Spectrum Analyses and Extracting Components of Ultrasonic Echo Signals Qiu-ze YE, Xi-zhong SHEN and Wei-wei CAO

Transcription

1 2016 International Conference on Advanced Manufacture Technology and Industrial Application (AMTIA 2016) ISBN: Spectrum Analyses and Extracting Components of Ultrasonic Echo Signals Qiu-ze YE, Xi-zhong SHEN and Wei-wei CAO School of Electrical and Electronic Engineering, Shanghai institute of Technology, Shanghai, China Keywords: Ultrasonic testing, Time-frequency analysis, Characteristic frequency, Digital filter. Abstract. On the basis of an analysis on ultrasonic signals collected by detecting type 6061 aluminum block and 60kg/m rail, a signal processing method is applied to separate echo signals from ultrasonic signal. Ultrasonic signals are extracted and analyzed based on the parameters such as sampling frequency, sampling time, band-pass width and so on, in the experiments. Then, on the time domain, the frequency analysis is carried out in the target frequency range from ultrasonic signal and the characteristic frequencies are obtained. Finally, the digital filter is designed to reduce noise in order to contrast and analysis. The experimental results indicate that noise signals are effectively suppressed in the ultrasonic signals and echo signals are obtained clearly. Introduction Ultrasonic target detection is how to recognize ultrasonic echo signal, including lots of information of defects and material properties. The expected information are mostly hidden in the achieved ultrasonic echo signal and echo signal is mixed with various noise in the process of ultrasonic detection, which is difficult to recognize and estimate. With the development of digitization and Intelligence, analyzing and processing ultrasonic signal has become an essential foundation in ultrasonic detecting system. It can improve the SNR of signal to get more information of echo signal and can be used in automatic judgment, data analysis, etc. At present, main noise removing methods used in ultrasonic signal are Split spectrum analysis [1], wavelet transform [2], adaptive filter [3], Wiener filter [4] and so on. Wavelet transform is applied in denoising separation of ultrasonic signal by Xiang Mu et al.[5] and Wei Dong et al. [6] separate ultrasonic signal with K-SVD algorithm and OMP algorithm. However, how to separate the noise of the ultrasonic signal by the obtained features needs more research. This paper use ultrasonic testing signals as targets in type 6061 aluminum block and 60kg/m rail, then separate noise components in signals and intercept the target s frequency band to spectrum analysis on the time domain. Finally, the digital filter is designed to reduce noise in order to contrast and analysis and echo signals are obtained clearly. Noise Components Analysis in Ultrasonic Testing Signals The noise in ultrasonic testing can be divided into acoustic noise and non-acoustic noise [7]. non-acoustic noise mainly includes electronic circuit noise, ringing noise and pulse noise, the coupling noise, environmental noise and so on. This section analysis ultrasonic signals in type 6061 aluminum block and 60kg/m rail. Ultrasonic signals are extracted and analyzed based on the parameters such as sampling frequency, sampling time, band-pass width and so on, in the experiments. According to analysis experiments, ultrasonic testing signal includes transmit signal, echo signal and noise signal. Thus, ultrasonic testing signal (S) can be described by the function: S S1 S2 N (1) where S 1 is the transmit signal, S 2 is the echo signal, N is the noise signal. The first experiment use a ultrasonic probe with the frequency of 5MHz and the dispensed interval time is 1 ms, which get two signals, the first one without testing (refer with: Figure 1) and the second one with testing type 6061 aluminum block (refer with: Figure 2).

2 transmit signal non-acoustic noise signal transmit signal noise signal echo signal Figure 1. Ultrasonic signal without testing. Figure 2. Ultrasonic time-domain signal with testing type 6061 aluminum block. According to Figure 1 and Figure 2, ultrasonic testing signal is mixed with noise signal, which makes extracting target frequency range to be difficult, in spite of getting some characteristics included roughly time point of echo signal and roughly signal envelope. Spectrum Characteristics In the previous section, ultrasonic time-domain signal is obtained in the first experiment and it is essential to get the ultrasonic frequency-domain signal (refer with: Figure 3). It can be founded that the characteristic frequency is almost 5MHz. Although the shape of the echo envelope is obtained, it is not clear. Figure 3. Ultrasonic frequency-domain signal with testing type 6061 aluminum block. P1 P2 Figure 4. Schematic diagram with testing 60kg/m rail. The 60kg/m rail is used in the second experiment, which tests two different locations on the upper surface with the 5MHz ultrasonic probe (refer with Figure 4). The first location, denoted as P1, above the hole on the rail and the second location, denoted as P2, above the point between the two neighboring holes on the rail. Therefore, the corresponding results of the time-frequency

