CASIA OpenIR  > 毕业生  > 硕士学位论文
基于热红外视觉的自动铺丝缺陷原位在线检测方法研究
陈梦娟
Subtype硕士
Thesis Advisor吴保林
2019-05
Degree Grantor中国科学院自动化研究所
Place of Conferral中国科学院自动化研究所
Degree Discipline控制工程
Keyword碳纤维复合材料,自动铺丝机,热红外视觉, 原位在线,缺陷检测
Abstract

碳纤维复合材料由于轻质、高比强度、高比模量、抗疲劳、耐腐蚀等优点,广泛应用于航空、航天、交通运输、建筑、新能源等领域。复合材料自动铺丝技术是各种大型复杂复合材料构件成型的主要制造方式,具有效率高、成本低、可重复性好、自动化程度高等特点。由于复合材料构件通常应用在对质量要求很高的行业,因此在自动铺丝生产过程中,质量监控是不可或缺的环节。铺丝缺陷是影响质量的主要因素,一般通过缺陷检测实现质量监控。原位在线缺陷检测是指不影响正常生产工艺流程、不占用生产以外时间的检测模式,是保证生产效率的前提。由于复合材料自动铺丝生产过程复杂,缺陷类型众多,特征不明显,且没有统一的判别标准,因此铺丝缺陷的自动检测是业内公认的难题。本文研究针对热红外视觉的自动铺丝缺陷原位在线检测问题,主要工作如下:
(1) 调研分析复合材料缺陷对构件质量的影响,确定影响构件性能的主要缺陷类型与尺寸阈值,为缺陷检测提供判别依据,包括缺陷类型和尺寸两个参量对复合材料构件的张力、压力强度等性能指标的影响。由于复合材料缺陷在可见光、近红外等频段的反射成像对比度低,缺乏有效特征。本文通过对自动铺丝生产过程中的热特性和缺陷形成机理分析,采用热红外辐射成像技术,并实验验证了热红外视觉对铺丝缺陷检测的有效性。同时,搭建了基于热红外视觉的缺陷检测硬件平台,使缺陷在温度曲线和热图像中显现出明显特征,为自动铺丝缺陷的检测提供基础。
(2) 提出了一种热红外视觉测量系统的标定方法。由于红外热像仪的硬件限制和热红外辐射成像的特殊性,使得热图像特征提取和红外热像仪标定成为难题,尤其是在需要考虑精度的工业测量中。本文研究了热红外视觉测量系统的标定技术,基于铺丝系统中的成像约束,建立了热红外视觉系统测量模型,提出了一种在小视场小景深下的测量系统参数标定方法,基于待检测点之间的几何约束与铺丝机器人的运动轨迹,标定红外视觉系统测量参数,并通过实验测定了标定后系统的测量精度。
(3) 提出了一种基于热红外视觉的自动铺丝缺陷原位在线检测方法。首先设计了检测系统结构,使检测结果与生产过程形成闭环,实现了可靠的质量控制与高效率的生产。采用卷积神经网络实现了对影响铺丝质量的主要缺陷类型的分类。针对热图像中缺陷成像模糊问题,提出了一种鲁棒的线段特征提取方法,实现了对可容忍类型缺陷大小和位置的测量,并通过实际的机器人在线铺丝实验验证了所提出方法的有效性和热红外视觉缺陷测量可行性。通过将检测结果与铺层模型匹配,构建了关联缺陷大小、位置、分布信息的三维缺陷分布模型,为铺丝构件提供产品身份证,为质量追溯提供依据。
(4) 开发了基于热红外视觉的自动铺丝缺陷检测软件系统。该系统具有自动铺丝热图像实时获取与显示、缺陷图像实时处理与显示、增长式三维缺陷分布模型实时构建与可视化,缺陷模型操作和检测报表生成等功能。设计多铺层复合材料自动铺丝实验,并提出主动缺陷注入法,在铺丝过程中注入已知类型、大小和位置信息的缺陷,作为铺丝实验的检测真值,测试了软件系统功能的有效性。
 

