摘要
基于拉曼散射的分布式光纤温度传感器(RDTS)能够沿光纤沿线测量数十公里的温度,被广泛用于监测关键设施的温度状况。由于受到温度传递滞后和单次解调运行时间的影响,系统在面对火灾等温度突变事故中还无法快速有效地实现对安全隐患的超前预警。为从源头上对安全隐患采取预防措施,做到安全准确和防患于未然,研究人员提出了阈值预警、温差预警以及建立基于多阶实时移动平均法的预警模型等多种方法。然而,这些方法都有其局限性,为此本文提出了一种基于分布式温度传感的在线学习自适应模糊温度预测法。实验结果表明,基于在线学习的自适应模糊温度预测方法能够在保证预测精度的前提下有效地提高预警时间。与传统方案相比,该方法不受系统硬件的限制,超前预警时间可根据需求自由设置,能同时对被测光纤任意位置进行分布式预测,其在实际应用中具有显著优势。当超前点数设定为2时,系统能够达到的平均预测温度的绝对误差为1.1℃,平均预测误差的百分率为3.06%,预测误差的波动范围约为±2.3℃。
The distributed optical fiber temperature sensor(RDTS)based on Raman scattering can measure the temperature of tens of kilometers along the optical fiber,and is widely used to monitor the temperature of key facilities.Due to the influence of temperature transmission lag and single demodulation operation time,the system cannot quickly and effectively realize the early warning of potential safety hazards in the face of temperature sudden change accidents such as fire.In order to take preventive measures against potential safety hazards from the source and achieve safety accuracy and prevention,the researchers have proposed threshold early warning,temperature difference early warning,and the establishment of early warning model based on multi-stage real-time moving average method.However,these methods have their limitations.Therefore,this paper proposes an online learning adaptive fuzzy temperature prediction method based on distributed temperature sensing.The experimental results show that the adaptive fuzzy temperature prediction method based on online learning can effectively improve the warning time on the premise of ensuring the prediction accuracy.Compared with the traditional scheme,this method is not limited by the system hardware,and the advance warning time can be set freely according to the demand.It can predict any position of the tested optical fiber at the same time,which has significant advantages in practical application.When the number of leading points is set to 2,the absolute error of the average prediction temperature that the system can achieve is 1.1°C,the percentage of the average prediction error is 3.06%,and the fluctuation range of the prediction error is about±2.3℃.
作者
李宁
尚雯珂
LI Ning;SHANG Wenke(College of Electrical Engineering,Xi’an Jiaotong University,Xi’an 712046,Shaanxi,China)
出处
《电气传动自动化》
2022年第4期52-57,共6页
Electric Drive Automation
关键词
分布式光纤温度传感
拉曼散射
在线学习
自适应模糊温度预测方法
distributed optical fiber temperature sensing
Raman scattering
online learning
adaptive fuzzy temperature prediction method
