Deteriorated equipment condition has a significant impact on theproduct quality and maintenance policies. It is well known thatonline diagnosis systems and intelligent maintenance strategy playan important role to support maintenance and production in themodern manufacturing industry. Among numerous issues related tomanufacturing maintenance, we address three major challengingproblems: 1) an adaptive anomaly detection algorithm forcondition-based maintenance, 2) more accurate stochastic modelsfor preventive maintenance, and 3) joint maintenance andproduction scheduling for a multiple product and multiple stationsystem.Adaptive anomaly detection allows us to conduct not only theonline degradation assessment but also anomaly diagnosis in thepresence of unknown faults. This algorithm is realized by usingthe hidden Markov model with reinforcement learning techniques.Online machine health information can further be investigated forthe relationship on the product quality and equipmentdeterioration. Based on impact of machine condition to the productquality, we develop an integrated maintenance and dynamic productsequencing policy that can be applied to a multiple product andmultiple station system.For preventive maintenance, the traditional degradation modelsonly focus on a single machine system and ignore maintenancedurations. We perform analytical and numerical examination ofproduction lines with the Markov process framework, focusing onthe more accurate dynamic behavior modeling and multiplemaintenance tasks. Non-exponential holding time distributions inMarkov chain are approximated by inserting multiple intermediatestates based on a phase-type distribution. By having an adequatemodel representing both deterioration and maintenance processes,we can find different optimal maintenance policies to maximize theavailability or productivity for different configurations ofcomponents.
【 预 览 】
附件列表
Files
Size
Format
View
Maintenance Strategies for Manufacturing Systems using Markov Models