Facility On/Off Inference by Fusing Multiple Effluence Measurements

摘要: We consider the problem of inferring on/off operational state a reactor facility by using effluence mea- surements three noble gases, namely, Ar-41, Cs-138, and Xe-138, which are collected on facilitys ventilation stack. first present classifiers based thresholding measurements individual types, then methods that combine their outputs or measurements. develop sample- implementations five fusers simple majority rule, Chow’s recognition function, physics-based radiation counts model, correlation-coefficient method, Fisher’s combined probability test. apply latter four to pairs all gas types. Our results show that: (i) these effective in status facility, for example, best achieve 97% detection at 1% false alarm rate, (ii) performance depends data classification particular, types models, correlation-coefficients method outperform rule as well thereby illustrating importance fuser choice.