Rail Inspection Research Articles

Non-contact detection of railhead defects and their classification by using convolutional neural network

Railhead defects must be detected and classified intelligently in order for railway transportation systems to operate safely. Rail defect identification and categorization can be automated by using machine learning models to process rail image data (acquired using cameras). However, such an automated method has significant drawbacks: it cannot detect subsurface defects, picture data requires a high-end GPU with a long computational time, and machine learning model training can be influenced by image quality, which is dependent on light intensity and shooting altitude. Rayleigh waves are a potential candidate for rail inspection because they can detect both surface and subsurface defects and travel long distances on curved surfaces (like a rail) at high speed. This article looks into the possibility of combining fully non-contact laser ultrasonic technology (LUT) and a deep learning approach for intelligent detection and classification of railhead surface and subsurface defects. The fully non-contact LUT was used to actuate and capture laser-generated Rayleigh wave signals on railhead specimens in order to create a database of A-scan signals from healthy, surface, subsurface, and edge defect railheads. The classification capabilities of a support vector machine (SVM), a fully connected deep neural network (DNN), and a convolutional neural network (CNN) were examined after they were applied to the preprocessed signals without extracting any statistical/signal processing-based characteristics. The comparative analysis demonstrates that CNN is robust in classifying railhead defects. As a result, when combined with CNN, the laser ultrasonic technology may ensure automatic defection and classification of railhead surface and subsurface flaws.

Guidelines for rail reprofiling

The condition of the rails affects the comfort and safety of railway transport. Managing the rail life cycle on modern railway networks has been playing an increasingly important role. Inspection, preventive and corrective maintenance of rails are significant activities in the integral maintenance of the railway superstructure and substructure. The competitiveness of railway transport on the traffic market requires the realization of maintenance activities in short periods of line closure. The paper presents guidelines for rail reprofiling based on the experiences of railway Infrastructure Managers and the recommendations of European technical regulations. The need for grinding new rails after laying the track was pointed out. The advantage of preventive over corrective reprofiling is emphasized. Different types of target profiles of the reprofiled railhead are presented. This paper aims to improve the knowledge of experts in the field of railway infrastructure maintenance.

Dynamic Correction of Sensitivity of Flaw Detection Channels in High-Speed Rail Inspection

Dynamic Correction of Sensitivity of Flaw Detection Channels in High-Speed Rail Inspection

PROBLEMS OF HIGH-SPEED RAIL FLAW DETECTION

The main factors that reduce the quality of non-destructive testing of rails during high-speed scanning are considered. The most significant factors are the quality and volume of obtaining primary information about the condition of the monitored rails. It is shown that, at high-speed (up to 120 km/h) control, the compression of the defect location zone is clearly manifested and the length of the areas with acoustic contact violations increases. Analysis of real flaw diagrams of diagnostic complexes shows that with an increase in the scanning speed, the length of the location zone even from such large reflectors in the rails, such as bolt holes, significantly decreases. The average length of the instability zone of the bottom signal over the welded joints of the rails also increases significantly. The compression of the location zones of the reflectors can be compensated by expanding the aperture of the ultrasonic transducers. Improving the quality of the acoustic contact requires further improvement of the design of the search system and the systems for supplying the contacting liquid to the transducers, depending on the scanning speed. The Magnetic Flux Leakage(MFL) method can effectively detect defects in the rail head up to 20 mm deep at high speeds. Mathematical modeling of the magnetic flux in the controlled rails is performed. This allows us to start creating a new rail magnetization system with an increased interpole distance for high speeds. The introduction of modern methods for processing a significant flow of flaw detection information using neural networks requires the formation of a large sample base of training signals from real defects in different sections of rails. This is a complex task in its own right. For the first time, the issues of checking the operability of flaw detection devices in real control conditions are raised. Testing of ultrasonic equipment at high speeds is proposed to be carried out with the help of special electronic-acoustic simulators of defects. They are installed on different surfaces of the rails on which the diagnostic complexes pass. Operational quality control of ultrasonic rail inspection can be evaluated by statistical analysis of signal parameters from structural elements (bolt holes). The choice between the monitoring performance and the required reliability of detecting rail defects must be made based on the results of real passes of diagnostic complexes at operating scanning speeds.

