全文下载： 
202201014.pdf
文章编号： 1672-6987（2022）01-0100-05； DOI： 10.16351/j.1672-6987.2022.01.014
孙晓1, 王清梅2， 楚敬敬1(1.青岛科技大学 自动化与电子工程学院，山东 青岛 266061；2.中国科学院国家天文台，北京 100012)
摘要： 为保障FAST大型射电望远镜安全工作，使用传感器对其关键点应力信息进行监测与评估，对应力信号进行预测，有助于故障的早期预警与处理，提高安全性能。基于物理模型和人工智能模型预测的方法，往往需要结构详细特性信息，或相关荷载信号作为输入，建模过程较为复杂。本研究使用时间序列分析法，基于ARIMA模型对FAST光纤Bragg光栅应变计实际输出信号进行了建模与测试，利用二阶差分将信号平稳化，并确定模型阶数为ARIMA(4,2,9)。通过模型训练与实际数值比对，长期预测中模型可反映信号变化趋势与规律，短期预测中，预测值与实际值间均方根误差仅为0001 3 nm，模型简单且仅依赖数据本身即可实现预测，预测精度满足FAST实际监测预测需求。
关键词： FAST大型射电望远镜; 时间序列分析; ARIMA模型; 信号预测
中图分类号： TP 277;TP 111.44文献标志码： A
引用格式： 孙晓，王清梅,楚敬敬.基于时间序列的结构应力监测信号预测方法［J］. 青岛科技大学学报(自然科学版), 2022, 43(1): 100-104.
SUN Xiao， WANG Qingmei, CHU Jingjing. Prediction method of structural stress monitoring signal based on time series［J］. Journal of Qingdao University of Science and Technology（Natural Science Edition), 2022， 43（1）： 100-104.
Prediction Method of Structural Stress Monitoring
Signal Based on Time SeriesSUN Xiao1, WANG Qingmei2, CHU Jingjing1
(1.College of Automation and Electronic Engineering, Qingdao University of Science and Technology, Qingdao 266061, China;
2.National Astronomical Observatories, Chinese Academy of Science, Beijing 100012, China)
Abstract: Five-hundred-meter aperture spherical telescope (FAST) uses sensors to monitor the stress information of key structural points to ensure its safety. The prediction of stress signal is helpful to the early warning and processing of fault, and improves the safety performance. Prediction methods based on physical models or artificial intelligence models often require detailed structural information or related load signals as input. And the modeling process is more complicated. This paper uses time series analysis method based on ARIMA model. The real output signal of fast fiber Bragg grating strain gauge is tested. Use second order difference to stabilize the signal, and determine the model ARIMA(4,2,9). By comparing the model predicted value with the actual value, the model can reflect the signal change trend and law in the long-term prediction. In the short-term prediction, the root mean square error of the predicted value and the actual value is only 0001 3 nm. The model is simple and only depends on the data itself to achieve predictions. The prediction accuracy meets the actual monitoring and prediction requirements of FAST.
Key words: five-hundred-meter aperture spherical telescope; time series analysis; ARIMA model; signal prediction
收稿日期： 2021-01-30
基金项目： 国家自然科学基金项目(11803053).
作者简介： 孙晓(1987—), 男,博士.
