Condition monitoring and diagnostics of systems and its relationship to some aspects of artificial intelligence.

Abstract

Artificial intelligence and condition monitoring and diagnostics of systems are emerging, computer-based transdisciplines. The origins of the theoretical kernels of both transdisciplines can be traced to the last century, but the richly structured theories in the body of knowledge that supports the present work in the cognitive sciences and our recently achieved ability to make real-time decisions related to the operation and maintenance of modern, large-scale, complex mechanical systems are due, in large part, to the explosion of synergistic activity in systems thinking and computers which has occurred in the past three decades. In artificial intelligence one strives to make machines (usually, digital computers) do the sorts of things that the human mind does, including making decisions and problem solving, which is an essential element in condition monitoring and diagnostics as well. Thus there is a strong link between the two transdisciplines. A generic definition of artificialintelligence is given above; however, there is no universally accepted definition of the term conditionmonitoringanddiagnostics, either in its general form or in one of its specific forms, say, the condition monitoring and diagnostics of machines. Nevertheless, it is believed that most practitioners of the condition monitoring and diagnostics art would agree that the activities encompassed by the term can be characterized as a two-phase sequence of processes designed to monitor the condition of an operating, functioning system for the purpose of diagnosing faults in the system. Monitoring is a passive, system-specific, information and data gathering process. Diagnostics is an interactive decision-making process in which the unique, system-specific information and data obtained in the monitoring phase are used by a human or computer-based decision maker to decide if faults exist in the system, where they are located, their criticality and what advisory and/or corrective action should be taken. In the diagnostic phase, computer-based decision making (read ‘‘artificial intelligence’’) can be used to support decisions made by a human or to make and implement decisions automatically. Both cases are discussed here, using the condition monitoring and diagnostics of a gas turbine as an illustrative example.