摘要
分布式测温传感器被广泛运用于温度监测。本文以此为背景,提出了一种具有温度监测和温度预测功能的分布式光纤系统。在实现温度监测的同时,建立了一种基于机器学习的趋势外推模型,通过不断更新的历史数据来预测未来温度。实验模拟了一般火灾的温度变化并进行预测。结果表明,当采样时间为5 s,预测1 min后的温度时,预测误差最大值为2.88℃,均方误差为0.958℃,说明模型准确度较高,对工程实践具有一定的参考价值。
Distributed temperature sensors are widely used for temperature monitoring.Based on the background,this article proposes a distributed fiber optic system with temperature monitoring and prediction functions.While implementing temperature monitoring,a trend extrapolation model based on machine learning has been established to predict future temperatures through continuously updated historical data.The experiment simulated the temperature changes of general fires and made predictions.The results show that when the sampling time is 5 seconds and the temperature is predicted after 1 minute,the maximum prediction error is 2.88 ℃,and the MSE is 0.958 ℃,indicating that the model has high accuracy and has certain reference value for engineering practice.
作者
刘成柱
郑宇斌
杨展
邓琴
施春龙
张聪慧
Liu Chengzhu;Zheng Yubin;Yang Zhan;Deng Qin;Shi Chunong;Zhang Conghui(Zhejiang Provincial Zhoushan Coal Power Company,Zhoushan,China;Huzhou Institute of Zhejiang University,Huzhou,China)
出处
《科学技术创新》
2024年第8期216-220,共5页
Scientific and Technological Innovation
关键词
光纤测温
趋势外推法
机器学习
多项式拟合
optical fiber temperature measuring
trend extrapolation method
machine learning
polynomial fitting
