舰船辐射噪声中鲸类联络叫声检测方法

Automatic detection of whale contact calls in background of ship-radiated noise

鲸类联络叫声的频段与舰船辐射噪声频段的重叠,使其易被误检测为舰船噪声而被被动声纳接收,影响后续对被动声纳检测到的信号进行目标自主识别等信号处理的性能。针对鲸类联络叫声干扰目标识别等声纳信号后续处理的问题,提出舰船辐射噪声中鲸类联络叫声检测算法,为削除鲸类联络叫声干扰作准备,算法首先通过理论与实测数据分析得出舰船辐射噪声频谱分布的短时稳定性特征,进而采用中值滤波方法对舰船辐射噪声进行白化处理,利用三阶多项式相位信号对鲸类叫声建模,然后通过基于匹配滤波器的广义似然比检测方法,实现了舰船辐射噪声中鲸类信号的检测。经实测信号的实验验证,提出的算法适用于不同类型和工况的舰船;通过与其他三种检测方法的比较分析,算法在给定虚警率下可以取得较高的鲸类叫声检测率,检测结果优于其他三种方法。

Some whale contact calls have been the interference of ship-radiated noise collected by passive sonar for their frequency bands overlapping. Therefore, that whale calls should be detected and eliminated to improve the further target autonomous recognition. So, in this paper, a two-stage whale calls detection algorithm is proposed, in which, in the first stage, the whale calls is modeled by polynomial phase signals and the ship-radiated noise is modeled by locally stationary Gaussian random process after prewhitening based on its frequency stability and then the whale contact calls are detected using the generalized likelihood ratio test calculating on a bank of match filters. The experiment results benchmarked on a large measured data show that the proposed method is available and can be used in various ship-radiated noise environments. The results of a further experiment comparing with another three methods show that the proposed method is outperformed the other three methods and can achieve a good whale calls detection result in a low false alarm.