A DIGITAL FEEDWATER CONTROL SYSTEM WITH SENSOR VALIDATION AND ANALYTIC REDUNDANCY

THIS PROJECT WILL ANALYZE AND PREPARE THE CONCEPTUAL DESIGN FOR A DIGITAL FEEDWATER CONTROL SYSTEM FOR A COMMERCIAL LIGHT WATER REACTOR POWERPLANT. THE CONTROL SYSTEM WILL FEATURE SENSOR VALIDATION AND REPLACEMENT OF FAILED SENSORS WITH SYNTHETIC VARIABLES. AN ADVANCED TECHNIQUE BASED ON PATTERN-RECOGNITION ALGORITHMS ALREADY PROVEN IN SURVEILLANCE APPLICATIONS WILL BE USED TO VALIDATE SENSOR INPUTS AND SYNTHESIZE THE REPLACEMENT VARIABLES. PRIOR WORKHAS DEMONSTRATED THE VALIDITY OF THE PRINCIPLE; THIS PROJECTWILL BEGIN THE WORK OF APPLYING THE PRINCIPLE TO A CRITICAL NUCLEAR PLANT APPLICATION. THE PROJECT WILL INCLUDE: SELECTION OF THE SET OF PLANT PARAMETERS TO BE MONITORED AND PROVIDED AS INPUT INTO THE PATTERN RECOGNITION ALGORITHM, TO ACHIEVE THE BEST ESTIMATION OF THE PARAMETERS TO BE VALIDATED AND SUBSTITUTED; DESIGN OF THE CONTROL SYSTEM HARDWARE AND SOFTWARE; AND DESIGN OF THE LOGIC THAT WILL CAUSE A PARAMETER TO BE REPLACED BY A SYNTHETIC VARIABLE. THE RELIABILITY AND CORRESPONDING METHOD OF IMPLEMENTATION OF THE CONTROL SYSTEM HARDWARE AND SOFTWARE WILL BE CONSIDERED IN VIEW OF THE ADVANCED CAPABILITIES OF THE PATTERN-RECOGNITION TECHNIQUES FOR VALIDATION AND SYNTHESIS OF VARIABLES. THE DESIGN OF THE VARIABLE SUBSTITUTION LOGIC WILL DEVELOP THE PROCESSS BY WHICH DECISIONS TO SUBSTITUTE ARE MADE BY THE CONTROL SYSTEM.