钢层下多层橡胶粘接界面脱粘的超声检测信号处理技术研究
文献类型：学位论文
| 作者 | 张建生 |
| 学位类别 | 博士 |
| 答辩日期 | 2000 |
| 授予单位 | 中国科学院声学研究所 |
| 授予地点 | 中国科学院声学研究所 |
| 关键词 | 钢-橡胶 分层结构 脱粘 超声 无损检测 信号处理 |
| 中文摘要 | 针对钢-橡胶分层粘接结构中的脱粘超声检测课题，本文从超声检测信号的处理角度，利用现代信号处理技术探索研究脱粘缺陷的检测和评价。这是中国科学院声学所李明轩研究员领导课题组承担的国家九五重大军工民口科技攻关项目中的有关应用基础研究课题之一，具有重要的应用价值和理论意义。“钢-橡胶分层粘接结构”在现代航空航天等高新技术工业中的重要应用，迫切需要现代的超声无损检测和评价技术来为其提供有力的支撑和保障，然而粘接缺陷的超声检测和评价还面临许多很大的困难和有力的挑战。钢层下多层橡胶粘接结构中脱粘的超声无损检测就是其中当今亟待解决的一个重大课题。而现代信号处理技术正是有望攻克这一堡垒的有力武器之一。现代信号处理技术不仅通过信号的分析和处理，认识和提取波的不同特征，而且通过压制噪声与干扰，最大可能地突出有效信号并提取可利用信息，并能对缺陷进行空间定位和成像。信号处理与信号分析现在用于粘接结构的超声检测，它已经不应仅仅是一种辅助技术，它应该是一种基本技术，并应向智能化发展。这是粘接结构超声检测进入自动化、智能化的重要要求。同时，信号处理研究有利于进一步认识超声检测中超声波在层状介质中的传播特性。现代信号处理在粘接结构超声检测的应用还有很大的局限性，还很不深入，亟待进一步研究。对于钢-橡胶多层粘接结构的超声检测信号，其处理方法的研究尚不多见。本文的研究内容、主要结论和创新之处分为以下五个方面：第一方面研究钢-橡胶分层粘接结构超声检测信号的理论建模和仿真。该部分在给出超声检测信号的卷积模型描述基础上，详细分析了超声波在分层介质中的传播特点，由此阐述了“层滤波器”观点，并用所提出的层滤波器理论，得到了钢-橡胶分层粘接结构的超声检测回波信号解析表达式。该层滤波器方法与求解偏微分方程法、传输矩阵法、传输线法等理论比较而言，具有形式简洁、过程明了、物理意义清晰等优点，适宜于信号处理分析和研究。第二方面研究脱粘界面的超声定征。由于超声检测探头发射的脉冲信号具有一定信号长度，各界面的多次反射回波信号叠加在一起，造成在时域中不容易或者不可能识别各界面回波的位置等特征。该部分利用解卷积信号处理技术研究了检测子波影响的解卷积消除和粘接结构系统响应的提取。首先利用子波反滤波技术，有效地对检测子波进行了压缩，提高了各界面特征多次反射的分辨率，实现对脱粘界面的鉴别；其次，利用同态解卷积技术在消除检测子波的影响后，提取出粘接结构的系统响应，由此实现了对脱粘界面的定征。第三方面研究零界面大幅度信号、一界面强多次反射回波干扰信号的消除和深层界面微弱反射回波信号的提取。由于零界面大幅度信号和一界面强多次反射回波的能量相对大得多，使得它们淹没了来自深层界面的弱反射回波信号，造成长期以来在时域和频域对该结构深层界面(二、三、四界面等)脱粘检测的困难。该部分分别利用同态滤波、指数加权匹配相消、相关加权匹配相消和维纳滤波匹配相消等技术实现了对零界面大幅度信号、一界面强多次反射回波干扰信号的消除，并且提取出深层界面的微弱反射回波特征信号。通过深层界面反射回波信号的位置和幅度等特征，达到对脱粘界面的识别和确定。第四方面研究不同界面粘接状况时的检测信号模式识别。由于“频率窗”检测技术的采用，选取了频率较低的超声探头，使得各界面多次反射信号的分辨率很小，以至于一般的信号处理技术不能取得有效的处理结果。该部分从界面脱粘的检测信号模式识别角度，在利用傅立叶变换(FFT)和离散余弦变换(DCT)提取的模式特征矢量基础上，通过人工神经BP网络实现了对各界面脱粘时检测信号模式的正确识别和分类。第五方面研究脱粘界面超声检测信号的小波多分辨率分析与重构定征。检测信号携带着反映粘接结构界面粘接质量的大量信息，包括时域特征信息和频域特征信息。该部分利用小波变换在时频两域的良好局部性能，通过多分辨率时间-尺度幅度分布和时间-尺度相位分布对各脱粘检测信号进行了分析，在此基础上实现了时域信号和频域特征分布的重构。该部分的研究不仅实现了脱粘信号的小波多分辨率分析和定征，而且提取出在时域和频域的特征信号。它对超声检测技术的发展具有促进作用和重要的应用前景。本文所用的信号处理技术都是第一次应用在钢-橡胶分层粘接结构超声检测中，研究工作目前国内外尚未见有报道。 |
| 英文摘要 | Multi-layered adhesive bonded structure between metal and rubber are gaining increasing acceptance in safety critical applications in advanced industries. There is a correspondingly growing need for reliable ultrasonic non-destructive testing (NDT) techniques that can ensure the integrity of a bond-line after manufacture and allow for periodic checks of bond integrity during use in intended applications. Unfortunately, successful ultrasonic pulse-echo NDT techniques for the inspection of adhesive bonds have been difficult to develop. The reason is that serious problem, due to the limited bandwidth and non-linear phase response of commonly used ultrasonic testing equipment, are met in the case of thin layer of steel and rubber. So, the debond detection, in multi-layered steel-rubber adhesive structure, is still one of the most difficult problem in the world. The purpose of this paper is to introduce modern DSP algorithms which enables the efficient detection and characterization of debonds in thin multi-layered composites using band-limited ultrasonic signals acquired by commonly used ultrasonic instrument. In particular, the algorithms exploiting modem signal processing techniques, such as deconvolution, echo cancellation, wavelet transform, pattern recognition and artificial neural networks, are analyzed. Comparison of the simulated and experimental results revealed the possibility of determining the character of the debond occurring in the five-layered steel-rubber adhesive structure. Five innovative research results in this paper are summarized. A simple layer-filter