sensitivity analysis method is introduced to accommodate the model. By simulating the generation, evolution and propagation of system failures, system reliability and mechanisms of system failure can be studied. Combined with significance tests and pseudo-Monte Carlo simulations, the Sobol´ method is used to do multi-parameter sensitivity analysis. Finally, a comparative analysis example of the servo-feed system was given to demonstrate the usability and advantage of the proposed model, and the parameter sensitivity analysis were performed after simulating the evolution of system failures.