Ship Target Classification and Identification Based on MVDR Spectral Parameters

In the modern ocean battle, how to extract features of ship targets during the complicated and changeable sea environment is a real key for passive sonar underwater target identification. This paper firstly discusses Minimum Variance Distortionless Response Modeling method of ship target. Then, Comparing with the MVDR, FFT and LP modeling method, MVDR is found to provide better spectrum envelop of ship radiated noise, and it analyses the influence of the order number on the performance which is found that it can provide better spectrum envelop of ship radiated noise with the increase of the filter order. Finally, the designed BP neural network is adopted to classify and identify MVDR spectrum characteristics and LP spectrum characteristics. Based on processing results of experimental data, the MVDR spectral parameters proposed in this paper has better classification effect and higher reliability than LP spectrum.