model for an ultrasonic non-destructive testing of 5-layered steel-rubber structure, which contain debond defects, is proposed. According to the model, the analytic expression, which is used to simulate testing ultrasonic signal of that structure, is derived. Deconvolution techniques are used to identify the debond defects. In ultrasonic nondestructive testing (NDT) debond, the reflection sequence convolved with the impulse response of the transducer results in masking closely spaced reflections. Deconvolution of these signals may reveal the reflection sequence and thus make the interpretation easier. In this paper, deconvolution techniques of wavelet inverse filtering and Homomorphic filtering are used to undo the effect of the convolution and extract the defect impulse response which is essential for debond defect identification. Synthesized ultrasonic signals and real signals obtained from artificial debond defects are used to show that the proposed technique is superior to conventional deconvolution techniques commonly used in NDT. Algorithms based on the echo cancellation techniques, such as exponential, across-correlation and Winner weighting, are proposed. Debond defects, occurring in rubber layer, cannot be detected effectively with normal incident ultrasonic testing methods used in industry, because the signal reflected from such a defect is so weak that may be overshadowed by the reverberate signal in steel layer. Through reverberation suppression in steel-layer, both simulated and experimental data shows encouraging debond detection capabilities by the use of the algorithms. Debonds are identified and classified by Modern Signal Processor. There is a need for effective inspection for both quality assurance (QA) and the assessment of condition in service. The acquired signals from 5-layered steel-rubber adhesive bonded structure are analyzed by modern signal processing techniques. The bond quality featurea (BQF) are extracted by DFT and DCT. The BQF data is used as the input to artificial neural networks (ANNs), which is trained to associate features in the input data with principal bond quality. The optimal network structure is examined, and the technique gives reliable classification of the features not included in network training. The interconnected weights of simplified networks provide evidence of the features in ultrasonic signals that underlie the successful preparation of the method. Ultrasonic signals of adhesive multi-layered structure are multi-resolution analyzed and reconstructed by wavelet transform. In this paper, the acquired ultrasonic signal of multi-layered steel-rubber bonded structure, which recorded by our efficient testing system, was studied by wavelet transform. The characterizations of debond at interfaces of steel-rubber or rubber-rubber were identified, extracted and synthesized. The results show the model and algorithm are promising for NDT &NDE in industry application. This elementary study of the application of DSP techniques to the inspection of the debond in multi-layered steel-rubber adhesive structure provides a solution to a common problem in NDE that is particularly significant in adhesive quality evaluation. |
| 语种 | 中文 |
| 公开日期 | 2011-05-07 |
| 页码 | 143 |
| 源URL | [http://159.226.59.140/handle/311008/674]  |
| 专题 | 声学研究所_声学所博硕士学位论文_1981-2009博硕士学位论文 |
| 推荐引用方式 GB/T 7714 | 张建生. 钢层下多层橡胶粘接界面脱粘的超声检测信号处理技术研究[D]. 中国科学院声学研究所. 中国科学院声学研究所. 2000. |
入库方式： OAI收割
来源：声学研究所
浏览0
下载0
收藏0
其他版本
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。