摘要
传统的被动目标识别主要依靠声呐员的作用,随着人工智能的迅速发展,水下目标智能识别成为未来发展的趋势。针对这一问题,根据舰船辐射噪声特性,提出基于特征融合的舰船目标识别方法,通过提取基于人耳听觉感知的梅尔倒谱系数特征、基于循环平稳分析的谱相关密度函数特征,构建特征层融合和决策层融合的特征融合模型,利用深度学习中的卷积神经网络进行舰船目标识别。利用4种舰船辐射噪声实测数据进行验证,结果表明,所提出的决策层融合算法能够明显提高舰船目标识别率。
The traditional passive target recognition mainly depends on the sonarman.With the rapid development of artificial intelligence,underwater target intelligent recognition becomes the future development trend.To solve this problem,according to the characteristics of ship radiated noise,a method of ship target recognition based on feature fusion is proposed.Based on the Mel frequency cepstrum coefficient feature and spectral correlation density function feature,a feature fusion model is constructed by using convolution neural network.The measured data of four underwater acoustic signal is verified.The result shows the decision level fusion algorithm can improve the recognition accuracy.
作者
王莹
侯朋
吴迪
王文冠
李光远
WANG Ying;HOU Peng;WU Di;WANG Wen-guan;LI Guang-yuan(Dalian Scientific Test and Control Technology Institute,Dalian 116013,China)
出处
《舰船科学技术》
北大核心
2022年第1期146-149,共4页
Ship Science and Technology
关键词
舰船辐射噪声
梅尔倒谱系数
谱相关密度函数
特征融合
卷积神经网络
ship radiation noise
Mel frequency cepstrum coefficient
spectral correlation density function
feature fusion
convolution neural network
