Application of Grey Linear Exponential Model in Fault Prediction

Abstract

Fault data of weapon systems are small-sample grey sequence,which often take on the wobbly characteristic. It is found by research that modeling of the grey wobbly sequence does not satisfy the condition of GM( 1,1) model. Therefore,it is proposed to use the dynamic exponent transformation for transforming the grey wobbly sequence into a monotonically increasing sequence with certain grey exponent law,and then to establish a GM( 1,1) model,which is called as grey linear power exponent function curve model( GIM( 1)). For GIM( 1),the unary linear regression modeling method is used for model parameter identification. The results prove that GIM( 1) model has good fitting and prediction accuracy for the grey wobbly sequence of the weapon system's fault sequence,which not only has the advantage of grey identification algorithm,but also can meet the identification requirement of general system.