Information processing device, information processing method, and recording medium

Abstract

In an image including a region of a slider end face in an overhead wire extension direction, of the slider of a pantograph in contact with an overhead wire, and a region of a predetermined member located in the vicinity of the slider, the region of the slider end face and the region of the predetermined member are detected using a region detection model. The thickness of the slider in the vertical direction is calculated on the basis of the ratio of the distance of the slider end surface in the vertical direction in the image and the distance in the predetermined direction in the image in the region of the predetermined member, and the actual dimension value in the predetermined direction in the region of the predetermined member.

Description

Information Processing Apparatus, Information Processing Method, and Recording Medium

[0001] The present disclosure relates to an information processing apparatus, an information processing method, and a recording medium.

[0002] A technology of an inspection support system for inspecting the wear state of a rubbing plate of a pantograph is disclosed in Patent Document 1.

[0003] Patent Document 1 discloses a technique for detecting an edge of a boat body that supports a rubbing plate to determine a measurement reference, and detecting an edge of a contact surface of the rubbing plate based on the measurement reference, thereby calculating a value representing the wear of the rubbing plate.

[0004] Japanese Patent Application Laid-Open No. 2022-92182

[0005] However, it has been demanded to detect the wear state of the rubbing plate of the pantograph with higher accuracy.

[0006] An object of the present disclosure is to provide an information processing apparatus, an information processing method, and a recording medium that solve the above problems.

[0007] An information processing apparatus according to an aspect of the present disclosure includes region detection means for detecting a region of an end face of a rubbing plate in a wire extension direction of a rubbing plate of a pantograph that contacts a wire and a region of a predetermined member located in the vicinity of the rubbing plate in an image, and a ratio between a distance in an image in the vertical direction of the end face of the rubbing plate and a distance in an image in a predetermined direction in the region of the predetermined member, and thickness calculation means for calculating the thickness in the vertical direction of the rubbing plate based on the ratio and an actual dimension value in the predetermined direction in the region of the predetermined member.

[0008] An information processing method according to an aspect of the present disclosure includes detecting a region of an end face of a rubbing plate in a wire extension direction of a rubbing plate of a pantograph that contacts a wire and a region of a predetermined member located in the vicinity of the rubbing plate in an image using a region detection model, and calculating the thickness in the vertical direction of the rubbing plate based on a ratio between a distance in an image in the vertical direction of the end face of the rubbing plate and a distance in an image in a predetermined direction in the region of the predetermined member, and an actual dimension value in the predetermined direction in the region of the predetermined member.

[0009] A recording medium according to one aspect of the present disclosure stores a program that causes the computer of an information processing device to function as: an area detection means that detects the area of ​​the pantograph contact strip end face in the direction of extension of the overhead wire and the area of ​​a predetermined member located near the pantograph contact strip using an area detection model; and a thickness calculation means that calculates the thickness of the pantograph contact strip in the vertical direction based on the ratio of the distance in the image of the pantograph contact strip end face in the vertical direction and the distance in the image of the area of ​​the predetermined member in a predetermined direction, and the actual dimensions of the area of ​​the predetermined member in the predetermined direction.

[0010] According to one embodiment described above, an information processing device is provided that can detect the wear condition of the pantograph's contact strip with greater accuracy.

[0011] This figure shows the configuration of a pantograph inspection system according to one embodiment of the present disclosure. This figure shows the hardware configuration of an information processing device according to one embodiment of the present disclosure. This is a functional block diagram of an information processing device according to one embodiment of the present disclosure. This is the first figure showing an image of a pantograph taken according to one embodiment of the present disclosure. This figure shows the processing flow of an information processing device according to one embodiment of the present disclosure. This is the first figure showing an overview of the processing of an information processing device according to one embodiment of the present disclosure. This is the second figure showing an overview of the processing of an information processing device according to one embodiment of the present disclosure. This is the third figure showing an overview of the processing of an information processing device according to one embodiment of the present disclosure. This is the fourth figure showing an overview of the processing of an information processing device according to one embodiment of the present disclosure. This is the fifth figure showing an overview of the processing of an information processing device according to one embodiment of the present disclosure. This is the sixth figure showing an overview of the processing of an information processing device according to one embodiment of the present disclosure. This is the second figure showing an image of a pantograph taken according to one embodiment of the present disclosure. This is the seventh figure showing an overview of the processing of an information processing device according to one embodiment of the present disclosure. This is a functional block diagram of another example of an information processing device according to the present disclosure. This figure shows a processing flow of another example of an information processing device according to the present disclosure.

