Beamspace Time-Frequency Analysis for Identification of Underwater Tone Noise Sources
Xu Ling-Ji,Yang Yi-Xin,Yang Long
DOI: https://doi.org/10.7498/aps.64.174304
IF: 0.906
2015-01-01
Acta Physica Sinica
Abstract:The noise emitted by an underwater vehicle consists of several strong tones superimposed on a broad-band radiated noise component. Among them, the stable low-frequency tone noise induced by the reciprocating movements of the auxiliary machines in the underwater vehicle, carries characteristic information of the vehicle and is necessary for long-distance detection. Therefore, identification of the tone noise sources of an underwater vehicle is significant for noise reduction. On the basis of the joint information of space-time-frequency, beamspace time-frequency analysis (TFA) scheme is proposed for identification of low-frequency tone noise sources of underwater moving vehicle. First, the Doppler signals formed when the tone noise sources pass through the closest point of approach (CPA) are separated in time domain, by using superdirectivity beamforming of a small aperture circular array. The output signals can be approximated in linear form, i. e., LFM signal. After the LFM signals from the narrow beam are processed by two TFA methods of pseudo Wigner-Ville distribution and chirplet transform (CT), the time-frequency images of the noise signals are obtained. Then, the CPA time of each tone noise sources can be estimated by using peak search of the time-frequency images. At last, by converting the time coordinate to space coordinate and comparing with a reference source whose CPA time and position are known in advance, the positions of the low-frequency tone noise sources on the underwater vehicle are identified. The proposed scheme is different from the focused beamforming method, which scans the beam angle after eliminating the Doppler effect. Besides, due to no need of decorrelation usually used in the focused beamforming method, beamspace TFA scheme resolves the problem that array aperture is limited for identification of coherent noise sources of an underwater vehicle. The aperture of the used array can be reduced to meter-scale even when the frequencies of the tone noise are low. Although the array gain of superdirectivity beamforming decreases in nonisotropic noise field, the main lobe of the beam still keeps the same shape. Therefore, the performance of the proposed scheme is robust. Simulation analysis shows the following results: 1) Both the two beamspace TFA methods can precisely identify the underwater tone noise sources through a small aperture circular array, the radius of which is equal to 1.6 m, and the localization errors are less than 1 m when the signal-to-noise ratios are moderate; 2) The higher the frequencies of the tone noises are, the better the localization accuracy of beamspace TFA methods obtain; 3) The proposed scheme is less sensitive to the velocity of the underwater moving vehicle, and the localization results just have very small difference under various velocities; 4) The localization accuracy is related to distance, and decade meters is a reasonable choose for actual noise measurement; 5) Beamspace CT has better resolving accuracy when the information of measurement system is given, so the choice of the two beamspace TFA methods can be decided according to the actual measurement condition.