Analysis and Prediction of Trough Track Superelevation Based on Big Data

The detection of the safety status of urban rail transit is an important part of ensuring the operation of the rail. Using the data detected by the rail to analyze and predict the quality of the rail is very important for the research of rail inspection. Based on the existing research, this paper uses big data to analyze the collected superelevation data, and builds a trough-track superelevation big data prediction model based on the combination of the stochastic oscillation sequence gray model and ALO-Elman network to analyze the historical superelevation data and mine information about superelevation trends. In this paper, the average value of superelevation data of the equal-spaced groove track in a certain interval is collected for verification. The experimental results show that the method can reasonably predict the change trend of the ultra-elevation. The change trend is basically in line with the original data change trend, and the deviation is controlled to be small within range.

Development and application of a multi-sensor integration detection and analysis device for metro gauge and track geometric state

In order to ensure the safe operation of subways, it is necessary to detect the metro gauge and track geometry in time. However, the manual detection and the large rail inspection car is inefficient and expensive. Therefore, we have developed a multi-sensor integration detection and analysis device. Sensors, such as a laser scanner, a gauge sensor, an angle sensor, and a single-chip computer, are integrated into a rail inspection trolley, and the functions of gauge detection, gauge analysis, rail gauge calculation, and superelevation calculation are realized. Through the calibration of an odometer and a sensor, the accuracy of the device can meet the requirements of the standard. The device is applied to Changsha Metro Line 3 to detect and analyze the gauge and track geometry state. Finally, the device realizes the functions of the gauge intrusion alarm, automatic input of graphics and gauge intrusion data, automatic identification of sleepers, and automatic collection of track geometry data.

Анализ параметров донных сигналов в зоне сварных стыков рельсов при высокоскоростном ультразвуковом контроле

The present research is a continuation of works aimed at analyzing the parameters of ultrasonic signals during high-speed inspection of rails. The article analyzes the zones of welded joints of rails in a wide range of inspection speeds. The change in the amplitudes of the bottom signals in the zone of welded joints is analyzed. At certain scanning speeds, many areas of welded joints can be attributed to untested rail sections with significant loss of bottom signals. The research results can be used to improve the designs of search systems for high-speed diagnostic systems, as well as to determine the permissible scanning speeds of the rail track.

Non-Contact Inspection of Railhead via Laser-Generated Rayleigh Waves and an Enhanced Matching Pursuit to Assist Detection of Surface and Subsurface Defects.

Laser ultrasonic technology can provide a non-contact, reliable and efficient inspection of train rails. However, the laser-generated signals measured at the railhead are usually contaminated with a high level of noise and unwanted wave components that complicate the identification of defect echoes in the signal. This study explores the possibility of combining laser ultrasonic technology (LUT) and an enhanced matching pursuit (MP) to achieve a fully non-contact inspection of the rail track. A completely non-contact laser-based inspection system was used to generate and sense Rayleigh waves to detect artificial surface horizontal, surface edge, subsurface horizontal and subsurface vertical defects created at railheads of different dimensions. MP was enhanced by developing two novel dictionaries, which include a finite element method (FEM) simulation dictionary and an experimental dictionary. The enhanced MP was used to analyze the experimentally obtained laser-generated Rayleigh wave signals. The results show that the enhanced MP is highly effective in detecting defects by suppressing noise, and, further, it could also overcome the deficiency in the low repeatability of the laser-generated signals. The comparative analysis of MP with both the FEM simulation and experimental dictionaries shows that the enhanced MP with the FEM simulation dictionary is highly efficient in both noise removal and defect detection from the experimental signals captured by a laser-generated ultrasonic inspection system. The major novelty contributed by this research work is the enhanced MP method with the developments of, first, an FEM simulation dictionary and, second, an experimental dictionary that is especially suited for Rayleigh wave signals. Third, the enhanced MP dictionaries are created to process the Rayleigh wave signals generated by laser excitation and received using a 3D laser scanner. Fourth, we introduce a pioneer application of such laser-generated Rayleigh waves for inspecting surface and subsurface detects occurring in train rails.

