PPG信号用于高温高湿伤害预警的可行性研究

来源期刊:中南大学学报(自然科学版)2020年第5期

论文作者:金龙哲 王梦飞 于露 徐明伟

文章页码:1424 - 1431

关键词:光电容积脉搏波;高温高湿;残差神经网络;预警;信号识别

Key words:photoplethysmography; high temperature and humidity; residual neural network; early warning; signal recognition

摘    要:考虑到高温高湿环境对人体产生不良影响,影响安全生产,对光电容积脉搏波(PPG)信号用于识别人体进入高温高湿环境的可行性进行研究。采集人体处于高温高湿环境和舒适环境的光电容积脉搏波信号,将信号人工标记后输入残差神经网络进行训练,获得预警模型。研究结果表明:人体在高温高湿环境和舒适环境下光电容积脉搏波信号存在差异,这种差异可以被预警模型识别;训练集识别准确率为99.8%,测试集识别准确率为98.3%,训练集和测试集对高温高湿环境的查准率分别为99.8%和96.9%,查全率分别为99.8%和99.7%;训练结果的受试者工作特征曲线(ROC)与横坐标所围成的面积AUC接近于1,因而认为该模型有较强的泛化能力;所使用的方法预警时间为0.512 s,即在人体生理参数变化后0.512 s对人员发出预警。

Abstract: High temperature and high humidity environment has adverse effects on human body and affects safety production. Photoplethysmography(PPG) signal was used to identify whether human body enters high temperature and high humidity environment. The photoplethysmography signals of human body in high temperature and humidity environment and comfortable environment were collected.The signals were labeled artificially and then input into residual neural network for training,and an early warning model was obtained. The results show that there are differences in photoplethysmography between human body in high temperature and humidity environment and comfortable environment,which can be recognized by early warning model. The recognition accuracy of training set is 99.8%, and that of test set is 98.3%. The precision of training set and test set to high temperature and humidity environment are 99.8% and 96.9%, respectively. The recall rate of training set and test set to high temperature and humidity environment are 99.8% and 99.7%, respectively. The area AUC enclosed by the ROC curve and the abscissa is close to 1, so the model has strong generalization ability. The early warning time of the method used in this paper is 0.512 s, i.e.,warning is given to human after 0.512 s of the change of human physiological parameters.

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号