EP4335717A1

EP4335717A1 - Device for inspecting defects in a railway

Info

Publication number: EP4335717A1
Application number: EP22382840.1A
Authority: EP
Inventors: Héctor Sanz Mateos; Francisco Javier Carvajal Pérez; Germán Castaño Roldán
Original assignee: Ingeniera Y Economia Del Transporte SME MP SA
Current assignee: Ingeniera Y Economia Del Transporte SME MP SA
Priority date: 2022-09-12
Filing date: 2022-09-12
Publication date: 2024-03-13

Abstract

The present invention relates to a device for inspecting defects in a railway, and more specifically to a device with image capture means configured to simultaneously acquire at least zenithal images and lateral images of a track of the railway and with processing means in data communication with the image capture means configured to receive and process the images acquired by means of an artificial intelligence model.

Description

Technical field of the invention

Background of the invention

Safety in rail transport is closely related to the condition of the railway. Some of the defects that can be found in railways are cracks or breaks in tracks, spalling, skidding, shelling, squats, corrugations, cracks in the fastenings and sleepers, corrosion in the metal parts, blows and other anomalies that can cause a catastrophic failure in the infrastructure. Most defects are found in the tracks, particularly at the head of the track, however, anomalies can also occur in the web and foot of the track, which means that it is of the utmost importance to inspect the track in its entirety.
Traditionally, the state of conservation of railways had to be periodically checked with visual inspection by railway company personnel; this inspection system not only requires a lot of personnel and labour time, but also depends on the capacity, training and experience of each of the individuals responsible for the inspection.
As an alternative, it is possible to carry out destructive and non-destructive tests on the elements of the railway, especially the tracks, which are the critical elements of the infrastructure. Although the tests produce homogeneous, standardised and reliable results, their great drawback is that they must be carried out by specialised personnel in laboratories or workshops that have the required equipment, and generally involves extracting samples of material, which in practice means the loss of the studied section.
Portable pieces of equipment have also been used to carry out non-destructive tests that can be transported and used on the same railway, but these processes are considerably slow, require specialised personnel and only offer a reliable reading of specific points on the railway, and given the current extension of the railway networks is unfeasible.
Obtaining images of the railway for subsequent inspection by an operator has also been considered; however, after capturing the images, the standards and attention to detail that the operator must have in front of the monitor when reviewing thousands of images generates fatigue and loss of concentration that produces false positives or, even worse, causes him/her to ignore problems due to lack of attention during viewing.
With the current increasing rate in railway traffic, carrying ever heavier loads at higher speeds, a faster, more reliable and more efficient way to inspect railways is a must. Therefore, there is still a great room for improvement in the field of railway inspection devices.

Description of the invention

The present invention proposes a solution to the foregoing problems by means of a device for inspecting defects in a railway as described below.
In a first inventive aspect, the invention provides a device for inspecting defects in a railway, wherein the railway comprises two tracks, wherein the device for inspecting defects is characterised in that it comprises:

a structure configured to move over the tracks of the railway,
image capture means arranged in the structure and configured to simultaneously acquire at least zenithal images and lateral images of a track of the railway,
lighting means arranged in the structure and configured to illuminate at least one track of the railway, and
processing means in data communication with the image capture means, wherein the processing means are configured to receive and process the images acquired by the image capture means

wherein the processing means comprise an artificial intelligence model configured to recognise defects in the railway from the images acquired by means of the image capture means.