Detection of Rail Geometry Irregularities Using Point Cloud Data from Terrestrial Laser Scanner (Study Case: Curve No. 106, Railroad between Cipatat-Tagogapu, Indonesia)

Geometry railroad irregularities can disrupt safety of the train that passes through it and the comfort of train users. In geometric inspection of rail, it requires reliable accuracy from survey data. Terrestrial Laser Scanner (TLS) offers the advantage of obtaining accurate measurements of objects or 3D model. This study aims to examine irregularities of widening gauge and cant rail in curve no. 106, railroad between Station Cipatat – Station Tagogapu based on Minister of Transportation Regulation No. 60 of 2012. The 3D rail model from the point cloud TLS can be used to extract changes of the gauge and height differences with the help of cross- sections. Extraction of gauge and height differences is carried out per 1 meter and given numbering. The results of both gauge and height differences are adjusted to the type of track from the results of the design of the curve, then compared with the value of the standard widening gauge and cant based on the Minister of Transportation Regulation No. 60 of 2012. The result extraction of the rail gauge in curve no. 106, shows the highest irregularities of widening gauge on the gauge number 24 with deviation value +0,017 m. The results of obtaining height differences rail extraction at curve no. 106, shows the highest irregularities of rail cant on the height differences number 15 and 16 with deviation value +0,022 m.

The Effect of Motion-Induced Eddy Currents on Three-Axis MFL Signals for High-Speed Rail Inspection

Nondestructive testing (NDT) methods are widely used in the rail industry to detect rolling contact damage (RCD) defects to ensure railway safety. With the need of monitoring millions of miles of rail tracks more effectively and efficiently that leads to improved rail safety, reliability, and optimized assets management, the research on high-speed and high-accuracy NDT methods that can obtain high signal-to-noise ratio (SNR) and high sensitivity defect signals at higher speeds is of vital importance. Magnetic flux leakage (MFL) is one of the NDT methods used for high-speed rail inspection, which is suitable for detecting surface and subsurface RCD defects on railhead, but the measured MFL signals are distorted severely by the effect of motion-induced eddy currents (MIEC). However, the fundamental understanding of the mechanism of MIEC generation, distribution, and magnitude is lacking, and the physical interaction between MIEC and MFL measurement during high-speed rail inspection has not been well studied. This article investigates the effect of MIEC on three-axis MFL signals through 3-D finite-element method (FEM) simulations. The 3-D distributions of MIEC and 2-D image signals of MFL measurements for high-speed rail MFL inspection are presented and analyzed. The simulations are conducted with an inspection speed range of 0-62.5 mi/h, and a surface hemisphere-type defect and subsurface hole-type defect are both considered. The results show that the motion between the magnetizer and the rail will generate two types of MIEC, which in turn decreases the peak values of three-axis MFL signals from the subsurface defect and affects the MFL signals spatially, especially the base values along rail transverse and vertical directions. The effect of MIEC on the three-axis MFL signals in terms of SNR, sensitivity, and asymmetry is discussed.

Analyzing Ultrasonic Signal Parameters during High-Speed Rail Inspection

Analyzing Ultrasonic Signal Parameters during High-Speed Rail Inspection

Analyzing Risk of Service Failures in Heavy Haul Rail Lines: A Hybrid Approach for Imbalanced Data.

