【作者】 张宏普；

【导师】 贺西平；

【作者基本信息】 陕西师范大学 ， 声学， 2010， 硕士

【摘要】 目前,超声无损检测技术已应用于很多领域,例如医学检测、海底探测、材料探伤、超声测量等。然而由于超声在传播时的固有特性,使得超声无损检测技术在实际应用时遇到一些干扰因素,比如在医学检测领域,超声在人体内传播时遇到不同组织界面会发生多重回波,除首次回波外其他的回波会对生物组织的结构成像造成阴影或增强,给检测人员的判断带来干扰。因此论文提出了一种算法,用于识别第一次回波和多重回波。这项工作对提高结果的准确性和精度、对超声无损检测技术的进一步推广和应用具有一定意义。1、超声脉冲波在检测多层介质时,反射回波会包含不同界面的多重回波,不利于对回波特征值的提取。因此论文提出了一种用于识别多层介质的多重回波的算法,以区别回波中首次回波和多重回波,并能够找出该多重回波的反射源。该算法基于多层介质的不同介质层对超声波表现出不同的反射率、不同的回波衰减、不同的延时等物理特性,得出穿越相同介质的超声回波比穿越不同介质的超声回波在频谱上有更大的相似性的结论,各回波频谱间的相似程度采用能量谱最小二乘(LMS)偏差表示,然后结合各个回波TOF的时间逻辑关系完成对首次回波和多重回波的识别。2、采用实验仿真和真实实验的方法对文中提出的算法进行了验证。仿真结果表明,10组80个回波的正确识别率达到98.75%。在真实实验中,两层介质的回波正确识别率100%。仿真中使用的回波采用基本信号在频域与指数衰减项相乘的办法得到,基本信号取自实际实验的铁块超声脉冲回波(高信噪比)。3、使用该算法对信噪比较低的多重回波进行了仿真实验。结果显示,回波正确识别率降低。根据小波变换的原理,信号做离散小波变换后,相当于信号通过了高通和低通滤波器,小波系数体现高频成分,表现信号的细节,尺度系数体现低频成分,表现信号基本特征。因此,论文提出用小波变换的尺度系数代替原信号分析,结果回波识别率有效提高。总之,超声脉冲回波的首次回波和多重回波识别算法,可以基于回波间的能量谱最小二乘偏差和各个回波TOF的时间逻辑关系进行判断。对于低信噪比的回波可以采用其小波变换后的尺度系数代替原信号使用本算法来识别。更多还原

【Abstract】 At present, ultrasonic nondestructive testing technology has been widely applied to many areas, such as medical testing、survey the sea、machines、materials, etc. However, according to the characteristics of ultrasound propagation, ultrasonic nondestructive testing technology encounters some interference factor in practical application. For example, in medical testing, reverberation echoes are tested except the primary echo when ultrasound propagating in human body. In addition to the primary echo, reverberation echoes make the structure of the organization as a shadow or enhanced for the inspection officer. Therefore, a new algorithm are proposed in this article used to identify and separate the primary echo and reverberation echoes. This study has great sense in improving the accuracy and precision of the results of ultrasound testing and in further developing the application of ultrasonic nondestructive testing technology.1、Echoes involve reverberation echoes when ultrasound pulses are used to test multilayer media. Reverberation echoes are unfavorable to extracting the features of echoes. Therefore, a new algorithm is proposed in this article to identify reverberation echoes and to find out the source of the reverberation echoes. Based on the physical characteristics of ultrasound propagating in multilayered media,such as different reflectivity、different decay and different time delay, using this algorithm we can get the result that echoes through the same media have greater similarity than echoes through multilayered media. And the similarity score of echoes spectrum is denoted by Least mean square deviation of energy spectrum, then identifying primary echo and reverberation echoes with the time logical relationship of different echoes.2、The algorithm proposed in this article is tested by simulation and real experiment. And simulation results show that identification rate reaches 98.75% in ten groups 80 echoes. Echoes used in simulation get from basic signal in frequency domain multiplying exponential decay. And the basic signal extracted from the practical experiments of ultrasound pulse waves in the iron block (high signal to noise ratio).3、With the algorithm proposed in this article, we simulate the reverberation echoes with low signal to noise ratio. The experimental results show that recognition rate of echoes are greatly reduced. According to the principle of wavelet transformation, signal adopted discrete wavelet transformation is equivalent to the signal through high pass filter or low pass filter. Wavelet coefficients indicate the high frequency components, expressing the signal details, and scale coefficients indicate the low frequency components, expressing the basic features of the signals. Therefore, in this article we use the scale coefficients of wavelet transformation to take the place of the basic signal, which can effectively improve the recognition rate of echoes.In short, the algorithm proposed in this article to identify primary echo and reverberation echoes can be judged by least mean square deviation of energy spectrum and the time logical relationship of different echoes. To identify the echoes of low signal to noise ratio, we can apply the algorithm by using scale coefficients of wavelet transformation in place of the basic signal.更多还原