摘要
针对复杂构件装配过程中装配工装应变敏感区域直接监测困难的问题,提出了工装应变高精预测模型,基于支持向量机设计了一种复杂结构应变动态监测与高精预测方法。利用光纤光栅传感器获取模型所需输入、输出应变信息,建立输入输出数据间的非线性映射,对模型加以训练,最终达到精准预测复杂构件在装配过程中感生应变的目的。该方法解决了空间受限时,装配工装复杂零件应变状态在线监控与预测难题。针对工装复杂零件的应变监测与预测试验结果表明,该预测模型的R2高达0.98,RMSE为0.2040με,具有较高的精度和较强的泛化能力,可满足复杂构件装配的实际检测需求。
Aimed at the problem of difficulty of directly monitoring the strain-sensitive areas of assembly tooling during the assembly process of complex components,a high-precision prediction model of tooling strain was proposed.Based on support vector machine(SVM),a dynamic monitoring and high-precision prediction of complex structure strain was designed.The fiber grating sensor was used to obtain the input and output strain information required by the model,established a non-linear mapping between the input and output data,and trained the model to achieve the purpose of accurately predicting the complex components induced strain during assembly.This method solved the problem of online monitoring and prediction of the strain state of complex parts of assembly tooling under space constraints.The experimental results of strain monitoring and prediction for tooling complex parts showed that the R 2 of the prediction model was as high as 0.98 and the RMSE was 0.2040με,which had high accuracy and strong generalization ability,and met the actual inspection needs of complex component assembly.
作者
姜昕彤
梁冰
刘坤
冯荻
贾振元
刘巍
JIANG Xintong;LIANG Bing;LIU Kun;FENG Di;JIA Zhenyuan;LIU Wei(College of Mechanical Engineering,Dalian University of Technology,Dalian 116024,China)
出处
《新技术新工艺》
2020年第5期41-46,共6页
New Technology & New Process
基金
辽宁省“兴辽英才计划”资助项目(XLYC1807086)
大连市高层次人才创新支持计划资助项目(2017RJ04)
国家自然科学基金辽宁联合基金资助项目(U1808217)
国家创新群体基金资助项目(51621064)。
关键词
装配工装
复杂构件
光纤应变传感器
光纤光栅
应变预测
支持向量机
assembly tooling
complex components
fiber strain sensor
fiber grating
strain prediction
support vector machine
作者简介
姜昕彤(1995-),女,硕士研究生,主要从事复杂构件应变监控与预测等方面的研究。