Other Abstract

Carbon Fiber Reinforced Polymer is widely used in the area of aviation, aerospace, transportation, construction, new energy and so on due to their light weight, high specific strength, high specific modulus, fatigue resistance and corrosion resistance. The automated fiber placement(AFP) is the major production way of composite,which has high processing efficiency, low manufacturing cost, good repeatability and high degree of automation. Since composite components are usually used in industries with high-quality requirements, quality inspection is an indispensable part in the production process of AFP. Since defects are the main factors affecting quality, quality inspection is generally achieved through defects detection. In-process online inspection that does not affect the normal production process and does not take up time outside production is a prerequisite for ensuring production efficiency. However, the AFP production process of composite is complicated and the types of defects are diverse and the feature of defects is not obvious. There is no unified industry standard up to now and the automatic defects detection during AFP production has been a long-standing problem. This article studies the in-process online detection method for AFP process based on the thermal infrared vision. The main work is as follows:
(1) The effect of defects on the composite quality is analyzed, including the defects type and defects size on the properties of composites such as tension and pressure strength. After analysis, the main defect types and size thresholds are determined. Composite material defects have low contrast in the visible light and lack effective features. So, this article uses thermal infrared vision imaging technology. After analyzing the thermal characteristics and formation mechanism of defect in the AFP process, thermal infrared vision is adopted in this article. The effectiveness of the detection method of defects based on thermal infrared vision is verified by experiments. A defect detection platform based on thermal infrared vision is built to help us to get the temperature curve and the thermal image in which the defects show obvious features in the temperature curve. This can provide a basis for the detection of defects in AFP process.
(2) The thesis proposes a calibration method of thermal infrared visual measurement system. Due to the hardware limitations of Infrared cameras and the special characteristics of thermal infrared imaging process, the feature extraction and the infrared cameras calibration face great challenges and thermal infrared vision has not been practically applied in industrial measurement. This paper studied the calibration technique of thermal infrared visual measurement system. Considering about the imaging constraints of the AFP system, the measurement model of the thermal infrared visual system is established. And a calibration method of the measurement system with small field of view and small depth of field is proposed. Based on the geometric constraints between the detected points and the trajectory of AFP, the parameters of the infrared vision system can be calibrated. The measurement accuracy of the system after calibration is determined by experiments.
(3) The thermographic in-process online detection method is proposed for AFP process. The in-process online detection system for AFP process is designed which realizes closed-loop of detection results and quality control. It makes the detection results can provide a basis for quality analysis and AFP production optimization. The classification of major defect types is realized and a thermal feature extraction method for tolerable defect types is proposed. Combined with the calibration results, the size and position measurement of the tolerable defect type can be realized, and the feasibility and effectiveness are verified by experiments. As the detection results are matched with the layering model, a three-dimensional defect distribution model associating the defect’s size, position and distribution information is constructed. This model provide identification for the components, which can be used for quality traceability.
(4) The defect detection software based on thermal infrared vision is developed for AFP. The software function consists of real-time acquisition and display of automated fiber thermal infrared images, real-time processing and display of defect images, a real-time construction of a growing 3D defect distribution model, defect model operation and detection report generation. The active defect injection method is proposed and the automated fiber experiment of multi-layer composite is designed. In the automated fiber process, defect with known type, size and position information is inject and the effectiveness of the software function is verified.

Pages113
Language中文
Document Type学位论文
Identifierhttp://ir.ia.ac.cn/handle/173211/23990
Collection毕业生_硕士学位论文
Recommended Citation
GB/T 7714
陈梦娟. 基于热红外视觉的自动铺丝缺陷原位在线检测方法研究[D]. 中国科学院自动化研究所. 中国科学院自动化研究所,2019.
Files in This Item:
File Name/Size DocType Version Access License
陈梦娟_硕士学位论文_基于热红外视觉的自(3992KB)学位论文 暂不开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[陈梦娟]'s Articles
Baidu academic
Similar articles in Baidu academic
[陈梦娟]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[陈梦娟]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.