Visit the Energy Systems Laboratory Homepage.
Optimization of Fault Detection/Diagnosis Model for Thermal Storage System Using AIC
MetadataShow full item record
The authors simulated the human ability of pattern recognition mathematically, through finding the state-change characteristics of the objective system from actual measurements using statistical analytical tech-niques. The Fourier analysis method with the space in-variance that can extract the changing characteristics of the system state, such as phase and frequency, is chosen as a typical technique. The distinction function based on Maharanobis' pan-distance is used to identify and judge the essence of the event. In addition, human learning, recognition, and optimal judgment process of any event can be simulated by optimizing the most effective pa-rameters and their numbers for detection and diagnosis by the use of variable selection method. In previous papers by authors , the two optimization methods for the most effective detection and diagnosis vector, the variable selection method and the differentia-tion rate increment method, in which linear distinction function based on Maharanobis' pan-distance was used, have been reported. In the present paper, a new method for optimal model selection based on AIC(Akaike In-formation Criteria) is examined. In addition, by exam-ining the influence to the distinction and diagnosis rate of the optimal detection and diagnosis vector, the best convergence criteria of AIC was confirmed.
Pan, S.; Zheng, M.; Nakahara, N. (2006). Optimization of Fault Detection/Diagnosis Model for Thermal Storage System Using AIC. Energy Systems Laboratory (http://esl.tamu.edu); Texas A&M University (http://www.tamu.edu). Available electronically from