Throughout this document, a railway will be understood as the railway infrastructure that enables a railway vehicle to transit, and which comprises, in a conventional manner, two tracks arranged in parallel at a fixed distance, struts or sleepers to which the tracks are attached and that help to keep them separated at a fixed distance, fastenings or attachments to join the rails to the struts; in some cases, the railway may comprise other elements, such as ballasts, signalling installations, electrification installations, switches and crossings, etc.
The structure will be understood as an element formed by one or more portions, the function of which is to serve as a support and framework for at least some of the other elements making up the inspection device, which allows the inspection device to be placed or arranged on the tracks, and which also allows the movement thereof along the railway as a railway vehicle; in a preferred embodiment, the structure comprises three or more wheels, with at least two of them arranged in parallel at a distance corresponding to the track gauge of the railway. In some embodiments, the structure is configured for adapting the wheelbase to various track gauges.
The image capture means will also be understood as any technological element capable of obtaining or capturing images; in a preferred embodiment, the image capture means comprise digital photographic or video cameras, as well as other auxiliary elements that allow the operation thereof, such as power supply means or data connection cables. The image capture means are preferably attached to the structure by means of an articulated joint, and are oriented such that during normal operation of the inspection device they allow at least zenithal images of the railway and lateral images of a track to be simultaneously obtained.
Zenithal images should be understood as images obtained at a zenithal angle with respect to the plane of the tracks of the railway, i.e., from a point of view located vertically and above the plane defined by the tracks; zenithal images show at least one head of the track, but may show the heads of both tracks, and/or other railway elements; preferably the zenithal images also show a portion or all of the strut, sleeper or plate that fulfils these functions. Lateral images of a track are to be understood as images showing the lateral portion or flank of at least one of the tracks; preferably, the lateral images show one side of the web of the track, and may also show attachment elements and other elements of the railway. To obtain lateral images, an image capture element can be arranged focussing, pointing to or oriented towards the side of the track.
Lighting means should be interpreted as one or more light sources capable of providing lighting in a controlled manner to a section of the railway or to a specific area of an element of this railway; in a preferred embodiment, the lighting means comprise one or more lamps oriented such that during normal operation of the inspection device they illuminate at least the zenithal area and the lateral area of the track.
Processing means are to be understood as any technological element capable of processing information, in particular images captured by the image capture means; in a preferred exemplary embodiment, the processing means are a digital computer arranged in the structure of the inspection device; in another example, the processing means is a computer capable of wireless communication that is separated from the structure of the device. The processing means are in data communication by conventional means, such as cables or radio frequency communication systems. In addition, the processing means comprise an artificial intelligence model obtained from an image recognition algorithm, so that it is capable of recognising different elements in an image of the railway, as well as detecting anomalies and defects in the elements of the railway, especially the tracks.
Advantageously, the inspection device is configured to be arranged on the tracks of the railway to be inspected like a conventional railway vehicle, capable of moving along the railway at a constant or approximately constant speed; once arranged or placed on the rail, the image capture means acquire zenithal images of the railway and, simultaneously, also acquire lateral images of one or both tracks; this process can be repeated periodically at regular segments of the rail, between which the inspection device moves, or it can be carried out continuously, without the device stopping, depending on the performance of the image capture means. Given that the lighting conditions of the railway are usually unpredictable and heterogeneous, the lighting means allow the area of the railway from which the image capture means obtain images to be illuminated under certain conditions. The images are transmitted to the processing means, which process them through the artificial intelligence model to detect defects in the railway.
In a particular embodiment, the image capture means comprise, for each track of the railway, a camera that is orientable towards the head of the track, and a camera that is orientable towards the web of the track. The cameras of the image capture means are arranged such that during the normal operation of the inspection device they are each oriented, focussed or pointed towards a specific portion of one of the tracks; in this way, a first camera points to the head of one track, a second camera points to the web of the same track, a third camera points to the head of the other track, and a fourth camera points to the web of this other track. In one embodiment, the cameras pointing to the web of the tracks are oriented towards the outside faces of the web, providing lateral images of the tracks. Orientable should be understood as the element being configured to be oriented during the normal operation of the inspection device towards a specific area of the railway; in one embodiment, the orientable condition is achieved by means of an articulated joint between the camera and the structure.
In a particular embodiment, the image capture means further comprise two other cameras that are each orientable to the web of a track of the railway, and one camera that is orientable to the space between the tracks of the railway. Preferably, the two additional cameras that are orientable to the webs of the tracks are arranged such that they can focus on the side of the web of the track that is not focussed on by the previous cameras, for example on the inside faces of the tracks. For its part, the camera that is orientable to the space between the tracks allows images to be obtained of both the tracks and one or more of the elements that are within the angle of focus, such as sleepers, struts and fastenings.
In a particular embodiment, the image capture means further comprise one or more cameras that are orientable to one or more of: a railway strut, a railway attachment, railway ballast, a railway signalling apparatus. Advantageously, the inspection device comprises additional cameras configured to obtain images of one or more of the listed elements. This enables specific elements of the railway to be exclusively or redundantly inspected.
In a particular embodiment, the lighting means comprise two lamps, each being orientable to a track of the railway, and a lamp that is orientable to the space between the tracks of the railway. Advantageously, the lamps enable specific portions of the railway to be illuminated in a controlled manner so that the images obtained have minimum brightness and contrast conditions.
In a particular embodiment, the device further comprises at least one light detection and ranging, LiDAR, distance measurement system in data communication with the processing means. With LiDAR systems, accurate distances can be obtained from a set of points to a reference position, which allows anomalies to be detected in the position of elements of the railway.
In a particular embodiment, the device further comprises positioning means for positioning the device in data communication with the processing means. Advantageously, the positioning means provide position coordinates of the inspection device that can be associated with each set of images obtained.
In a particular embodiment, the positioning means comprise one or more of: a GNSS satellite navigation module, a real-time kinematic, RTK, satellite navigation module, an inertial navigation system, INS. A combination of the foregoing positioning systems allows for improved position reading, which in turn contributes to a more effective anomaly detection.
In a particular embodiment, the device further comprises a self-calibration system configured to detect the metering of the railway. Advantageously, the self-calibration system allows the correction of possible position errors of the inspection device.
In a particular embodiment, the device further comprises propulsion means configured to propel the device in a controlled manner. The inspection device can be used to statically inspect a single section of railway, but its application is more advantageous if the inspection device is moved along a long segment of railway, which can be achieved by means of an external traction, for example, by pushing the device along the rail, or more conveniently, by propulsion means that propel the inspection device at an approximately constant and adjustable speed. In a preferred embodiment, the device comprises an electric motor operatively connected to at least one wheel, wherein the motor is powered by lithium batteries. In another embodiment, the electric motor comprises a controller configured to be remotely operated by a remote control.
In a particular embodiment, the structure comprises at least one telescopic joint configured for adapting the device to various track gauges. Advantageously, the telescopic joint allows the inspection device to be used in different types of gauges, from 1400 to 1700 mm; in a preferred embodiment, the structure comprises fastening systems to secure the track gauge when it is in operation.
In a particular embodiment, the processing means comprise a data storage unit. Advantageously, the data storage unit allows non-transitory storage of the images obtained by the image capture means together with additional data, such as the date and time of acquisition and the coordinates of the place where it was acquired. This information can be used as a log for later reference.
These and other features and advantages of the invention will be evident in light of the description of preferred, but not exclusive, embodiments which are illustrated by way of non-limiting example in the drawings which are attached.

