摘要
针对飞机装配过程中型架定位器的位置变化,提出一种基于光纤应变传感器的定位器位置检测方法。通过光纤应变传感器对型架主体及定位器组件关键特征区域的应变和定位器位置变化量进行实时采集,对采集的数据进行分析和处理,采用BP(back propagation,反向传播)神经网络建立飞机型架不同区域应变量与定位器位置变化量之间的数学模型。在某型工艺装备上搭建型架定位器位置检测试验平台,根据试验数据分别建立了定位器在平面及空间范围受载下的位置检测神经网络模型。试验结果表明:当神经网络样本数量一定时,模型对定位器位置的预测精度与加载偏移角度有关,偏移角度越小,加载覆盖率越高,模型的预测精度越高。其中,平面内以30°间隔、空间以45°间隔加载时,建立的神经网络模型对定位器位置的预测精度满足定位要求,验证了本文方法的可行性。该方法摈弃了用数字化设备直接测量的传统方式,弥补了因视线遮挡而造成个别定位器位置无法被检测的缺陷,能对装配过程中所有定位器的位置进行检测,解决定位器位置偏差过大造成的飞机部件检测质量差的问题,提升了飞机的装配质量。
A position detection method of positioner based on fiber-optic strain sensor was proposed for the position change of positioner in aircraft assembly process.Through the fiber-optic strain sensor,the strain of jig body and key characteristic areas of the positioner components and positioner position variation were collected in real time,and the collected data were analyzed and processed.BP(back propagation)neural network was used to establish the mathematical model between the stress variables in different areas of aircraft jig and the positioner position changes.A test platform for position detection of jig positioner was built on a certain type of process equipment,and the neural network model of position detection with positioner loaded in a plane and at space was established based on the test data.The experimental results showed that when the number of neural network samples was fixed,the prediction accuracy of the model to the position of the positioned was related to the loading migration angle.The smaller the migration angle,the higher the load coverage and the higher the model prediction accuracy.When the load position was 30°in the plane and 45°in the space,the prediction accuracy of the established neural network model for the position of positioner met the positioning requirements,which verified the feasibility of the proposed method.This method abandoned the traditional way of direct measurement with digital equipment,made up for the defect that individual positioned position couldn’t be detected due to line of sight occlusion.It can solve the problem of poor inspection quality of aircraft parts caused by excessive positioning deviation of positioner,and improve the assembly quality of aircraft.
作者
李现坤
曾德标
楚王伟
伍承旭
LI Xian-kun;ZENG De-biao;CHU Wang-wei;WU Cheng-xu(ChengduAircraft Industrial(Group)Co.,Ltd.,Chengdu 610092,China;SichuanAeronautical Intelligent Manufacturing Equipment Engineering Technology Research Center,Chengdu 610092,China)
出处
《工程设计学报》
CSCD
北大核心
2020年第3期287-292,共6页
Chinese Journal of Engineering Design
基金
成都飞机工业(集团)有限责任公司创新专项资金资助项目(40113000010X)。
关键词
定位器
应变传感器
虚拟仿真
神经网络
positioner
strain sensor
virtual simulation
neural network
