Enhancing preventive maintenance using data mining technique

Abstract

Maintenance is an important part in any operational system, keeping equipment in good condition. This report proposed a maintenance model, multi criterion decision making grid (DMG), to provide decision support for different components and parts in equipment, applicable in both single and multi-unit system. The author will then attempt to apply this model to an industrial case study, using deterministic criterion gathered from the case study historical data in making this decision analysis. Decision support capability of the model is further improved by employing fuzzy inference system (FIS). Discussion and implementation of fuzzy logic would also have to be made. Also, a simplify data performance analysis is presented with the decision making grid model.

Following on, emphasis is made on formulating a predictive maintenance model for condition-based monitoring (CBM) maintenance policy in DMG. This predictive maintenance will uses application of data mining technique, in which association rule mining is chosen to determine correlation of preceding fault and alarm events for prediction of the investigated CBM failures. Utilizing the result, informed decision can be made beforehand and render the CBM policy failures.

To summarize, the significance of this report is to design and development of decision support system along with integrating knowledge discovery system for predictive maintenance, thereby enhancing the maintenance management and saving cost for the company.