Underwater Cylindrical Shell In Different Thickness Recognition Using Biomimetic Dolphin Clicks
Gang Qiao,Xin Qing,Donghu Nie,Shuai Ma,Yi Zhang,Jiangsheng Tang
DOI: https://doi.org/10.1109/OCEANS.2016.7761433
2016-01-01
Abstract:For the traditional narrowband sonar signal, recognizing underwater small targets is an extremely difficult task. Therefore, a series of biomimetic dolphin clicks, a kind of broadband and transient-like sonar signal, were used as the transmitting waveform for recognition tasks, and the tasks were presented from experiments in an anechoic pool. In this experiment, three biomimetic dolphin clicks, differing in the energy distribution in the time-frequency domain, were used to recognize three copper cylindrical shells which have a diameter of 10 centimeters, 40 centimeters high and different thickness. The echoes from three targets were collected by receiving system, and were analyzed in the time domain and frequency domain. In addition using wigner-ville distribution, the echoes were projected to the time-frequency space for feature observation. Then the time-frequency features of echoes were extracted by singular value decomposition, and these features were classified by a support vector machine for recognizing target echoes. Experimental results indicated that these cylindrical shells in different thickness can be recognized by the biomimetic dolphin clicks, and different clicks obtain different echo responses and different identification results. Therefore, the echo responses have strong dependence with the time-frequency distribution of transmitting waveform. Furthermore the results show a promising way to improve underwater acoustical recognition in an intelligent waveform design manner by a dynamic closed-loop feedback sonar system.