Research On Decision Models And Application Of Condition Based Maintenance

Condition Based Maintenance (CBM) is a maintenance technique that has been researched abroad in recent years. Firstly, the basic concept of CBM is introduced, and then the condition models, decision methods and application are discussed in detail. The main research content is as following.The condition model based on small sample is studied to overcome the lack of samples in experiments, which appears as the difficulty of experiments. The Weibull Proportional Hazards Model which parameters are estimated through Bayesian method is used to describe the health of system. To obtain the prior distributions of parameters by Bootstrap, I proposed the integrated sampling method to generate Bootstrap samples. The method simulates a Markov Chain to describe the state changing procedure based on the failure time sampled by Monte Carlo method. And the posterior distributions of parameters are obtained by Markov Chain Monte Carlo (MCMC) method.The current maintenance decision method is based on system reliability, in which the mission to be earned out is seldom considered, and would cause decreasing of running efficiency. To overcome the disadvantages, I proposed the Mission Oriented Condition Based Maintenance.. First, the concept and analyzing procedure are given. Then the method working out the policy based on mission and state is studied. On the optimization of CBM policy for repeated single mission, the Markov Decision Process (MDP) is applied, and also a numerical method is proposed in which the maintenance threshold of state is determined according to the relation between the average availability or the average cost and system states when the condition data is not periodical. For particular mission, we give the condition based maintenance policy on optimizing cost, availability and mission reliability in mission time. Making decision with lack of running state information, I proposed a method by which the maintenance time could be obtained based on working age and current condition data. I proposed the sequential inspection policy based on the Markov Chain, which works out the next inspection time based on the current condition. The CBM policy of identical components serial system is studied based on MDP, the condition based maintenance policy optimizing cost and mission is given.Based on the researches above, the software accomplishing the method in this paper has been developed. The CBM policy for engine based on the Spectrometric Oil Analysis is worked out by the software.