Research On The Key Technologies Of Active Detection Of Underwater Small Target

With the transition of human exploration from land to ocean,a great majority of countries have realized the importance of marine resources because of its crucial role in military confrontation,national defense construction,and economic development et al.Numerous expenses and endeavors have been invested in underwater technologies.Among them,sonar designed for detecting small targets such as frogman,underwater robot and UUV?Unmanned Underwater Vehicle?is of uppermost priority.Small target's low reflectivity and good mobility,coupled with severe underwater interferences and limited visible region,makes the detection,localization and tracking of such objects a challenge.Along with further investigation on the detection technique and growing application requirements in underwater acoustic signal processing,it is of great importance for the improvement of underwater equipment to develop new theory and methods of underwater small target detection under complex environment.Focusing on practical applications of underwater target detection,this dissertation investigates theories,methods,and anechoic tank experiment for underwater small target detection,mainly from the following aspects:scattering model of underwater diver,acoustic characteristics of small targets,coherent integration and nonlinear integration of acoustic signal,and range ambiguity resolution for moving target.The main contents are described as follows.1.Two models,statistical target strength model of bubble cloud and acoustic scattering model of diver's lung,are established after a thoroughly analysis of the diver's target characteristics and several main factors that contribute to its target strength?TS?.Through the study of the dynamic characteristics of bubble in water and the acoustic scattering properties of bubble cloud,a statistical model with fluctuating TS values is established for underwater bubble cloud by means of mathematical statistics method.The model illustrated the cause of the fluctuation of bubble cloud TS clearly and described the fluctuating properties of the bubble cloud more precisely than the non-fluctuating models.Numerical simulation results show that the TS of a bubble cloud with a 2L gas volume is about-20dB.Considering the compressional and shear waves propagated in human organs and tissues,the diver's lung is simplified to be a sphere composed by two layers of elastic mediums,and a double-layered elastic sphere model is constructed.Based on this model,the scattering acoustic field of human lung is presented and formula of its form function is deduced.From the form function,the echo wave form could be predicted.Computer simulation results indicate that the echo reflected by human lung is a series of delayed copy of the incident wave.According to this model,the TS of diver's lung is approximately-25dB.2.Motion compensated coherent integration method for arbitrary waveform is proposed.Based on the Gaussian noise assumption,the multi-pulse echo model for arbitrary waveform is established and its coherent integration expression is deduced.The analytical integration expressions of commonly used signal waveforms,CW signal,LFM signal,m-sequence phase-coded signal and Costas frequency hop coded signal,are given,and computer simulations for the four waveforms are conducted.The results verified the adaptability and correctness of the coherent integration method for arbitrary waveforms.Coherently integrating M pulses returned from the target will improve the signal-to-noise ratio?SNR?by a factor of M.Owing to the randomness of the noise,the integration gain cannot be exactly equal to the theoretical gain,but fluctuating around it.The fluctuating extent enlarges with the increase of the number of integrated pulses M.3.Multi-pulse non-linear integration algorithm for detecting underwater target from impulsive noise environment is proposed.According to the characteristics of underwater acoustic environment,an impulsive mixed noise model is established by viewing the underwater ambient noise as the combination of a Gaussian noise component and an impulsive noise component.The probability density function of this noise model follows?-mixture distribution.According to convolution theorem in frequency domain,which means multiplication in the time domain equals point-wise convolution in frequency domain,the multiplication operation is introduced into the multi-pulse integration theory.Through production between small sample and large amplitude interference,impulsive noise is suppressed and the robust detection of weak echo from severe interference is realized.Target can be detected from impulsive mixture noise when the input signal-to-interference ratio?SIR?is no less than-38dB,and from Gaussian noise when the SNR is above-38dB.The algorithm can mitigate the random impulsive interference effectively,improve the detection threshold,and reduce the false alarm rate.4.Range ambiguity resolving algorithm based on correcting range cell migrations is given through utilizing a set of pulse repetition intervals that are not pair-wisely co-prime.For locating moving target,a multi-pulse echo model based on multiple-burst sonar signal is established,and the effects of the target motion have on the detection performance of Chinese Remainder Theorem is analyzed.Traditional ambiguity resolving method is used to obtain a preliminary target range at first,which coarse estimation and is not exactly in the case of moving target.The proposed method could correct the reconstruction errors by evaluating the relationships among pulse repetition intervals.This algorithm performs well in resolving range ambiguity and compensating range migration when the target velocity satisfy the condition|v|?27??27?c(m_i?10?1-m_i)/?2m_i?.Furthermore,even though measuring errors?is unavoidable,the target range can also be reconstructed by the algorithm when??27?d/4,and the reconstruction accuracy is equivalent to the measuring errors.When the modulo meet the requirementm_i?29??29???c,the target velocity can be precisely estimated.5.The real time information processing and massive data transmission software for the underwater small target detection sonar is designed,and anechoic water tank experiments are conducted to investigate the acoustic characteristics of underwater target and the performance of the multi-pulse non-linear integration algorithm and the motion compensated coherent integration method for arbitrary waveform.The first water tank experiment,which is to confirm the proposed underwater target's acoustic characteristics models,demonstrates that the pulse reflection from the target is similar to the incident wave and the calculated TS of the target has fluctuations with the frequency and time.The TS of a bubble cloud?1L in volume?is about-24dB at 100kHz and a shot?5cm in radius?is-27dB.The results of the second experiment indicate that the multi-pulse non-linear integration algorithm can distinguish the target echo pulse from severe impulsive noise interference efficiently.Its performance outperforms the traditional linear integration method in that it can detect static target from impulsive noise when SIR-36dB and moving target when SIR-11dB.This algorithm can be also implemented in Gaussian noise scenario,it can detect static and moving target when SNR exceed-37dB and-14dB respectively.The third experiment show that the motion compensated coherent integration method is applicable to every waveform.These anechoic water tank experiments testified the fluctuating property of the TS of underwater small target,and proved the correctness and efficiency of the multi-pulse non-linear integration algorithm and the motion compensated coherent integration method.Those theories and methods supply an essential foundation for their promising engineering applications.