College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China.
The Key Laboratory of Non-Destructive Testing and Monitoring Technology for High-Speed Transport Facilities of the Ministry of Industry and Information Technology, Nanjing 211106, China.
Sensors (Basel). 2022 Sep 16;22(18):7023. doi: 10.3390/s22187023.
A method based on the high-frequency ultrasonic guided waves (UGWs) of a piezoelectric sensor array is proposed to monitor the depth of transverse cracks in rail bottoms. Selecting high-frequency UGWs with a center frequency of 350 kHz can enable the monitoring of cracks with a depth of 3.3 mm. The method of arranging piezoelectric sensor arrays on the upper surface and side of the rail bottom is simulated and analyzed, which allows the comprehensive monitoring of transverse cracks at different depths in the rail bottom. The multi-value domain features of the UGW signals are further extracted, and a back propagation neural network (BPNN) is used to establish the evaluation model of the transverse crack depth for the rail bottom. The optimal evaluation model of multi-path combination is reconstructed with the minimum value of the root mean square error (RMSE) as the evaluation standard. After testing and comparison, it was found that each metric of the reconstructed model is significantly better than each individual path; the RMSE is reduced to 0.3762; the coefficient of determination R reached 0.9932; the number of individual evaluation values with a relative error of less than 10% and 5% accounted for 100% and 87.50% of the total number of evaluations, respectively.
提出了一种基于压电传感器阵列的高频超声导波（UGW）的方法，用于监测轨底横向裂纹的深度。选择中心频率为 350 kHz 的高频 UGW 可实现对 3.3 mm 深度裂纹的监测。模拟和分析了在轨底上表面和侧面布置压电传感器阵列的方法，可实现对轨底不同深度横向裂纹的全面监测。进一步提取 UGW 信号的多值域特征，并使用反向传播神经网络（BPNN）建立轨底横向裂纹深度的评估模型。以均方根误差（RMSE）的最小值为评价标准，重建了多路径组合的最优评价模型。经过测试和比较，发现重建模型的每个指标都明显优于各个路径；RMSE 降低到 0.3762；确定系数 R 达到 0.9932；相对误差小于 10%和 5%的个体评估值的数量分别占总评估数量的 100%和 87.50%。
