Optimization Of Equipment Maintenance Criterion Involving Minimum Average Cost And Remaining Lifetime Based On Indirect Monitoring

Equipment is a very important part for enterpriee in modern manufacturing industry, in other words, no equipment no production, no equipment could operate smoothly in all its lifetime, so when it breaks down, proper maintenance is needed. As a rule, more enterprise would adopt condition based maintenance, which means change the wear part or exponent preventively according to the condition that is diagnosed by the monitored values and information. But when to replace just become a big problem. As for direct monitoring which the monitored information could directly reveal the exact degradation state, many scholars give the decision criteria, however, seldom gives the method about replacement policies as for indirect monitoring. This thesis is to draft out an optimal policy involving minimum average cost and mean remaining lifetime to help those practitioners make a proper decision about equipment maintenance.To formulate replacement policies, the failure rate is needed. In order to overcome the defects of only consider the current observations without involoving history ones, a proportional hazards model is introduced, which not only consider the age of equipment but also the effects caused by degradation state.Besides, the thesis introduces a concept of decision criterion. As a matter of fact, the decision criterion is based on the age, the distribution of the degradation and minimum average cost of maintenance, then, the minimum average cost is solved by introducing a recursive method. Last a numeric example is given as the verification of theory proposed above, the average cost is compared as for the two condition monitorings, besides, the effects which is caused by varying parameters is analyzed.In traditional maintenance there is a problem that only consider the cost of replacement, but seldom involving the availability, so in chapter 4, the method of calculating the mean remaining lifetime is introduced, the mean remaining lifetime is the function of reliability of equipment, so at every monitoring moment, an observation is collected, then accordingly the distribution of state and the reliability of the equipment is calculated using Bayes rules and Partial Observable Markov Decision Process, at last equipment's mean remaining lifetime is obtained.