No abstract

In machinery condition monitoring, weak signal is a type of signal that has very low signal-to-noise ratio (SNR). Therefore, it is necessary to develop effective methods so as to enhance or extract useful information from the noisy signal. These methods aim to solve the problem of de-noising and separation of mixed-signal. This paper investigated the current progress on weak signal detection methods, including traditional and emerging, linear and non-linear methods. The combination of various methods is presented according to their advantages and disadvantages of these methods. In the end, the future development of weak signal detection technology is summarized.

In order to investigate long-range spatial affects of high ambient-noise levels generated at the ice-water boundary by the interaction of ocean waves with sea ice [O. I. Diachok and R. S. Winokur, J. Acoust. Sec. Am. 55, 750–53 (1974)], coincident measurements of the spatial variability of ambient noise, transmission loss, and thermal structure were made along several 300-NM open water tracks normal to the boundary. The measurements were made from two aircraft with sonobuoys and expendable bathythermographs deployed and monitored every 20 NM, and explosive signals, deployed every 4 NM from one aircraft. The signals were detected on modified sonobuoys, which were monitored from a second aircraft. The resultant transmission loss data were corrected for cylindrical spreading and the attenuation coefficient was computed for comparison with the rate of change of the ambient noise versus distance measurements. An analysis of the results indicates that the ice-water boundary may be considered as a line source of sound and that the observed variations of ambient-noise signals with range may be quantitatively related to the computed attenuation coefficient.

Measurements of transmission-loss and sea ice ridge parameters were made in the Arctic Ocean to test predictions of an under-ice reflection loss model, which assumes that ridges may be represented as randomly distributed elliptical half cylinders. The acoustic measurements were accomplished with explosive signals deployed from aircraft, and detected on hydrophones deployed from ice island T-3, and at other sites on sonobuoys monitored from a second aircraft. Ice ridge keel characteristics were inferred from airborne laser profiler measurements of ridge sails, using geometrical and statistical models. Numerical values of reflection loss as a function of keel parameters and grazing angle were incorporated into ray theoretical computations of transmission loss for comparison with measurements. The good agreement at high frequencies (f>200 Hz) is in accordance with previous such comparisons, and may be considered a demonstration of the validity of the model. The good agreement at low frequencies (f<50 Hz), however, may in part be fortutitions due to uncertainties in the ridge model, reflection-loss model, ray theoretical computations, and source levels.

No abstract