首页|期刊导航|金属加工(热加工)|多模态传感融合技术下火电厂高温管道焊接微裂纹检测方法

多模态传感融合技术下火电厂高温管道焊接微裂纹检测方法OA

Multimodal sensing fusion technology for detecting welding microcracks in high-temperature pipelines of thermal power plants

中文摘要英文摘要

火电厂管道材料在高温环境下易发生蠕变、相变等非线性行为,导致其力学性能随时间产生退化,在裂纹扩展的连续状态与离散决策之间呈现逻辑缺口,产生微裂纹漏检率高、误判风险大等问题.为此,提出多模态传感融合技术下火电厂高温管道焊接微裂纹检测方法.通过超声波、红外、机器视觉3种传感器协同采集管道焊接数据,经A/D转换为数字信号后,利用卡尔曼滤波构建微裂纹位置与扩展速度的状态方程,结合方差计算和线性加权平均融合多模态信息,并通过误差协方差矩阵优化融合结果.设计卡尔曼滤波与CART分类回归树算法的协同框架,填补从连续状态到工程决策的逻辑链条,利用分类回归树构建微裂纹类别分类器,将融合后的状态估计值及多源特征输入分类器,实现微裂纹位置、类型的精准检测.试验结果表明:多模态传感融合技术可准确检测出高温管道焊接微裂纹位置,且多分类对数损失较低.

Pipeline materials in thermal power plants are susceptible to nonlinear behaviors such as creep and phase transformation under high-temperature environments,leading to the degradation of their mechanical properties over time.A logical gap emerges between the continuous state of crack propagation and discrete decision-making,resulting in a high missed-detection rate and a significant risk of misjudgment for microcracks.To address this issue,a welding microcrack detection method for high-temperature pipelines in thermal power plants based on multimodal sensor fusion technology is proposed.Three types of sensors-ultrasonic sensor,infrared thermal imager,and machine vision sensor-collaboratively acquire pipeline welding data.After A/D conversion into digital signals,the Kalman filter is employed to construct state equations for microcrack position and propagation velocity.Multi-modal information is fused through variance calculation and linear weighted averaging,and the fusion result is optimized via the error covariance matrix.A collaborative framework integrating the Kalman filter and the CART algorithm is designed to bridge the logical gap from continuous state estimation to engineering decision-making.A microcrack category classifier is built using the classification and regression tree,into which the fused state estimation values and multi-source features are input to achieve accurate detection of microcrack position and type.Experimental results demonstrate that the multimodal sensor fusion technology can accurately detect welding microcrack positions in high-temperature pipelines,with a relatively low multi-class log loss.

郭剑;王宗喜;岳晓光;李鑫;刘智

中国电力工程顾问集团华东电力设计院有限公司 上海 200300安徽钱营孜发电有限公司 安徽宿州 234000安徽钱营孜发电有限公司 安徽宿州 234000中国电力工程顾问集团华东电力设计院有限公司 上海 200300施耐德电气(中国)有限公司 北京 100020

多模态传感融合技术火电厂高温管道焊接微裂纹检测卡尔曼滤波分类回归树

multimodal sensing fusion technologyhigh temperature pipelines in thermal power plantswelding microcrack detectionkalman filterclassification and regression tree

《金属加工(热加工)》 2026 (10)

68-74,83,8

评论