An incident in which a rail defect of size over a threshold value is noticed and the track is taken out of service is known as a service failure. This article aims at building accurate prediction models with binary outcome for risk of service failures on heavy haul rail segments. An analysis of the factors that influence the risk of a service failure is conducted and quantitative models are developed to predict locations where service failures are most likely to occur until the next inspection. To this end, data are collected from a Class I U.S. Railroads for six years from 2011 to 2016. Four prediction models (i.e., logistic regression, decision tree, multilayer perceptron, and gradient boosting classifier) are implemented and their results are compared. To account for the imbalanced classes between the normal operation and service failure, two treatments have been used including undersampling and oversampling. To improve the model performance, the parameters of each method are tuned using random search hyperparameter optimization. Later, bootstrap aggregation (or bagging) is incorporated into each method. The findings of the study show that the prediction performance is the highest when using bagging and oversampling as treatments with gradient boosting method. It was also identified that gross tonnage, presence of geometry defects, ambient temperature, segment length, and rail defect presence are the most important factors for predicting the risk of service failures. The results of this study are useful for railroads to develop effective strategies for rail inspections, preventive maintenance, and capital planning.

Inductive Thermography as Non-Destructive Testing for Railway Rails

Inductive thermography is a non-destructive testing method, whereby the specimen is slightly heated with a short heating pulse (0.1–1 s) and the temperature change on the surface is recorded with an infrared (IR) camera. Eddy current is induced by means of high frequency (HF) magnetic field in the surface ‘skin’ of the specimen. Since surface cracks disturb the eddy current distribution and the heat diffusion, they become visible in the IR images. Head checks and squats are specific types of damage in railway rails related to rolling contact fatigue (RCF). Inductive thermography can be excellently used to detect head checks and squats on rails, and the method is also applicable for characterizing individual cracks as well as crack networks. Several rail pieces with head checks, with artificial electrical discharge-machining (EDM)-cuts and with a squat defect were inspected using inductive thermography. Aiming towards rail inspection of the track, 1 m long rail pieces were inspected in two different ways: first via a ‘stop-and-go’ technique, through which their subsequent images are merged together into a panorama image, and secondly via scanning during a continuous movement of the rail. The advantages and disadvantages of both methods are compared and analyzed. Special image processing tools were developed to automatically fully characterize the rail defects (average crack angle, distance between cracks and average crack length) in the recorded IR images. Additionally, finite element simulations were used to investigate the effect of the measurement setup and of the crack parameters, in order to optimize the experiments.

ОЦЕНКА ПОКАЗАТЕЛЕЙ НАДЕЖНОСТИ ЭЛЕКТРОМЕХАНИЧЕСКОГО ОБОРУДОВАНИЯ СПЕЦИАЛЬНОГО САМОХОДНОГО ПОДВИЖНОГО СОСТАВА

The study highlights the factors affecting the basic reliability indices of special self-propelled rolling stock (SSPRS), rail service cars and rail inspection cars, such as mean time to failure, the probability of failure-free operation, the failure rate, and gamma-percentage resource. Methods for their determination are indicated, and their quantitative assessment is made. It is shown that individual units of electromechanical equipment working in conjunction with an asynchronous motor, clutches and actuators have a superimposed exponential and normal law of probability distribution to failure. The formula to determine the distribution density, which is necessary to adjust the SSPRS reliability is described.

A New Model to Distinguish Railhead Defects Based on Set-Membership Type-2 Fuzzy Logic System

This paper focuses on the new model for the classification of railhead defects, through images acquired by a rail inspection vehicle. In this regard, we discuss the use of set-membership concept, derived from the adaptive filter theory, into the training procedure of an upper and lower singleton type-2 fuzzy logic system, aiming to reduce computational complexity and to increase the convergence speed. The performance is based on the data set composed of images provided by a Brazilian railway company, which covers the four possible railhead defects (cracking, flaking, head-check and spalling) and the normal condition of the railhead. Additionally, we apply different levels of additive white Gaussian noise in the images in order to challenge the proposed model. Finally, we discuss performance analysis in terms of convergence speed, computational complexity reduction, and classification ratio. The reported results show that the proposal achieved improved convergence speed, slightly higher classification ratio and remarkable computation complexity reduction when we limit the number of epochs for training, which may be required under real-time constraint or low computational resource availability.