[0012] The pantograph inspection system, including the information processing device disclosed herein, will be described below with reference to the drawings.

[0013] Figure 1 shows the configuration of a pantograph inspection system according to one embodiment of the present disclosure. The pantograph inspection system 100 comprises at least an information processing device 1 for inspecting wear on the pantograph 2, an imaging device 3 for photographing the pantograph 2, a light 4 for illuminating the pantograph with light, and a PoE (Power over Ethernet) hub 5 for communication connection between the information processing device 1 and the imaging device 3 and the light 4.

[0014] The information processing device 1 communicates with the camera 3 and light 4 via the PoE hub 5, thereby enabling it to communicate with the camera 3 and light 4 and supply power to them. The information processing device 1 analyzes the pantograph 2 contained in the image captured and generated by the camera 3 to inspect the wear condition of the contact strip that contacts the overhead wire on the pantograph 2.

[0015] Figure 2 shows the hardware configuration of an information processing device according to one embodiment of the present disclosure. As shown in Figure 2, the information processing device 1 is a computer equipped with hardware such as a CPU (Central Processing Unit) 101, ROM (Read Only Memory) 102, RAM (Random Access Memory) 103, storage device 104, and communication module 105. The information processing device 1 may be a PC (Personal Computer) or a cloud server.

[0016] Figure 3 is a functional block diagram of an information processing device according to one embodiment of the present disclosure. The information processing device 1 executes a pantograph inspection program that is stored in advance. As a result, the information processing device 1 performs the functions of the control unit 11, area detection unit 12, linear approximation unit 13, over-detection removal unit 14, measurement location identification unit 15, thickness calculation unit 16, and output unit 17.

[0017] The control unit 11 controls other functional units. The region detection unit 12 uses an image that includes the region of the pantograph's contact strip end face in the direction of the overhead wire extension and the region of a predetermined member located near the contact strip to detect the region of the contact strip end face and the region of the predetermined member. The linear approximation unit 13 linearly approximates the contours of the region of the contact strip end face and the region of the predetermined member, as well as the estimated point cloud. The overdetection removal unit 14 identifies and removes the overdetection region from the processing area if an overdetection region occurs in the machine learning inference based on the results of detecting the region of the contact strip end face and the region of the predetermined member using the region detection model. The measurement location identification unit 15 performs processing such as identifying the shortest distance between the upper edge (upper contour) and the lower edge (lower contour) of the contact strip end face as the distance in the vertical direction of the image of the contact strip end face. The thickness calculation unit 16 calculates the vertical thickness of the abrasive plate based on the ratio of the distance in the vertical image of the abrasive plate end face to the distance in the image in a predetermined direction within the region of the predetermined member, and the actual dimensions in a predetermined direction within the region of the predetermined member. The output unit 17 outputs inspection results indicating the wear condition of the abrasive plate to an output destination such as a monitor.

[0018] Figure 4 shows an image of a pantograph according to one embodiment of the present disclosure. The imaging device 3 photographs the pantograph. The imaging device 3 generates an image of the pantograph and transmits it to the information processing device 1. The information processing device 1 receives the image. The image received by the information processing device 1 shows at least the end face of the contact strip 21 of the pantograph 2 in the direction of the extension of the overhead wire and in the direction of vehicle travel. The image also shows at least the end face of the shoe body 22, which is a support member that supports the contact strip 21 of the pantograph 2 from below, in the direction of the extension of the overhead wire and in the direction of vehicle travel. The shoe body 22 is equipped with a spring mechanism that allows the contact strip 21 to move vertically relative to the shoe body 22, so that the contact strip 21 can always contact the overhead wire when the vehicle to which the pantograph 2 is attached is in motion. Both the contact strip 21 and the shoe body 22 are elongated plate shapes and constitute a part of the pantograph 2. The camera device 3 only needs to be pre-installed in a position where the pantograph's contact strip 21 and the front end of the pantograph's shoe 22 in the direction of vehicle travel are included in the field of view.

