An innovative fuzzy-logic-based methodology for trend identification

A new fuzzy-logic-based methodology for on-line signal trend identification is introduced The methodology may be used for detecting the onset of nucle ar power plant (NPP) transients at the earliest possible time and could be of great benefit to diagnostic, maintenance, and performance-monitoring pro grams. Although signal trend identification is complicated by the presence of noise, fuzzy methods can help capture important features of on-line sign als, integrate the information included in these features, and classify inc oming NPP signals into increasing, decreasing, and steady-state trend categ ories. A computer program named PROTREN is developed and tested for the pur pose of verifying this methodology using NPP and simulation data. The resul ts indicate that the new fuzzy-logic-based methodology is capable of detect ing transients accurately, it identifies trends reliably and does not misin terpret a steady-state signal as a transient one.