文献类型：期刊文献

中文题名：基于AFSA-GPR的超声测厚温度补偿研究

英文题名：Research on Temperature Compensation for Ultrasonic Thickness Measurement Based on AFSA-GPR

作者：杨冬旭[1];李祥[1];贾九红[1]

机构：[1]华东理工大学承压系统与安全教育部重点实验室

年份：2025

期号：2

起止页码：91

中文期刊名：仪表技术与传感器

外文期刊名：Instrument Technique and Sensor

收录：;北大核心:【北大核心2023】；

基金：国家自然科学基金(52175138)。

语种：中文

中文关键词：高温承压设备;超声测厚;温度补偿;高斯过程回归;人工鱼群算法

外文关键词：high-temperature pressure equipment;ultrasonic thickness measurement;temperature compensation;Gaussian process regression;artificial fish-swam algorithm

摘要：使用超声信号对高温承压设备进行壁厚在线监测时,温度的变化会影响壁厚测量结果。针对这一问题,提出一种基于人工鱼群算法优化高斯过程回归的AFSA-GPR温度补偿模型。采用人工鱼群算法对高斯过程超参数进行寻优以提高模型预测精度。在室温(25℃)至500℃环境下进行超声测厚试验研究,结果表明,该温度补偿模型能显著提升高温环境下壁厚测量精度,其MAE为0.014 8 mm, RMSE为0.022 3 mm。
When using ultrasonic signals for online monitoring of wall thickness of high-temperature pressure equipment,changes in temperature can affect wall thickness measurements.To solves this problem,an AFSA-GPR temperature compensation method based on Gaussian process regression optimized by artificial fish-swam algorithm was proposed.The artificial fish-swam al-gorithm was used to optimize Gaussian process hyperparameters to improve model prediction accuracy.The ultrasonic thickness measurement experiment was conducted from room temperature(25℃)to 500℃,and the result shows that the temperature compensation model significantly improves the accuracy of wall thickness measurement in high-temperature environments.The MAE is 0.0148 mm and the RMSE is 0.0223 mm.