Forecast Method of Track Irregularity of Heavy-haul Railway Based on BP Neural Network

The track irregularity is an important factor for the transport safety of heavy-haul railway. Limited by the technique of data mining and analysis, the data obtained by rail inspection vehicles cannot fully identify the status of track irregularity. In this paper, based on the characteristics of track irregularity, the BP neural network is used to predict the geometric irregularity parameters of heavy-haul railway tracks, identify the status of track irregularity, and provide support for the decision-making of maintenance strategy. In order to further verify the accuracy of the BP neural network, the single and double hidden layer networks are established to predict 20 sets of the 7 indicators of track irregularity. According to the prediction results, the mean square errors of the single and double hidden layer networks are 0.064 and 0.051, respectively. The result shows that the multi-hidden layer BP neural network has higher accuracy, which provides a new idea for the research on the prediction model of track irregularity.

Research on Combined Location Method of Dual Rail Inspection Vehicle Based on Adaptive Kalman Filter

Dual rail inspection vehicle has the function of inspecting rail defect points automatically and an odometer is required for its localization. Since real-time localization of rail defect points still needs manual calibration, it cannot meet the requirements of automatic vehicle operation. In this paper, the automatic localization methods are discussed. The Adaptive Kalman Filter (AKF) is employed for the multi-sensor data fusion of the odometer and the Global Navigation Satellite System (GNSS). Then the GNSS/odometer combined localization model has been built to improve the accuracy and reliability of the inspection vehicle localization. In addition, the hardware board is designed. Finally, the on-line test has been performed on the Shanghai-Hangzhou railway line. The experimental results show that our combined localization method can significantly improve the accuracy compared with that of the single GNSS method from 1.848m to 1.093m.

The need to consider searcher efficiency and carcass persistence in railway wildlife fatality studies

To improve mitigation planning on railways, it is crucial to accurately assess the number of fatalities resulting from this type of infrastructure. To illustrate the relevance of considering the two main errors from the sampling process (searcher efficiency and carcass persistence), we corrected the estimation of mammal fatalities (> 1 kg) on a 750-km railway located in the Brazilian Savanna accounting for these errors. Observed fatalities were collected by two observers in a rail inspection vehicle (RV). We estimated searcher efficiency comparing the number of carcasses found during the searcher efficiency trials on foot and by RV on the same subsections of the railway. Carcass persistence was estimated based on trials with a subset of carcasses that had their persistence verified over three consecutive days. Searcher efficiency was 29.8% and daily carcass persistence probability was 99.7%. Nine surveys on the entire railway (2015–2016) recorded 1950 carcasses and, after correcting for the sampling errors, we estimated that 4286 mammals died on this railway in 23 months. This estimate is 2.2 times larger than the observed number of fatalities. These errors should not be neglected, as our results demonstrate that the number of observed carcasses is a misperception in relation to the true number of fatalities caused by the railway. However, our results also indicate that dealing with these errors in railway ecology is still a challenge. We stress the need to test more suitable sampling designs for fatality estimates on railways.

Research on the three-dimensional detection system of the rail full profile

To solve the problem of slow speed, low precision and low efficiency of manual measurement in the current inspection of rail. Based on the non-contact laser profilers, a three-dimensional detection system of the full profile of rail is proposed by using the integrated combination of four laser profilers. The top straightness of the rail is detected by two-point connection method and the bottom distortion of the rail is detected by the common-plane method. After field testing, the machine measurement accuracy of the rail height, rail head width, rail bottom width and rail waist width in both rail head and tail are 0.039, 0.037, 0.029, 0.036mm and 0.047, 0.037, 0.039, 0.042mm respectively; The distortion are 0.035 and 0.039mm respectively. Experiments show that the system can complete the measurement of rail efficiently and steadily, and improve the speed, accuracy and efficiency of measurement.

A new feature for detecting defects in ultrasonic high-speed rail inspection

In high-speed testing of railway rails using ultrasonic transducers, there are significant deviations of the carrier frequency of echo signals from the frequency of filling the probing pulses. They are caused by the Doppler Effect. Since inclined ultrasonic transducers have a certain width of the directional diagram, the Doppler frequency of the echo signal from any reflector in the rail has a deviation that reaches 14-30% of the average value of the Doppler shift. Analysis of this frequency modulation of received signals allows detecting possible defects in rails with high noise immunity.