摘要
针对当前井下不安全行为识别率低的问题,提出了一种基于信道状态信息的井下人员行为识别方法。首先利用SOM聚类算法与多种滤波方法去除井下噪声,其次利用窗口函数和卷积神经网络优化状态判断模型,最后利用卷积神经网络输出行为识别结果。实验结果表明,所提方法数据分割准确率达98.4%,优于阈值法;行为识别准确率达到93%,优于HDFi,Wi-ACR算法。
In order to solve the problem of low recognition rate of underground unsafe behavior, a method of underground personnel behavior recognition based on channel state information is proposed.First of all, the SOM clustering algorithm and a variety of filtering methods are used to remove the downhole noise, then the window function and convolution neural network are used to optimize the state judgment model, and finally the convolution neural network is used to output the behavior recognition results. The experimental results show that the data segmentation accuracy of the proposed method is98.4%, which is better than the threshold method, and the behavior recognition accuracy is 93%, which is better than HDFi and Wi-ACR algorithms.
作者
梁晨阳
华钢
LIANG Chenyang;HUA Gang(School of Information and Control Engineering,China University of Minging and Technology,Xuzhou 221000,China)
出处
《煤炭技术》
CAS
北大核心
2022年第11期182-186,共5页
Coal Technology
基金
国家自然科学基金面上资助项目(51574232)。
关键词
井下人员
行为检测
信道状态信息
特征重构
数据分割
卷积神经网络
underground personnel
behavior detection
channel state information
feature reconstruction
data segmentation
convolution neural network