3 characteristics are obtained with the two locations in the second experiment (refer with: Figure 5, Figure 6, Figure 7, Figure 8). Figure 5. Ultrasonic time-domain signal at P1. Figure 6. Ultrasonic frequency-domain signal at P1. Figure 7. Ultrasonic time-domain signal at P2. Figure 8. Ultrasonic frequency-domain signal at P2. In the experiment, the sampling time is 0.5μs and the sampling frequency is 2GHz. Although the noise signal is always existed, some conclusions are drawn from the results of the experiment. The characteristic frequency range is between 0.1MHz and 5MHz in rough, which coincides with the ultrasonic probe with the center frequency of 5MHz. However, due to the different locations, there are many differences about the spectrum, the time of the echo signals. In addition, there are the rest conclusions: in the range between 0μs and 2μs, the first echo signal is obtained from the lower surface of rail-head; in the range between 7μs and 8μs, the second echo signal is obtained from the lower surface of the rail. Extracting Components According to the two experiments, the noise signal is always existed and there are some results in rough. Especially, the noise signal impact on the echo signals in the second experiment, that we just identify the first and the second echo signals roughly. In order to avoid the interference of the noise signal, this paper design a 6-step band-pass filter, which the cut-off frequencies are 0.09MHz, 9.91MHz and the center frequency is 5MHz. This designed scheme is up to the mustard on the basis of multiple tests. By applying the above method to the two sets of experiments, the following results can be obtained (refer with: Figure 9, Figure 10, Figure 11).

4 Figure 9. Ultrasonic signal with testing type 6061 aluminum block. Figure 10. Ultrasonic signal at P1 with testing 60kg/m rail. Figure 11. Ultrasonic signal at P2 with testing 60kg/m rail. Finally, some conclusions are drawn from the analysis of the results. In the first experiment, transmit signal and echo signal have become clearly on time domain and noises, especially high frequency noise, have been filtered out in abundance on frequency domain. In the second experiment, it has uncovered similar results, characteristic frequency bands are preserved. Summary This paper takes type 6061 aluminum block and 60kg/m rail as testing targets, three sets of

5 ultrasonic signals are obtained, it can be got the characteristic frequency band after analyzing the noise components, then design a suitable filter scheme. The noise components are effectively suppressed, and a clear ultrasonic signal is obtained. Acknowledgement This research was financially supported by Shanghai Science and Technology Commission of Shanghai Municipality, No. 15ZR References [1] Gustafsson M G, Stepinski T. Studies of split spectrum processing optimal detection and maximum likelihood amplitude estimation using a simple clutter model[j]. Ultrasonics (1): [2] Drai R, Khelil M, Benchaala A. Time frequency and wavelet transform applied to selected problems in ultrasonics nde[j]. NDT and E International (8):567. [3] Kim J, Udpa L, Udpa S. Multistage adaptive noise cancellation for ultrasonic NDE[J]. NDT and E International (5): 319. [4] Izquierdo M A G, Hemandez M G, Graullera O, et al. Time frequency wiener filtering for structural noise reduction[j]. Ultrasonics (1-8):259. [5] Mu Xiang, Zheng Bin,Guo Hualing.Features detection of laser ultrasonic flaw detection signal based on wavelet[j]. ELECTRONICS WORLD,2015,22: [6] Wei Dong, Zhou Jianpeng. Application of K-SVD and OMP algorithm on ultrasonic signal denoising[j]. Journal of Applied Acoustics, 2016,02: [7] J. Chen, Y Shi, S Shi. Noise Analysis of Digital Ultrasonic Nondestructive Evaluation System.Inter- national Journal of Pressure Vessels and Piping, 1999, 76: