Maintenance grouping optimization with system multi-level information based on BN lifetime prediction model
| Autor: | Yuan Zhang, Jingbin Wang, Jianxing Lu, Lizhi Wang, Xiaohong Wang |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
0209 industrial biotechnology
High security Computer science Bayesian network System safety 02 engineering and technology Industrial and Manufacturing Engineering Task (project management) Reliability engineering Group maintenance 020901 industrial engineering & automation Hardware and Architecture Control and Systems Engineering 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Software Reliability (statistics) |
| Zdroj: | Journal of Manufacturing Systems. 50:201-211 |
| ISSN: | 0278-6125 |
| Popis: | Group maintenance for multi-level systems is necessary to ensure task success and system safety. However, many group maintenance models, which only consider the health of components without regard for reliability information at system-level, have difficulty meeting the increasing system task-performance demands. Based on system multi-level information, an age-based group maintenance method that trades off cost and system reliability is proposed. The method considers different failure mechanisms of units and system structures, and achieves a grouping strategy and maintenance decision-making approach according to multi-level lifetime prediction data. The reliability information at system- level is predicted by Bayesian network (BN) from life information of units, and multi-objective programming of cost and system reliability is used to optimize maintenance grouping strategies. This method is applicable to multi-level systems of varying sizes. A simulation example and a solar-powered unmanned aerial vehicle (UAV) application illustrate the method. The results verify the feasibility and superiority, and meet the high security and reliability standards. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect