TIME SERIES INTERVENTION ANALYSIS WITH APPLICATION TO MULTI-UNIT SYSTEM MAINTENANCE MANAGEMENT

Abstract

Maintenance management is important in order for complex systems to continue operation with minimal interruption. A primary concern of maintenance management is the failure pattern exhibited by increasing failure rate (IFR) systems. In the existent literature, a common assumption about component behavior is that after replacement the new components work "as good as new". In this study, we relax this assumption and employ intervention analysis based on time series formulation to study the failure behavior of the system under preventive/predictive maintenance policies. We explicitly consider different preventive maintenance and replacement policies, the aging characteristics of the units as well as other operational issues. The proposed methodology provides the basis for a new approach to the vexing problem of the realistic characterization of IFR systems in the presence of maintenance intervention. The model is verified using data collected from a real life system and the parameter estimation is carried out using the refined instrumental variables method.