Real-Time Fault Detection and Diagnosis: The Use of Learning Expert Systems to Handle the Timing of Events

The successful performance of real-time, sensor-based fault detection and diagnosis in large and complex systems is seldom achieved by operators. Examples of operator and system failures are presented and analyzed. The lack of an effective method for handling temporal data is seen as one of the key problem in this area. As part of the solution to these problems, a methodology is introduced that is able to make good use of temporal data to perform fault diagnosis in a subsystem of a Navy ship gas turbine engine propulsion unit. The methodology is embedded in a computer program designed to be used as a decision aid to assist the operator. It utilizes machines learning, is able to cope with uncertainty at several levels, work in real-time, and is developed to the point of possible application. Data are presented and analyzed with regard to the effectiveness of this approach. Relevance and applicability to other process control and classification problems are discussed. The approach is put forth as an example of how relatively simple existing techniques can be assembled into more powerful real-time diagnostic tools. Keywords artificial intelligence multisensor integration.