[0019] The information processing device 1 of this disclosure detects the region of the end face of the sliding strip 21 (divided sliding strip) in the direction of vehicle travel, and the region of the end face of the boat body 22 in the direction of vehicle travel, using a machine learning model (region detection model) generated using machine learning. The region detection model is a machine learning model obtained by learning a large number of correct regions of the region of the end face of the sliding strip 21 in the direction of vehicle travel and the region of the end face of the boat body 22 in the direction of vehicle travel, as seen in the image, using machine learning. Alternatively, the region detection model may be a machine learning model for detecting the region of the end face of the sliding strip 21 in the direction of vehicle travel and the region of the end face of the boat body 22 in the direction of vehicle travel, obtained by machine learning many of the images taken by the camera 3 under multiple different weather conditions and shooting times (sunny, cloudy, rainy, night, daytime, etc.). This makes it possible to generate a machine learning model that can detect the region of the end face of the sliding strip 21 in the direction of vehicle travel and the region of the end face of the boat body 22 in the direction of vehicle travel with high accuracy. The image shows the end faces of the contact strips 21 and the shoe body 22 in the direction of the overhead wire extension and rear in the direction of vehicle travel. The information processing device 1 may use a machine learning model (region detection model) to detect these end faces of the contact strips 21 and the shoe body 22 in the direction of the overhead wire extension and rear in the direction of vehicle travel, and detect the wear state (thickness) of the contact strips 21 based on the detected region.

[0020] In this disclosure, the predetermined member visible in the image for calculating the thickness of the sliding strip 21, as described above, is defined as the area of ​​the end face of the boat body 22 in the direction of vehicle travel. Hereinafter, the area of ​​the end face of the sliding strip 21 in the direction of vehicle travel will be referred to as the sliding strip end face 210, and the area of ​​the end face of the boat body 22 in the direction of vehicle travel will be referred to as the boat body end face 220. The information processing device 1 calculates the vertical thickness of the sliding strip end face 210 based on the ratio of the vertical distance in the image of the sliding strip end face 210 to the vertical distance in the image of the boat body end face 220, and the actual vertical dimensions of the area of ​​the boat body end face 220.

[0021] The information processing device 1 of this disclosure can automatically calculate the vertical thickness of the contact strip end surface 210 in a short time by detecting the region of the contact strip end surface 210 and the boat end surface 220 using a region detection model. Furthermore, when measuring the vertical thickness of the contact strip end surface 210, the information processing device 1 of this disclosure can calculate the vertical thickness of the contact strip end surface 210 with higher accuracy by not approximating the upper edge of the contact strip end surface 210 with a straight line. Furthermore, when measuring the vertical thickness of the contact strip end surface 210, the information processing device 1 of this disclosure can calculate the vertical thickness of the contact strip end surface 210 with higher accuracy by identifying the position with the shortest distance between the parallel upper and lower edges of the contact strip end surface 210 and measuring the thickness of the contact strip end surface 210 at that position. Furthermore, in order to confirm whether the upper and lower edges of the abrasive plate end face 210 are parallel, if the abrasive plate 21 is a segmented abrasive plate, the information processing device 1 of the present disclosure can exclude the trapezoidal left and right edge portions of each independent abrasive plate end face 210 and identify the position of the worn portion within the width of the parallel upper and lower edges, thereby accurately determining the vertical length (wear) of the abrasive plate end face 210. In addition, the information processing device 1 of the present disclosure can similarly consider not only segmented abrasive plates but also integrated abrasive plates as a single abrasive plate end face and accurately determine the vertical length (wear) of the end face of the abrasive plate 21.

