AUTOMATED LEARNING FOR REAL-TIME EXPERT SYSTEM IN MONTORING AND CONTROL

High-speed, real-time diagnostic and advisory systems are part of ICS?s heritage. Unstructured and labor-intensive knowledge capture systems have been the downfall of many monitor and control systems. We believe that an adaptive learning system can greatly aid in the population and refinement of automated reasoning systems. Our proven, man-rated expert system must be populated with the fault monitoring rules for a given domain. Exhaustive path analysis for complex systems has proven to be labor-intensive and error prone without a structured methodology or an automated approach. ICS and FIT believe that an automated learning system can define relationships between sensors and effectors and automatically generate rules for the Expert System to aid the operator in real-time monitor and control. This new tool would be a leap forward in extending the capabilities of the system to perform system validation procedures and offer advisories to operators and other web-based users in real-time. FIT?s research into adaptive learning and data mining systems holds great promise for non-invasive applications to mine knowledge from data streams. Using ICS and FIT?s research, we believe a derivative architecture can correlate data measurands and auto-generate rules for an expert system for monitor and control.