Equipment installed in buildings for heating,ventilation, and air conditioning is becoming increasingly complex. Operation with conflicting goals is getting more difficult. Methods are needed to significantly improve the real-time detection and diagnosis of faults in building equipment. Current approached to solving this problem do not workwell in nonlinear systems such as VAV, and can give incorrectresults due to sensor bias, model inaccuracies and modeleddisturbances or other noise. Extensions of traditional technologyto nonlinear systems are often not robust. Furthermore, the implementations are time consuming and aredifficult for nonspecialists to apply, limiting widespread use.The proposed research will develop a new methodology for creatingfault detection observers (FDO's) for detecting and diagnosingthese problems. It will overcome the weaknesses of both puremodel-based approaches and pure pattern recognition approaches,by a combination of model-based and neural network techniques. Itwill embed these techniques in an overall object orientedgraphical expert system environment to simplify future widespreaddeployment. For Phase 1, a prototype will be programmed as partof the research. Phase 2 will resolve additional outstandingquestions, prior to ultimate commercialization.Commercial Applications:Gensym, with its large customer baseexpects to commercialize this innovative technology as a productfor building monitoring, process, and aerospace markets. Mostindustries need to readily generate reliable diagnostic systems.
