Research on underwater acoustic target recognition based on LOFAR spectrum and deep learning method

Underwater acoustic target recognition has long been an internationally recognized problem. This paper combines the method based on LOFAR spectrogram and the deep convolutional neural network of deep learning hot spot method for underwater target radiated noise recognition, making full use of the separability information of target noise in different dimensions, and using the time spectrum chart as the depth Learn to enter data. In this paper, the basic principle and specific application process of the method are analyzed in detail, and the measured data are simulated and classified, and the simulation diagram and test results are given. The results show that the recognition rate of underwater target recognition with this method is high.