摘要
文章研究了无人机环境监测系统中的声电传感器信号分析方法,通过UrbanSound数据集进行实证验证。首先,介绍无人机环境监测系统的基本原理。其次,采用循环神经网络(Recurrent Neural Network,RNN)模型对声电传感器信号进行分类,同时引入丢弃法等正则化手段以优化模型性能。最后的实验表明,相较于传统RNN,正则化RNN在准确率、精确度和召回率等性能指标上均取得了显著提升。
In this paper,the signal analysis method of acoustoelectric sensor in unmanned aerial vehicle environmental monitoring system is studied and verified by UrbanSound data set.Firstly,the basic principle of unmanned aerial vehicle environmental monitoring system is introduced.Secondly,Recurrent Neural Network(RNN)model is used to classify the acoustic and electric sensor signals,and regularization methods such as discard method are introduced to optimize the model performance.The final experiment shows that compared with traditional RNN,regularized RNN has achieved significant improvement in accuracy,precision and recall rate.
作者
孙军辉
SUN Junhui(Zhengde Vocational and Technical College,Nanjing 210000,China)
出处
《电声技术》
2024年第2期113-115,共3页
Audio Engineering
关键词
无人机
声电传感技术
循环神经网络(RNN)
正则化
unmanned aerial vehicle
acoustic and electrical sensing technology
Recurrent Neural Network(RNN)
regularization