Sonar locates underwater targets by receiving reflected sound waves.However,the complex marine environment makes it difficult to set targets in appropriate locations and conditions.The sonar echo simulator has the fun...Sonar locates underwater targets by receiving reflected sound waves.However,the complex marine environment makes it difficult to set targets in appropriate locations and conditions.The sonar echo simulator has the function of simulating sonar detection target echo signals.As a cutting-edge technology,underwater backscatter has led to the emergence of array based acoustic reflection systems.The research on sonar echo simulators based on backscatter technology has promoted the solution of problems such as target echo modeling,sonar arrival direction estimation,and echo directional transmission.In response to the above problems,this paper designs an end-to-end sonar echo simulator system array phase estimation sonar echo simulation system(APE-SESS),which can independently complete highresolution real-time direction of arrival(DOA)estimation and generate directional simulation echo based on the array structure.Dual branch convolutional neural network(DB-CNN)is proposed in the system to estimate the direction of the signal array and directly obtain the phase weights containing azimuth information.Comparing DBCNN with conventional methods and classic underwater DOA network models based on classification problems,the results show that DB-CNN exhibits stability,small error,and high real-time performance under different signal-to-noise ratio(SNR).The proposed APE-SESS has end-to-end characteristics,real-time angle estimation,and azimuth simulation functions.展开更多
混响是主动声纳检测的主要背景干扰,由于它是一种非平稳的有色噪声,使得工作在白噪声条件下的检测器性能受到极大限制。在混响背景下实现目标回波检测,常采用自回归(AR)模型对宽带回波预白化处理,但在强混响背景条件下,白化后直接进行...混响是主动声纳检测的主要背景干扰,由于它是一种非平稳的有色噪声,使得工作在白噪声条件下的检测器性能受到极大限制。在混响背景下实现目标回波检测,常采用自回归(AR)模型对宽带回波预白化处理,但在强混响背景条件下,白化后直接进行匹配滤波检测的结果不甚理想。针对此问题,在AR模型预白化基础上,提出一种改进方法,对白化后信号先进行二分奇异值分解(SVD)处理,有效去除大部分混响干扰,然后再作匹配检测。仿真实验分析表明,相比于仅白化后的匹配滤波检测,该方法可提高信混比约3 d B,匹配检测效果得到了明显改善。展开更多
基金supported by the Joint Funds of the National Natural Science Foundation of China under Grant U22A2009in part by the National Key Research and Development Program under Grant 2021YFC2803000.
文摘Sonar locates underwater targets by receiving reflected sound waves.However,the complex marine environment makes it difficult to set targets in appropriate locations and conditions.The sonar echo simulator has the function of simulating sonar detection target echo signals.As a cutting-edge technology,underwater backscatter has led to the emergence of array based acoustic reflection systems.The research on sonar echo simulators based on backscatter technology has promoted the solution of problems such as target echo modeling,sonar arrival direction estimation,and echo directional transmission.In response to the above problems,this paper designs an end-to-end sonar echo simulator system array phase estimation sonar echo simulation system(APE-SESS),which can independently complete highresolution real-time direction of arrival(DOA)estimation and generate directional simulation echo based on the array structure.Dual branch convolutional neural network(DB-CNN)is proposed in the system to estimate the direction of the signal array and directly obtain the phase weights containing azimuth information.Comparing DBCNN with conventional methods and classic underwater DOA network models based on classification problems,the results show that DB-CNN exhibits stability,small error,and high real-time performance under different signal-to-noise ratio(SNR).The proposed APE-SESS has end-to-end characteristics,real-time angle estimation,and azimuth simulation functions.
文摘混响是主动声纳检测的主要背景干扰,由于它是一种非平稳的有色噪声,使得工作在白噪声条件下的检测器性能受到极大限制。在混响背景下实现目标回波检测,常采用自回归(AR)模型对宽带回波预白化处理,但在强混响背景条件下,白化后直接进行匹配滤波检测的结果不甚理想。针对此问题,在AR模型预白化基础上,提出一种改进方法,对白化后信号先进行二分奇异值分解(SVD)处理,有效去除大部分混响干扰,然后再作匹配检测。仿真实验分析表明,相比于仅白化后的匹配滤波检测,该方法可提高信混比约3 d B,匹配检测效果得到了明显改善。
