数据驱动技术的光传感器状态智能检测研究  

Research on intelligent detection of optical sensor state based on data-driven technology

作　　者：李景富[1] 李福荣[1] LI Jingfu;LI Furong(Huanghuai University,Zhumadian Henan 463000,China)

机构地区：[1]黄淮学院,河南驻马店463000

出　　处：《激光杂志》2022年第12期164-168,共5页Laser Journal

基　　金：河南省重点科技攻关项目(No.192102210285)。

摘　　要：为了及时准确地检测出光传感器的各种故障,以解决当前光传感器检测方法存在的局限性,设计了一种基于数据驱动技术的光传感器状态检测方法。该方法首先采集光传感器状态信号,并采用经验分解模态法和独立分量法对光传感器状态信号进行预处理,然后提取光传感器状态检测特征向量,最后采用数据驱动技术将特征向量作为输入信息进行建模,输出光传感器状态类型,具体的光传感器状态智能检测结果表明,本方法可以有效描述各种光传感器状态,光传感器状态检测正确率超过92%,而光传感器状态拒检率和误检率均低于5%,光传感器状态信号时间少于5 ms,光传感器状态智能检测性能要明显好于当前经典光传感器状态检测方法,具有一定的实际应用价值。In order to detect various faults of optical sensor timely and accurately and solve the limitations of current optical sensor detection methods,an optical sensor state detection method based on data-driven technology is designed.Firstly,the optical sensor state signal is collected,and the empirical decomposition mode method and independent component method are used to preprocess the optical sensor state signal,and then the optical sensor state detection feature vector is extracted.Finally,data-driven technology is used to model the feature vector as the input information and output the optical sensor state type,The specific intelligent detection results of optical sensor state show that this method can effectively describe various types of optical sensor state.The detection accuracy of optical sensor state exceeds 92%,while the rejection rate and false detection rate of optical sensor state are less than 5%,and the signal time of optical sensor state is less than 5 ms,The intelligent detection performance of optical sensor state is obviously better than the current classical optical sensor state detection methods,which has a certain practical application value.

关 键 词：数据驱动技术 光传感器 状态信号 极限学习要机 状态拒检率 状态误检率 

分 类 号：TN929[电子电信—通信与信息系统]