Underwater ship noise characteristic extracting method based on IMF energy entropy and PCA

Abstract

The invention provides an underground ship noise characteristic extracting method based on IMF energy entropy and PCA. The method comprises steps of extracting time domain characteristic parameters of the average, the variance, the peak value and the skewness of ship noise signals and constructing a time domain characteristic matrix; using the EEMD method to decompose the original ship noise signals to obtain IMF components, converting the IMF components into energy characteristic vectors, thereby observing change of frequency band energy characteristics, and constructing a frequency domain characteristic variable to form a characteristic matrix A together with the time domain characteristic vectors; using the PCA dimension reduction method to carry out dimension reduction processing on the characteristic matrix A, and mapping the characteristic matrix with the high dimension into characteristic matrix with the low dimension to serve as a new characteristic so as form a characteristic matrix B; and training LSSVM parameters, according the risk minimization principle, adjusting the parameter gamma and the kernel width sigma, finally inputting the characteristic matrix B subjected to the dimension reduction into a classifier and testing a classification result. According to the invention, time domain information can be combined with the frequency domain information and information complementation is formed, so quite complete characteristic information can be provided and classification precision can be improved.

Description

technical field [0001] The invention relates to the field of underwater target recognition, in particular to an underwater ship noise feature extraction method based on IMF energy entropy and PCA. Background technique [0002] At present, my country's demand for marine resources continues to increase, and the scale of development and utilization continues to increase. Underwater target recognition technology has also developed rapidly and has been applied to many aspects, such as fish detection, marine biological research, and underwater submarine detection. Underwater target recognition plays an important role in biological protection and strengthening marine defense security, and is of great significance to marine development and sustainable development. The identification and classification of underwater ship noise signals is the focus and difficulty of underwater target recognition and classification. The ship's engine noise is mainly used to judge the ship's category. ...

AI Technical Summary

Problems solved by technology

The traditional methods are based on the assumption that the signal and noise are stationary and Gaussian random processes. With the improvement of the vibration and noise reduction performance of underwater acoustic targets, it is difficult for the traditional signal processing method based on the Fourier transform to accurately extract the underwater radiation. noise characteristics

Moreover, the amount of feature vector data extracted by the traditional method is large, which is easy to cause the curse of dimensionality, and it is easy to cause over-fitting in the classifier and reduce the classification accuracy.

Images