QQ客服 官方微博 反馈留言

题名	基于红外热波检测的缺陷识别方法研究
作者	张志强
学位类别	硕士
答辩日期	2011-05-27
授予单位	中国科学院沈阳自动化研究所
授予地点	中国科学院沈阳自动化研究所
导师	赵怀慈
关键词	红外热波序列图像 缺陷 奇异值分解 压缩传感 自动识别
其他题名	The Research of Defect Recognition Based on Technology of Infrared Heat Waves
学位专业	模式识别与智能系统
中文摘要	热障涂层是航空发动机叶片隔热材料的重要组成部分。由于传统的检测方法不能满足对检测的精度和安全性要求，本文采用红外热波无损检测技术对热障涂层缺陷进行检测。针对采用红外热波无损检测技术建立热传导模型来计算材料热物理特征分布的方法难以实现定量分析的问题，本文采用从图像处理与目标识别的角度进行缺陷的分析，并通过探索缺陷的分类识别算法实现对被检测物内部缺陷的快速分析。本文通过对缺陷红外序列图像的研究，从以下三个方面进行了研究与实现： (1) 基于奇异值分解的缺陷特征提取方法。该特征提取方法采用对缺陷的红外序列图像进行相空间变换构造缺陷矩阵，通过对重构的缺陷矩阵进行奇异值分解把矩阵投影到时间基与空间基向量。由于奇异值分解后的空间与时间基向量包含了缺陷静态空间与动态热量变化的特征信息，因此，在提取缺陷矩阵的代数特征基础上,通过提取缺陷空间与时间基向量的特征信息，进一步丰富缺陷的特征表征。RBF神经网络作为分类器，实现对热障涂层缺陷样本的分类识别，分类结果验证了缺陷的特征提取算法的有效性。在验证分析中，本文采用 (2)基于压缩传感的缺陷识别方法。压缩传感理论是一种全新的数据采集技术，压缩传感采用非自适应线性投影来保持信号的原始结构, 能通过数值最优化问题准确重构原始信号。基于压缩传感的缺陷识别是依据压缩传感的思想，采用提取缺陷的全部信息，通过变换缺陷序列图像建立缺陷特征矩阵，对经过测量矩阵投影后测量值进行分析实现缺陷的类别判断。本文采用基于压缩传感的缺陷识别算法在热障涂层的分类识别中取得了很好的分类结果，通过实验验证了本算法的有效性。 (3)本文根据项目需求，完成了热障涂层内部缺陷识别系统的设计与实现。该系统由缺陷的定位与分割、缺陷样本库生成、缺陷分类识别以及缺陷评估与分析等模块组成，实现对缺陷的快速识别功能，具有较大的工程应用价值
索取号	TP391.4/Z36/2011
英文摘要	Thermal barrier coating is an important part of insulation materials which is aviation engine blades. Because the traditional test methods cannot meet the precision and safety defects requirements for thermal barrier coating, we use the infrared thermal wave nondestructive testing technology to detect thermal barrier coating defects.  Through the research of the defect by image sequences through infrared hot wave, for the difficult of calculate the distribution of material thermal physical characteristics to realize the method of quantitative analysis by establish heat transfer model, the paper puts forward through image processing and target recognition to analysis the perspective of defect analysis, and by exploring a classification algorithm defects to analysis the internal defects quickly. Based on the study of infrared image sequences defects, and puts forward the following two methods: (1) the algorithm of defect feature extraction by singular value decomposition. Through the study of typical defects characteristics, we put forward the method of singular value decomposition to get the feature extraction. The feature extraction method is adopted to phase space transform to reconstruct the infrared image sequences, and get the vector of space and time by singular value decomposition, which contains the information of defect static space and dynamic heat change characteristics. In validating analysis, this paper adopts the RBF neural network classifier to realize the classification of flaw feature extraction algorithm by the flaw stylebook results of thermal barrier coating  (2) the identification algorithms based on compression sensor. Compression sensing theory is a revolutionary technology of data collection. compression sensing using an adaptive linear projection to keep the original structure which can accurate reconstruction original signal through numerical optimization problem. The method of defect recognition based on compression sensor can extract defect of the information by transform defect feature matrix of image sequences, and realize judgment defect category by establishing matrix. This method in the analysis of classification of thermal barrier coating made very good classification and the result shows the algorithm is effective through experiment. (3) According to the project needs, the paper complete and design the internal defect recognition system for thermal barrier coating .The system of defects contains the module of positioning and segmentation, flaw stylebook library generation, defect classification and defect evaluation. This system realize the function of fast identification which have larger value for engineering application.
语种	中文
产权排序	1
公开日期	2012-07-27
分类号	TP391.4
内容类型	学位论文
源URL	[http://ir.sia.ac.cn/handle/173321/9232]
专题	沈阳自动化研究所_光电信息技术研究室
推荐引用方式 GB/T 7714	张志强. 基于红外热波检测的缺陷识别方法研究[D]. 中国科学院沈阳自动化研究所. 中国科学院沈阳自动化研究所. 2011.

个性服务

查看访问统计

相关权益政策

暂无数据

收藏/分享

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。