[0022] Figure 5 is a diagram showing the processing flow of an information processing device according to one embodiment of the present disclosure. Next, the processing flow of the information processing device 1 will be described in order. First, the information processing device 1 receives an image from the imaging device 3 (step S101). The control unit 11 outputs the image to the region detection unit 12.

[0023] Figure 6 is a first diagram showing the processing overview of an information processing device according to one embodiment of the present disclosure. The region detection unit 12 inputs an image to a neural network using a region detection model (step S102). As a result, the region detection unit 12 detects the region of the sliding plate end surface 210 and the region of the boat end surface 220 in the image, as shown in Figure 6 (step S103). At this time, the detection results of these regions may include overdetected regions. Overdetected regions are the result of processing in which the sliding plate end surface 210 and the boat end surface 220 are detected using the region detection model, and are regions of overdetection in machine learning inference. The region detection unit 12 identifies the regions of the sliding plate end surface 210 and the boat end surface 220 detected in the image, as well as the coordinates of each pixel indicating the contour of said region (step S104).

[0024] The region detection unit 12 outputs the image and detection results, which include the region between the sliding plate end face 210 and the boat-shaped end face 220 detected in the image, as well as the coordinates of each pixel of the contour of the region, to the linear approximation unit 13 and the overdetection removal unit 14. The region between the sliding plate end face 210 and the boat-shaped end face 220 is the result of region detection on a pixel basis. In this disclosure, the region of the sliding plate end face 210 is an example where the sliding plate 21 is a segmented sliding plate, so as shown in Figure 6, multiple trapezoidal cross-sectional regions are connected horizontally. The region of the boat-shaped end face 220 has a horizontally elongated rectangular shape. In other embodiments, the region detection unit 12 may detect regions other than the boat-shaped 22 as a reference region for calculating the vertical thickness of the sliding plate 21. The predetermined member used as a reference for calculating the vertical thickness of the contact strip 21 may be any member within the pantograph shown in the image (for example, the screws that fix the contact strip 22, or the horns that are attached to the ends of the contact strip 22), as long as a non-wearing member can be selected, in addition to the shoe body 22.

[0025] The linear approximation unit 13 acquires the detection result of the region between the sliding plate end face 210 and the boat end face 220. Based on the coordinates indicating the sliding plate end face 210 included in the detection result, the linear approximation unit 13 uses the coordinate points (pixels) indicating the contours of the lower edge, upper edge, left edge, and right edge of the sliding plate end face 210 to linearly approximate each contour of the lower edge, upper edge, left edge, and right edge (step S105). Note that the linear approximation unit 13 only needs to linearly approximate the contours of at least the lower edge and upper edge using the coordinate points (pixels) indicating the contours of the lower edge and upper edge of the sliding plate end face 210. In other disclosures, linear approximation of the contours of the left edge and right edge is not required. In other disclosures, the detection result of the region of the sliding plate end face 210 may be input to a contour extraction model generated by machine learning, and the coordinates of the contour of the region of the sliding plate end face 210 obtained as a result may be linearly approximated. Furthermore, the linear approximation unit 13 uses coordinate points (pixels) that indicate the contours of the lower edge, upper edge, left edge, and right edge of the boat end face 220 to linearly approximate the contours of the lower edge, upper edge, left edge, and right edge (step S106). The linear approximation unit 13 only needs to linearly approximate the contours of at least the lower edge and upper edge using coordinate points (pixels) that indicate the contours of the lower edge and upper edge of the boat end face 220. In other disclosures, linear approximation of the contours of the left edge and right edge of the boat end face 220 is not required. In other disclosures, the detection results of the region of the boat end face 220 may be input to a contour extraction model generated by machine learning, and the coordinates of the contour of the region of the boat end face 220 obtained as a result may be linearly approximated. This allows the linear approximation unit 13 to extract the contour of the region between the sliding plate end face 210 and the boat end face 220. Linear approximation can be performed using methods such as RANSAC (Random Sample Consensus), Hough Transform, or least squares method.

