TARGET CLASSIFICATION FROM JEM SIGNAL USING FREQUENCY MASKING

This paper deals with a technique for classifying jet aircrafts from JEM (Jet Engine Modulation) signal. A novel method to recognize an engine model by analyzing JEM spectrum using frequency mask is proposed. The frequency mask extracts and analyses the spectral component at the frequencies that are predicted from the blade number of a jet engine and the estimated spool rate. The proposed method does not need a complicated logical algorithm for finding the chopping frequency or the pre-simulated engine spectra used in previous methods. In addition, we suggest a method to precisely estimate the spool rate in the spectrum domain of JEM signal, which plays an important role in generating the frequency mask. The classification experiments using the JEM signals measured from two fabricated engine models verify that the proposed algorithm has good performance in the recognition of jet aircrafts.