Abstract:

The high rate of false alarm is the main problem that influences the performance of builtin test (BIT) system and its wide application. A combined false alarm reducing method for BIT system based on time stress analysis is introduced. A support vector machine (SVM) model is used to correlate the false alarm of BIT to time stress information. On the other hand, a fuzzy clustering analysis based on kernel based principal component analysis is made for false alarm recognition. And then, an optimization multilevel decision model is put forward to reduce the false alarm rate of BIT system. Finally, in order to prove the validity of the false alarm reducing method introduced, an experimental study on BIT system of a helicopter attitude indicator is executed in detail.

Key words: time , stress,  , BIT,  , false , alarm,  , SVM,  , kernel , based , principal , component , analysis,  , fuzzy , clustering , analysis