摘要
地震动传感器是无人值守系统非常重要的组成部分之一,在地面目标侦查工作中,地震动传感器有着非常重要的预警功能。针对传统传感器采用单一时域或频域分析法中出现的问题,研究创新性地选用时频域联合分析法,解决了传感器分析非平稳信号时准确率较低的问题。采用改进的BP算法,在系统中避免了频率特征匹配时出现网络功能函数中出现局部最小点的问题。研究设计的地震动信号识别传感器系统实现了3s、10s内对25m距离内的人和车以及100m范围内的各种车辆震动信号的识别,正确识别率分别达到96%和90%。
Ground motion sensor is one of the most important components of unattended system. In ground target detection,ground motion sensor has a very important early warning function.Aiming at the problems of traditional sensors using single time-domain or frequency-domain analysis method,the joint time-frequency domain analysis method is innovatively selected to solve the problem of low accuracy of sensors when analyzing nonstationary signals.The improved BP algorithm avoids the problem of local minimum in network function when frequency feature matching occurs in the system.The seismic signal recognition sensor system designed in this paper realizes the recognition of the vibration signals of people and vehicles within 25m and 100 m in 3s and 10s,and the correct recognition rate is 96% and 90% respectively.
作者
岳辉
YUE Hui(China Coal Institute Group Xi'an Research Institute Co.,Ltd.,Xi'an Shaanxi 710077,China)
出处
《粘接》
CAS
2019年第9期117-121,共5页
Adhesion
基金
国家重点研发计划:井下随掘巷道地质异常体动态探测技术与装备(2018YFC0807804)
关键词
震动信号
BP神经网络
模式识别
传感器
vibration signal
BP neural network
pattern recognition
sensor
