Concepts of condition-based maintenance for intelligent ships

Historically, the maintenance of complex machinery such as engines, transmissions, or pumps has proceeded in accordance with one of two basic philosophies: (1) schedule or use-based maintenance, in which preventive maintenance is performed after a specified period of time or history of machine use (e.g., hours of operations); or (2) restorative maintenance in which repairs are made to machinery after it fails (or exceeds operational bounds). Schedule or use-based maintenance is performed based on mortality statistics (e.g., statistical studies of wear and failure which allow prediction of the expected time to failure of a population of similar mechanical systems). A new philosophy, termed Condition-Based Maintenance (CBM), is evolving in which a system is instrumented with one or more sensors, and maintenance is performed based on the observed condition of the individual system. and on projections of failure events (usually based on simple trending techniques). CBM provides the potential for improved safety, reduced lifecycle maintenance costs? and extended use of existing equipment. Indeed, if properly implemented, smart CBM systems could be used to reduce the manpower required for ships. CBM is becoming increasingly viable due to recent advances in sensors, digital signal processing, mechanical modeling, and automated reasoning techniques. This paper introduces the concept of CBM systems, and provides an overview of research currently being performed at the Applied Research Laboratory, The Pennsylvania Stare University (ARL Penn Stare). The research involves five main areas: (1) sensors and sensing technology; (2) signal processing and multisensor data fusion; (3) micro-mechanical modeling; (4) nonlinear dynamical modeling; and (5) intelligent reasoning and control. In each area. new advances art being made to provide the basis for reliable condition monitoring and failure prediction. Advances include. new smart sensors fabricated via nanofabrication techniques; development of new tools for signal processing and fusion of non-commensurate sensor data; development of improved understanding of the micro-mechanical failure phenomena; new models for prediction of failure evolution; and hybrid approximate reasoning techniques for automatic, contextual, interpretation of the output of sensors and models. This research is being performed as part of a portfolio of research projects funded by the Office of Naval Research (ONR), including a Multidisciplinary University Research Initiative (MURI) on Integrated Predictive Diagnostics (IPD). The paper will present an overview of mis research and an assessment of the applicability of that research to Smart Ship applications.