Key Techniques Of Data Processing In Underwater Acoustic Environments

Ocean acoustic surveys are the basis not only of physical regularities research of underwater acoustic channels, but also of design and development of sonar equipment, generally including measurements of source level (SL), sound transmission loss (TL), noise level (NL), reverberation level (RL), sound speed profile as well as acoustic parameters of sediments. The data processing of ocean acoustic survey aims to acquire acoustic database of processing results and provide a basis data set support for marine environmental information system construction. The amount of data of ocean acoustic survey is large, traditional processing methods require a person who has professional knowledge, to deal with the data manually and hence slow speed. The paper focuses on the key technology of automatic data processing of measurements for sound level, sound transmission loss, noise level, in order to improve processing efficiency, reduce the degree of human intervention and improve the quality of results.Beginning with the sonar equation, the paper discusses the meaning of the sonar parameters and features of sound geometrical spread and noise in typical ocean sound channel, and introduces the typical survey methods of acoustic and computational methods of acoustic parameters with one-third octave. The source level is estimated by reversing the receiver signal level with spheric geometrical spread law, the transmission loss is estimated by source level subtracting receiver signal level, and the noise level is estimated by periodogram method.The paper focuses on an automatic detection method of transient signal in ocean. The spectrum characteristic of ocean noise is analyzed firstly, to conclude that the main energy concentrates under 100Hz. The distribution of ocean noise in specific sea which lasts for a few seconds is Non-Gaussian through Hinich method and Histogram method, however it is approximate Gaussian after passing a high pass filter whose cutoff frequency is 100Hz, as a result, data will pass a high pass filter before detection. Modeling oceanic noise as a linear superposition of transient interference and ambient noise, we propose a model-based adaptive threshold setting method of energy detector. The threshold is high when the hydrophones are close to source for low false alarm probability, is low when the hydrophones are far from the source for high detection probability in the same survey line.We showed a simulation of hydrophone depth fluctuation induced TL errors in shallow water, which indicates that can't be ignored in the studied cases. The acoustic propagation model is then used to obtain the varying curve of TL over receiver depth at a fixed range, and a polynomial is used to fit the curve of model TL. Finally the fitted polynomials are used to correct TL to some fixed depth to get a TL curve over range. Experiment data from ASIAEX are processed to verify that the method is feasible. This is another emphasis of the paper.At last, we construct the processing system of ocean acoustic parameters measurements, which includes the calculation of SL, TL and NL. The system consists of normalizing of raw data, high pass filter, interception of noise and explosion wave, information synchronization during preprocessing period, and divides into the 1/3 octave band and evaluation of SL, TL and NL during the processing period.