Switch rails are weak but essential components of high-speed railway systems that have urgent nondestructive testing requirements owing to aging and the associated potential for fatigue damage accumulation. This study presents a multi-feature integration and automatic classification algorithm for switch rail damage using guided wave monitoring signals. A combination of piezoelectric transducers and magnetostrictive patch transducers is adopted to improve the monitoring performance and meet actual monitoring requirements. Furthermore, multiple features extracted from various signal processing domains—such as the time domain, power spectrum domain, and time–frequency domain—are proposed and defined according to the structure and characteristics of the switch rail and guided wave to represent the complex nature of the damage. A damage index is defined to eliminate the influence of various environmental and operational conditions, signal power, and other factors. In addition, a feature selection method based on binary particle swarm optimization with a new fitness function is proposed to select the most damage-sensitive features and eliminate irrelevant and redundant features to improve the classification performance. Moreover, considering that the results are easily influenced by experts’ subjective judgment and experience, the least-squares support-vector machine is used to construct automatic classification models to reduce the probability of artificial incorrect diagnosis and improve the generalization ability to unknown environments. Finally, three types of experiments on the foot of a switch rail are presented to evaluate the proposed method. The results indicate that the proposed method is capable of identifying damage in challenging cases and is superior to conventional methods.

Data-driven approaches in structural health monitoring have received increasing attention, especially advances in deep learning-based methods, which have further driven the development of data-driven damage detection. Due to the limited availability of guided wave samples and the imbalance between data classes, this study proposes a deep convolutional neural network-based transfer learning (DCTL) approach for the structure monitoring of switch rails using guided wave monitoring signals. A pretrained model based on Inception-ResNet-V2 was adopted and fine-tuned. Different methods for converting 1D signals into 2D images were investigated to find the optimal approach that meets practical monitoring requirements. Affine transformations were used for data augmentation to improve generalization ability and to avoid the overfitting of the training model. Two types of guided wave monitoring experiments on the foot and web of switch rails were conducted to evaluate the proposed method against different conventional methods in the field of switch rails. In addition, the DCTL method was investigated, with and without pretrained weights, along with different frozen layers. The classification results show that the proposed method can identify damage in challenging situations and outperforms conventional methods.

Temperature changes are a major challenge in outdoor guided wave structural health monitoring of rails. Temperature variations greatly impact the waveform of guided wave signals, making it challenging to diagnose and characterize defects. Traditional temperature compensation methods, such as signal stretch and scale transform, are restricted to use in regular structures, such as plates and pipes. To solve the temperature compensation problem in long rails with serious mode conversion and complex structure echo, we propose a temperature compensation and defect monitoring method, namely, sliding window dynamic time-series warping (SWDTW), which overcomes the challenges of mass computation and overcompensation of dynamic time-series warping (DTW). The basic idea of SWDTW is to utilize sliding windows to accelerate the computation and identify defects from subsequence scales. Then, an index, window subsequence Teager energy (WSTE), is used to indicate the local abnormality of guided wave signals, and a sliding window net (SWnet) is devised to monitor the occurrence of defects automatically. Outdoor monitoring of turnout rails showed that the proposed method can effectively reduce the temperature noise and recognize an artificial defect with 1.16% and 0.36% cross-sectional change rates (CSCRs) on the switch and stock rails, respectively, at different temperatures; moreover, the defect signals processed by SWDTW showed better defect identification performance than those processed by scale transform and DTW.

Switch rails are weak but essential components of high-speed rail (HSR) systems. In the condition-based maintenance of HSR, ultrasonic guided wave (UGW) on-line monitoring technology is widely used in judging real-time operating conditions; however, it always generates large amounts of data. Too much data bring significant challenges, such as too many unnecessary costs of energy, storage, and network bandwidth in the structural health monitoring of switch rails, making it challenging to realize embedded sensor networks with high durability and low power consumption. Furthermore, the structural damage occurs relatively less over the long-term, indicating that these measurements are inherently sparse. The sparseness of structural damage makes them attractive for the compressed sensing technique. This study proposes a novel data compression and reconstruction method to meet the challenges and reduce the amount of data transmitted by sensor networks and maintain their accuracies simultaneously. First, a lightweight data dictionary is constructed to perform a sparse decomposition of UGW signals according to the characteristics of UGW propagation. Second, an effective sampling method based on a sparse random matrix is designed for sub-Nyquist sampling and compressing UGW signals. Third, a novel block adaptive matching pursuing algorithm is proposed to reconstruct UGW signals from compressed data. Finally, numerical signals, finite element simulation, and several actual monitoring experiments on the foot of a switch rail are conducted to verify the effectiveness and accuracy of the proposed method. The influence of different compression ratios and block sizes on the reconstruction performance of guided wave signals is investigated. The results indicate that the proposed method can sample UGW signals with much lower requirements than the Nyquist sampling theorem and is superior to other novel algorithms.