Predictive Maintenance (PdM) strategy is a more logical concept than corrective or preventive maintenance strategy in that it carries out maintenance tasks based on actual operating condition of a sophisticated system. PdM program is composed of system analysis, monitoring techniques, fault detection and diagnosis techniques. In a PdM program, one performs system analysis to select cost-effective monitoring parameters that reflect lots of information about the system condition, and analyzes signal-data obtained by monitoring these parameters to provide information for detecting faults early and isolating failure causes. These activities enable maintenance to be planned and scheduled to minimize maintenance cost and impact on plant operation. This thesis presents CAPMS (Computer-Aided Predictive Maintenance System) as an integrating concept of PdM program, which is composed of a) Condition-Monitoring part, b) Fault Detection and Diagnosis part, c) Maintenance Management part, and d) Data base Management part. The state-of-art for each part is surveyed, and research issues are identified for a successful implementation of PdM.
