A classification and recognition method of underwater acoustic signal targets based on deep learning

Abstract

The invention belongs to the technical field of underwater acoustic signal processing, and in particular relates to a method for classifying and identifying underwater acoustic signal targets based on deep learning. The present invention includes the following steps: (1) feature extraction of the original underwater acoustic signal through the Gammatone filtering cepstral coefficient GFCC algorithm; (2) proposed to use the improved empirical mode decomposition MEMD algorithm to extract the instantaneous energy and instantaneous frequency, and extract the instantaneous energy and the instantaneous frequency with the GFCC algorithm (3) Establish Gaussian mixture model GMM to retain the individual characteristics of underwater acoustic signal targets; (4) Use deep neural network DNN to complete underwater target classification and recognition. The invention can solve the problems of single feature extraction and poor anti-noise capability of the traditional underwater acoustic signal target classification and identification method, and can effectively improve the accuracy of underwater acoustic signal target classification and identification. It still has some adaptability under other circumstances.

Description

technical field [0001] The invention belongs to the technical field of underwater acoustic signal processing, and in particular relates to a method for classifying and identifying underwater acoustic signal targets based on deep learning. Background technique [0002] The development and utilization of marine resources is an important way to achieve sustainable development. Underwater target classification and identification technology can help to better conduct marine biological surveys and marine exploration. The marine environment is complex and changeable, electromagnetic waves are greatly attenuated in water, and most underwater targets have their own specific acoustic characteristics, so it is more appropriate to choose the signal form of acoustic waves to study underwater targets. In the military field, the underwater acoustic signal target classification and identification technology can accurately, timely and covertly detect underwater targets, and provide accurate ...

AI Technical Summary

Problems solved by technology

In the classification and recognition of underwater targets, classification and recognition methods such as neural network, hidden Markov model and support vector machine (SVM) are usually used. These traditional underwater acoustic signal target classification and recognition methods use shallow structure algorithms to complete the classification and recognition of targets. Lack of the ability to learn the essential characteristics of the data set from a large number of data samples. At present, the types of underwater acoustic signal targets are complex, coupled with the complex environment of the seabed and a lot of noise, which makes the traditional classification and recognition methods unable to accurately complete the underwater acoustic signal. Target Classification and Recognition

Images