基于短时频谱分布统计特征的切割机声信号识别算法  

Acoustic Signal Recognition Based on Short-term Spectrum Distribution Statistical Characteristics for Cutting Machine

作　　者：王玮[1] 王建中[1] 杨三伟 

机构地区：[1]杭州电子科技大学,浙江杭州310018

出　　处：《工业控制计算机》2016年第2期10-12,共3页Industrial Control Computer

摘　　要：电缆在使用中常被挖掘设备(挖掘机、切割机、液压冲击锤等)所破坏。因此,建立一个监测系统可以自动检测识别开挖设备的监测系统至关重要。切割机作为开挖设备的主要组成部分,在开挖现场得到广泛的运用。针对它的特点,提出了一种基于统计特性的分类方法。该算法不需要建立声指纹库,核心是设计将它与马路汽车噪声分类识别的分类器。通过理论分析及实验表明,该算法正确率能达到95%以上,同时噪声的误判率控制在5%以下。说明该算法用以切割机的分类识别是可行的。Constructing a surveillance system which can automatically detect potential damages caused by excavation equipment is essential and urgent for power supply system.Cutting machine is the main part of excavation equipment,which is widely used. Based on its characteristics,this paper proposes a classification method based on statistical feature to recognize it.The algorithm does not need to build sample library,the core idea is to set up mathematical model according to their statistical features,and design the corresponding classifier.

关 键 词：切割机 统计特性 分类器 分类识别 

分 类 号：TN911.7[电子电信—通信与信息系统]