摘要
为了提升煤矿环境下周界安防系统的电磁兼容和信号识别能力,基于三维残差(R3D)神经网络,提出了一种抗干扰的振动信号分类方法。针对煤矿复杂电磁环境对周界信号的干扰问题,利用传感技术结合深度学习算法,高效地对振动异常信号进行特征提取和识别。在方法上,系统对信号进行傅叶变换,并通过R3D模型提取了时空特征,以判定信号类别。在实验过程中,基于信号真实数据集进行了验证。实验结果显示,该算法在强电磁干扰环境下的识别准确率达到92.5%,误报率下降至4.3%,论证了其在复杂煤矿场景下的抗干扰能力和实时性。研究表明,该方法为煤矿周界安防系统的智能化和抗干扰性提升提供了技术支持。
In order to enhance the electromagnetic compatibility and signal recognition capability of perimeter security systems in coal mine environments,this paper proposes an anti-interference vibration signal classification method based on Three-Dimensional Residual(R3D)neural network.In response to the interference problem of surrounding signals in the complex electromagnetic environment of coal mines,sensing technology combined with deep learning algorithms is used to efficiently extract and identify abnormal vibration signals.In terms of methodology,the system performs Fourier transform on the signal and extracts spatiotemporal features through the R3D model to determine the signal category.During the experiment,validation was conducted based on a real dataset of signals.The experimental results show that the algorithm achieves a recognition accuracy of 92.5%in strong electromagnetic interference environments,with a false alarm rate reduced to 4.3%,demonstrating its antiinterference ability and real-time performance in complex coal mine scenarios.Research has shown that this method provides technical support for the intelligence and anti-interference improvement of coal mine perimeter security systems.
作者
林洋
LIN Yang(Shanghai Coal Tech Inspection Co.,Ltd.,Shanghai 200003,China)
出处
《自动化应用》
2025年第16期252-256,共5页
Automation Application
关键词
定位型振动光纤
视频分类
R3D模型
positioning type vibration optical fiber
video classification
R3D model