Brief description of the drawings

Figures 1a, 1b: show a preferred embodiment of the inspection device, from two different angles.

Detailed description of an exemplary embodiment

In the following detailed description, numerous specific details are set forth in the form of examples to provide a thorough understanding of the relevant teachings. However, it will be apparent to those skilled in the art that the present teachings can be implemented without such details.
Figures 1a and 1b show a preferred exemplary embodiment of the device (1) for detecting defects in a railway from two different points of view. In the embodiment shown, the detection device (1) is built around a structure (2) or chassis formed by a set of structural aluminium profiles; in other embodiments, the structure (2) is formed using 3D printing with SLS technology with polyamide-glass fibre (PA-GF) reinforced powder material.
The structure (2) of the example is configured as a beam that is placed on the tracks of the railway to be inspected, substantially perpendicular to the axis of the rail; the structure (2) rests on the tracks by means of two wheels mounted at the ends of a crossbar arranged at one end of the beam, and by means of a set of two wheels in tandem arranged at the opposite end of the beam. A set of profile segments serves to join the different means and elements of the device (1) to the structure (2).
In some examples, the beam comprises two portions joined by means of a telescopic joint that enables the wheelbase to be selectively modified, so that it can be adapted to different track gauges. In a particular example, the structure (2) comprises locking elements that allow the telescopic joint to be interlocked and the position of the device (1) to be locked in track gauges of 1400 mm or 1700 mm during operation.
In the embodiment described, the image capture means (3) comprise seven digital cameras, with three cameras focussing on each track from three different angles, and another focussing on the space between the tracks; for each track there is a camera in a zenithal position with respect to the track, and two cameras oriented to the sides of the track, one focussing on the outside face of the web of the track, and another camera focussing on the outside face of the web of the track. In the example shown, the cameras are mounted by articulated joints at the ends of prismatic profiles that are joined to the beam of the structure (2). The cameras can work continuously as video cameras, or acquiring images at regular time intervals, in a discreet manner.
To ensure that the images have homogeneous brightness and contrast conditions, the device comprises lighting means (4) that in the example are implemented by means of three pairs of lamps, one pair for each track, and another pair focussing on the space between the tracks. In the example shown, the lamps are mounted on some of the profiles that support the cameras.
The acquired images are transmitted by cables to the processing means (5) which in this example are made up of a computer; this computer is powered by rechargeable lithium batteries and comprises a data storage unit in which the acquired images are stored for later reference; in this example, the data storage unit is a conventional hard drive. In some examples, the computer further comprises wireless communication means, for example wireless communication means capable of connecting to a data network.
In this example, the device (1) comprises positioning means (6), which comprise a GPS module and an odometer; the positioning means (6) provide the current coordinates of the device (1) at the time the images of the railway are acquired, and they are stored together with them. In other examples, the positioning means (6) comprise other systems or modules that offer a reading of the position of the device (1) on the railway, such as satellite navigation modules, GNSS, systems other than the GPS system, real-time kinematic satellite navigation modules, or RTK, or inertial navigation systems, INS.
In the example described, the computer comprises an artificial intelligence model configured to identify the elements that make up the railway in the images and to detect anomalies or defects in them. The artificial intelligence model is obtained through a process of training an image recognition algorithm with a set of labelled images of railways, including images with anomalies and images with the railway in good condition. This model allows, firstly, identifying each of the elements of the railway that appear in the image, and secondly, locating possible anomalies or defects in real time; given the reading of the positioning means (6), the presence of anomalies is associated with a specific position of the railway to facilitate its inspection and validation by qualified personnel. In one example, the artificial intelligence model provides an assessment of the severity of the anomaly, for example, the length of a crack in a strut; the assessment can be represented, for example, by means of a colour mask superimposed on the anomaly in the image. In other embodiments, the computer transmits the images via a data network to another computer that comprises the artificial intelligence model and executes the described image analysis.
In this example, the device (1) further comprises a LiDAR system, not shown in the figures, which allows the dimensions of the elements of the railway to be accurately calculated, and to determine whether there is any deformation that exceeds the tolerance limits.
It also comprises a self-calibration system; by detecting the rail metering placed in the web of the track, an image capture system is arranged that allows the device (1) to be repositioned on the rail, eliminating possible errors accumulated by the positioning means (6) between metering marks.
To propel the device (1), it comprises propulsion means (7), implemented by means of an electric motor that transmits power to the wheels arranged at the end of the beam. The propulsion means (7) in this example can be controlled remotely by a remote control that allows the personnel responsible for its operation to establish a constant speed.

