Application Cases

Experiment Name: Nonlinear Ultrasonic Detection of Rail Joint Defects

Experimental Principle:
This experiment detects bolt hole cracks by leveraging the sensitivity of nonlinear systems to initial conditions. Even minor changes in materials or structures can be identified using nonlinear ultrasonic indicators, significantly improving the detection accuracy for micro-defects. In this experiment, PZT material is used as both the excitation and receiving sensor. The excitation signal is amplified by a power amplifier to drive the piezoelectric patch, generating ultrasonic waves in the rail. Independent piezoelectric patches are attached near the excitation patch to act as receiving sensors. Specifically, PZT-2 serves as the excitation sensor, while PZT-1 and PZT-3 act as receiving sensors, capturing resonant signals in the horizontal and vertical directions, respectively. This setup allows for better observation of differences in response signals across different planes.

Testing Equipment:
Signal generator, ATA-8202 RF power amplifier, PZT sensors, test specimen, etc.

Figure 1: Schematic Diagram of the Experimental System

Figure 2: Rail Specimen

Experimental Procedure:
First, the time-domain response on the oscilloscope must be measured when no excitation signal is applied. This is because noise is unavoidable in experiments, and it is necessary to estimate its magnitude and impact in advance. A sinusoidal modulated signal is generated and input into the signal amplifier. The length of the input signal is adjusted to ensure it does not overlap with the response signal, thereby maximizing the zero-input duration of the input signal. Here, the modulated signal length is set to 2,000 samples, and the total input signal length is set to 10,000 samples. It is important to completely delete any residual signals from previous inputs after changing the input signal; otherwise, the new signal may become a mixture or overlap of the two. The rated voltage of the signal generator is 10V, but the piezoelectric material requires approximately 200V to generate voltage. Therefore, a signal amplifier is needed to amplify the voltage. After amplification, the signal is input to the piezoelectric patch. Receiving piezoelectric patches are placed near the excitation patch to capture the response signals, which are then connected to an oscilloscope. The response signals are extracted from the oscilloscope and processed to obtain time-domain and frequency-domain diagrams. Finally, damage-sensitive nonlinear parameters are extracted from these results.

Figure 3: Time-Domain Signal Comparison for Crack-Free and Cracked Specimens

Figure 4: Frequency-Domain Signal Comparison for Crack-Free and Cracked Specimens

This experiment uses an excitation signal with 10 cycles and a frequency of 0.5 MHz. Figure 3(a) shows the time-domain signal received from an intact rail, while Figure 3(b) shows the time-domain signal after introducing a 1 mm crack near the bolt hole. Although differences in the signals are observable, it is difficult to directly extract defect information from the time-domain signals alone. Figure 4 shows the frequency spectrum obtained after performing Fourier transform and logarithmic scaling on the response signals. It can be seen that the defect causes waveform distortion, making the second harmonic components in the output signal more pronounced.

Experimental Results:
In the smoothing analysis of experimental results, the number of points used is a critical parameter. If too few points are used, the smoothing effect is insufficient, but if too many points are used, the amplitude-frequency curve may lose local features. Figure 5 compares the original amplitude-frequency curve of a rail specimen with a 1 mm crack near the bolt hole and the results after smoothing with 5, 15, 25, and 50 points. As shown, the 5-point smoothing result shows no significant difference from the original curve. As the number of smoothing points increases, the noise influence decreases. The 50-point smoothing achieves good noise reduction while preserving the local features of the second harmonic. Therefore, all conditions are smoothed using 50 points before extracting nonlinear parameters.

Figure 5: Frequency Spectrum Results After Smoothing with Different Numbers of Points

Figure 6: Relationship Between Nonlinear Parameter β and Damage Degree

Here, a and b represent the logarithmic results of the fundamental and second harmonic amplitude peaks shown in Figure 5. The β values for all conditions are calculated and plotted in Figure 6. It can be observed that the nonlinear parameter β exhibits an approximately linear relationship with the crack length near the bolt hole. This indicates that the defined nonlinear parameter β can serve as an indicator for characterizing the damage degree of rail bolt hole cracks, providing a basis for crack size evaluation.

The experimental materials in this article were compiled and released by Xi’an Aigtek Electronics. For more experimental solutions, please continue to follow the Aigtek official website.