应用原型网络的小样本次声信号分类识别方法  

A method for classification of few-shot infrasound signals applying prototype network

作　　者：赵子杰 程巍 姬培锋[1,2] 滕鹏晓[1,2] 吕君[1,2] 杨军[1,2,3] ZHAO Zijie;CHENG Wei;JI Peifeng;TENG Pengxiao;LYU Jun;YANG Jun(Institute of Acoustics,Chinese Academy of Sciences,Beijing 100190,China;Key Laboratory of Noise and Vibration Research,Institute of Acoustics,Chinese Academy of Sciences,Beijing 100190,China;University of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]中国科学院声学研究所,北京100190 [2]中国科学院噪声与振动重点实验室,北京100190 [3]中国科学院大学,北京100049

出　　处：《应用声学》2024年第6期1193-1202,共10页Journal of Applied Acoustics

基　　金：国家自然科学基金项目(11874389)。

摘　　要：地震、闪电、火箭发射、爆炸等活动都会伴随着次声信号的产生。为提升次声事件的监测能力,需要对小样本的次声信号进行正确分类识别。针对小样本集的次声事件的有效识别问题,结合长短期记忆模型提出了一种应用原型网络的次声信号分类方法。使用该方法分别对公开的次声信号数据集和实地采集的地震、爆炸、闪电、火箭再入产生的4类次声信号进行分类实验。实验结果表明,该方法相对于传统方法,简化了特征提取的过程,有效解决了小样本集次声信号的特征分析问题,取得较好的分类结果和泛化效果。Events such as earthquakes,lightning,rocket launches,and explosions are accompanied by infra-sound signals.In order to improve the monitoring capability of infrasound events,it is necessary to correctly classify small samples of infrasound signals.For the problem of effective identification of infrasound events with small samples and variable duration,a classification method of infrasound signals applying prototype net-work is proposed in combination with a long and short-term memory model.The method is used to conduct classification experiments on publicly available infrasound signal datasets and four types of infrasound signals generated by earthquakes,explosions,lightning,and rocket re-entry collected in thefield.The experimental results show that the method simplifies the process of feature extraction and effectively solves the problem of feature analysis of variable duration infrasound signals compared with the traditional method,and achieves better classification results and generalization effects.

关 键 词：次声 小样本 原型网络 长短期记忆模型 

分 类 号：TN911.7[电子电信—通信与信息系统]