Claims

A device (1) for inspecting defects in a railway, wherein the railway comprises two tracks, wherein the device (1) for inspecting defects is characterised in that it comprises:
a structure (2) configured to move over the tracks of the railway,

image capture means (3) arranged in the structure (2) and configured to simultaneously acquire at least zenithal images and lateral images of a track of the railway,

lighting means (4) arranged in the structure (2) and configured to illuminate at least one track of the railway, and

processing means (5) in data communication with the image capture means (3), wherein the processing means (5) are configured to receive and process the images acquired by the image capture means (3);
wherein the processing means (5) comprise an artificial intelligence model configured to recognise defects in the railway from the images acquired by the image capture means (3).
The device (1) for inspecting defects according to the preceding claim, wherein the image capture means (3) comprise, for each track of the railway, a camera that is orientable towards the head of the track, and a camera that is orientable towards the web of the track.
The device (1) for inspecting defects according to the preceding claim, wherein the image capture means (3) further comprise two other cameras that are each orientable to the web of a track of the railway, and one camera that is orientable to the space between the tracks of the railway.
The device (1) for inspecting defects according to any of claims 2-3, wherein the image capture means (3) further comprise one or more cameras that are orientable to one or more of: a railway strut, a railway attachment, railway ballast, a railway signalling apparatus.
The device (1) for inspecting defects according to any of the preceding claims, wherein the lighting means (4) comprise two lamps, each being orientable to a track of the railway, and a lamp that is orientable to the space between the tracks of the railway.
The device (1) for inspecting defects according to any of the preceding claims, wherein the device (1) further comprises at least one light detection and ranging, LiDAR, distance measurement system in data communication with the processing means (5).
The device (1) for inspecting defects according to any of the preceding claims, wherein the device further comprises positioning means (6) of the device (1) in data communication with the processing means (5).
The device (1) for inspecting defects according to the preceding claim, wherein the positioning means (6) comprise one or more of: a GNSS satellite navigation module, a real-time kinematic, RTK, satellite navigation module, an inertial navigation system, INS.
The device (1) for inspecting defects according to any of the preceding claims, wherein the device (1) further comprises a self-calibration system configured to detect the metering of the railway.
The device (1) for inspecting defects according to any of the preceding claims, wherein the device (1) further comprises propulsion means (7) configured to propel the device in a controlled manner.
The device (1) for inspecting defects according to any of the preceding claims, wherein the structure (2) comprises at least one telescopic joint configured for adapting the device (1) to various track gauges.
The device (1) for inspecting defects according to any of the preceding claims, wherein the processing means (5) comprise a data storage unit.