Leak Acoustic Emission Signal Classification and Diagnosis Based on the Fractional-Order Fourier Transfer and Ensemble Learning

Fractional-order Fourier transform is a representational method of the fractional domain formed by signal rotated any angle counterclockwise about origin on coordinate axis in time-frequency plane. This paper intends to use process signals collected from acoustic emission device, and then train results through ensemble SVME, KNN Softmax, so as build model that can predict size location leak holes device. The has good accuracy predicting whether or not are empty. If you only need it leaks, reaches 75.6 %, compared trained original data, classification increased 25.6% 66.8%. In particular, Softmax classifier, addition FFRT increases more than 200%.