Reliability Modelling And Analysis Of Hydraulic System Based On Evidence Theory And Bayesian Network

In order to solve the uncertain problems caused by poor field data, complex system structure, low cognition level of failure mechanism and the polymorphisms problems of multiple property and multi-fault state in hydraulic systems, the reliability modeling and analysis based on static Bayesian network and evidence theory is proposed, in this method, the evidence theory is adopted to describe uncertain information, and the bidirectional reasoning ability of Bayesian network is made use. However, the static Bayesian network can not describe the dynamic system with time parameters, aiming at the dynamics and sequential characteristics of priority, functional dependency, sequence enforcing and the logical relationships of spare parts(cold spare, hot spare, warm spare, etc) caused by system redundancy, the reliability modeling and analysis method based on discrete time Bayesian network and evidence theory is proposed and applied in the hydraulic system, the method provides effective basis for the reliability analysis of hydraulic systems with uncertainty, polymorphisms, dynamics and sequential.Aiming at the problems of uncertainty and polymorphisms that exist in hydraulic systems, the method of reliability modeling and analysis based on the static Bayesian networks and evidence theory is proposed, the likelihood values and trust probability of the evidence theory are used to describe the upper and lower limits of the fault probability of the root node respectively, furthermore, the calculation methods of leaf nodes’ failure probabilities and fault rate, leaf nodes’ uncertainty degree, root nodes’ posterior probability and posterior fault rate, root nodes’ importance degree, cognitive importance degree and sensitivity, are introduced. The strong ability of this method to describe uncertain information is proved by case studies of hydraulic system.Aiming at the uncertainty and dynamics and sequential of the hydraulic system, the method of reliability modeling and analysis based on the discrete time Bayesian network and evidence theory is proposed in this article. The solution method of the elements’ fault probability at different times, the method of the transition from the static logic gate and dynamic logic gate to discrete time Bayesian networks are given. The evidence model is adopted to describe the fault probability of root nodes, also, the reliability indexes including the trust reliability degree, likelihood reliability degree of the system, trust posterior probability and likelihood posterior probability of the root nodes at different times are given. Finally, the reliability of the hydraulic luffing system of bucket wheel reclaimer is modeled and analyzed.