基于智能用电网络的低成本实时人员检测方法  

Low-cost Real-time Personnel Detection Method Based on Intelligent Power Network

作　　者：郭治远 刘伟峰[1] 何光宇[1] 周欢[1] 范帅 Guo Zhiyuan;Liu Weifeng;He Guangyu;Zhou Huan;Fan Shuai(Key Laboratory of Control of Power Transmission and Conversion Control,Ministry of Education,Shanghai Jiao Tong University,Shanghai 200240,China)

机构地区：[1]上海交通大学电力传输与功率变换控制教育部重点实验室,上海200240

出　　处：《电气自动化》2023年第3期61-63,67,共4页Electrical Automation

基　　金：国家重点研发计划(2019YFE0122600)。

摘　　要：房间有无人员状态是实施节能和安全等智能用电行为的基础信息,通常基于房间内的信道状态信息(channel state information,CSI)来间接检测有无人员状态。为此,对不同场景的CSI幅值进行了统计分析,并提出其数据的修正瑞利分布特性。在此基础上,以瑞利分布形状参数、右偏修正系数和形状参数变化率为特征,选取敏感子载波构建特征指纹矩阵,最后以实时特征指纹与有人、无人时的历史特征指纹矩阵之间的F范数来判定有无人员状态。此外,为提高实时人员检测的时效性,提出了CSI检测终端与边缘计算网关的协同计算模式。算例分析以典型办公室为试验地,与基于随机森林的判定方法进行对比。结果表明所提方法及其技术架构不仅更适宜实现长时间的高精度检测,还具备低成本的特性。The presence or absence of occupants in a room is the basic information for implementing smart power consumption behaviors such as energy saving and safety.The presence or absence of occupants is usually detected indirectly based on the channel state information(CSI)in the room.To this end,the CSI amplitudes of different scenarios were statistically analyzed and the modified Rayleigh distribution characteristics of the data were then proposed.On this basis,the Rayleigh distribution shape parameters,right-bias correction coefficients and shape parameter change rates were characterized,and the sensitive sub-carriers were selected to construct a feature fingerprint matrix.The F-norm between them was adopted to determine the presence or absence of personnel status.In addition,in order to improve the timeliness of real-time personnel detection,a collaborative computing mode of CSI detection terminal and edge computing gateway was put forward.The example analysis took a typical office as the experimental site,and compared it with the judgment method based on random forest.The results show that the proposed method and its technical architecture are not only more suitable for long-term high-precision detection,but also have the characteristics of low cost.

关 键 词：信道状态信息 特征指纹 人员检测 实时检测 低成本 

分 类 号：TM714[电气工程—电力系统及自动化]