多重超声信号处理技术及应用研究

发布时间：2018-04-10 15:05

本文选题：超声信号 + EMD　；参考：《南京邮电大学》2017年硕士论文

【摘要】：超声检测技术凭借穿透力强,灵敏度高,设备轻便等优点被广泛应用于现代工业各个领域中。在金属构件的超声探伤过程中,我们主要是通过分析接收到的多重超声信号来对缺陷进行定位、定量、定性分析,然而目标信号往往都夹杂在噪声等一些干扰回波中,影响了我们对有用信号的准确判断,所以要想得到清晰有效的目标信号,必需对接收到的多重超声波进行加工和预处理,因此研究多重超声信号的处理分析技术,具有非常深远的意义。多重超声信号处理技术研究主要集中于对采集到的信号进行噪声处理和目标信号特征提取这两个方面。由于多重超声信号属于非稳态的时变脉冲信号,分析研究这类信号时,需要获取其瞬时频率信息,因此本课题从适合非线性非稳态信号分析和处理的小波变换方法和经验模态分解方法(Empirical Mode Decomposition,EMD)入手,在研究上述两种时频分析方法的根本上,重点对这两种方法应用在多重超信号处理中进行分析研究,并进行了比较。实验仿真结果证明了EMD方法根据信号本身不同尺度的波动或趋向一层层解析出来,对其进行平稳化处理,特别适合非线性非稳态信号的分解处理。在上述研究的基础上,本文提出了基于EMD的多重超声回波阈值去噪新方法,并且利用仿真实验表明采用新方法去噪后的信号,信噪比明显提高,均方误差减小,平滑度提高,去噪效果好于传统的去噪方法。最后提出基于能量算子的多重超声信号时间识别方法,利用能量算子提取降噪后信号包络,最后对包络进行参数估计,实验证明,信号的降噪预处理使得该算法适用于低信噪比的状况下,而且基于包络的参数估计减少了计算量,精度也得到了提高,具有良好的工程应用性。
[Abstract]:Ultrasonic testing technology has been widely used in various fields of modern industry by virtue of its strong penetration, high sensitivity and portable equipment.In the process of ultrasonic flaw detection of metal components, we mainly analyze the received multiple ultrasonic signals to locate, quantify and qualitatively analyze the defects. However, the target signals are often mixed in some interference echo such as noise.In order to get a clear and effective target signal, we must process and preprocess the received multi-ultrasound signal, so we study the processing and analysis technology of multi-ultrasound signal.Has very profound significance.The research of multiplex ultrasonic signal processing mainly focuses on the noise processing of the collected signal and the feature extraction of the target signal.Because the multiplex ultrasonic signal belongs to the non-steady time-varying pulse signal, it is necessary to obtain the instantaneous frequency information when analyzing and studying this kind of signal.Therefore, this paper starts with wavelet transform method and empirical mode decomposition method, which is suitable for nonlinear unsteady signal analysis and processing, and studies the fundamental of these two time-frequency analysis methods.The application of these two methods in multiplex supersonic signal processing is analyzed and compared.The simulation results show that the EMD method is suitable for the decomposition of nonlinear unsteady signals.Based on the above research, a new method of multi-echo threshold de-noising based on EMD is proposed. The simulation results show that the signal-to-noise ratio (SNR), mean square error (MSE) and smoothness of the signal denoised by the new method are obviously improved, the mean square error is reduced and the smoothness is improved.The denoising effect is better than the traditional denoising method.Finally, an energy operator based time recognition method for multiple ultrasonic signals is proposed. The envelope of the signal is extracted by energy operator, and the parameters of the envelope are estimated, which is proved by experiments.The pre-processing of signal denoising makes the algorithm suitable for low SNR, and the parameter estimation based on envelope reduces the calculation cost and improves the precision. It has good engineering application.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TB559;TN911.7

【参考文献】