薄板复合材料粘接缺陷超声检测量化识别研究

项目名称： 薄板复合材料粘接缺陷超声检测量化识别研究

项目编号： No.60862004

项目类型： 地区科学基金项目

立项/批准年度： 2009

项目学科： 建筑科学

项目作者： 张泽

作者单位： 内蒙古大学

项目金额： 28万元

中文摘要： 针对目前超声检测领域无法对金属/非金属薄板复合材料界面缺陷进行量化识别、检测的难题，在深入研究超声检测回波信号的时域能量衰减性、频谱偏移特性以及全面的时-频域特性的基础上，提取能反映粘接缺陷量化程度的信息特征，联合构建分级化（10级）特征向量。以特征向量为输入，用神经网络算法建立模糊识别的隶属度函数，进一步产生标准模糊子集和待识别模糊子集。采用欧氏距离公式，计算待识别模糊子集与各个标准模糊子集的距离及贴近度，按择近原则确定待识别信号的归属，最终实现对粘接界面缺陷的量化识别及量化检测。本研究对于薄板复合材料粘接缺陷的量化识别以及缺陷检测的智能化、自动化具有重要的理论意义和实用价值。

中文关键词： 薄板复合材料；超声检测；特征向量；量化识别；模糊识别

英文摘要： According to the problem of quantification detection and recognition of the metal/nonmetallic compound thin-plate materials cementation contact surface by ultrasonic wave, the information characteristics that reflect quantification degree of the cementation flaws is extracted as well as joint mapping classfication (10 levels) characteristic vector constructed. Those are based on the deep reasearch to the time-domain energy decay characteristics, frequency spectrum displacement characteristic and comprehensive time-frequency characteristic of ultrasonic detecting echo signal. Take the characteristic vector as inputting, while establishing the fuzzy recognition membership function with the neural network algorithm, the standard fuzzy subset and awaiting-recognition fuzzy subset are produced. The distance and closeness degree between the awaiting-recognized fuzzy subset and each standard fuzzy subset are calculated by using the Euclidean distance formula. Assignment of the awaiting-recognized signal is confirmed according to the principle of choosing the nearest, then quantification recognition and detection of the cementation contact surface flaws are realized. This reasearch has great theoretical significance and practical value to the quantification recognition of compound thin-plate materials cementation flaws as well as the intellectualization and automation of flaw detection.

英文关键词： Compound thin -plate materials; Ultrasonic detection; Characteristic vector; Quantification recognition; Fuzzy recognition