[0026] In addition, if the upper edge of the contact strip end face 210 is extremely worn down due to friction with the overhead wire, the linear approximation section 13 may not be parallel to the linear approximation obtained by the linear approximation of the coordinates of the contour of the upper edge, and may have an extreme inclination. In such cases, the linear approximation section 13 does not need to perform linear approximation of the upper edge of the contact strip end face 210. Furthermore, in this embodiment, even if the linear approximation section 13 performs linear approximation of the upper edge of the contact strip end face 210, the linear approximation data is not used in the subsequent processing to calculate the vertical thickness of the contact strip end face. Instead, the vertical thickness of the contact strip end face is calculated using the contour of the region at the upper edge and the identified point clouds, thereby enabling a more accurate calculation of the vertical thickness of the contact strip end face.

[0027] Figure 7 is a second diagram showing the processing overview of an information processing device according to one embodiment of the present disclosure. The over-detection removal unit 14 also acquires the detection result of the region between the rubbing plate end face 210 and the boat end face 220. The over-detection removal unit 14 determines whether there is an over-detection region (step S107). If there is an over-detection region, the over-detection removal unit 14 deletes the information of the over-detection region from the detection result (step S108).

[0028] Specifically, the over-detection removal unit 14 identifies areas that do not have the predetermined shape and area assumed to be the sliding plate end face 210 or the boat end face 220, among the areas detected by the area detection unit 12 as being part of the sliding plate end face 210 or the boat end face 220, as over-detection areas. The over-detection removal unit 14 also identifies areas detected by the area detection unit 12 below the boat end face 220 in the image. Furthermore, the over-detection removal unit 14 identifies areas that do not intersect with a perpendicular line drawn downward from the lower edge of the area detected as the sliding plate end face 210 as areas that are not part of the boat 22. The over-detection removal unit 14 deletes information about these identified areas (information such as the contour of the area and the coordinates that identify pixels inside the area) from the detection results. Figure 7 shows the over-detected areas. The over-detection removal unit 14 deletes such over-detected areas (over-detection a, over-detection b) from the detection results. The over-detection removal unit 14 outputs the detection result with the over-detection area removed to the measurement location identification unit 15.

[0029] Figure 8 is a third figure showing an overview of the processing of an information processing device according to one embodiment of the present disclosure. Figure 9 is a fourth figure showing an overview of the processing of an information processing device according to one embodiment of the present disclosure. Figure 10 is a fifth figure showing an overview of the processing of an information processing device according to one embodiment of the present disclosure. Figure 11 is a sixth figure showing an overview of the processing of an information processing device according to one embodiment of the present disclosure.

[0030] An example where the sliding plate 21 is a divided sliding plate will be explained using Figures 8 to 11. In the detection results, the upper edge of the sliding plate end face 210 is represented by reference numeral 211, the lower edge by reference numeral 212, the right edge by reference numeral 213, and the left edge by reference numeral 214. The upper edge 211 of the sliding plate end face 210 may be defined as the contour of the sliding plate end face 210 that is substantially parallel to the lower edge 212 of the sliding plate end face 210. The upper edge 221 of the boat-shaped end face 220 is represented by reference numeral 221, and the lower edge by reference numeral 222. The upper edge 221 of the boat-shaped end face 220 may be defined as the contour of the boat-shaped end face 220 that is substantially parallel to the lower edge 222 of the boat-shaped end face 220. The measurement location identification unit 15 acquires the coordinates of each pixel of the region of the boat-shaped end face 220 and the contour of that region included in the detection results. Based on these coordinates, the measurement location identification unit 15 sequentially identifies perpendicular derivation points 31 at predetermined intervals on the upper longitudinal edge 221 of the rectangular area indicated by the boat end face 220 (step S109). When the first perpendiculars 32 derived from each of the perpendicular derivation points 31 are set at predetermined intervals, the measurement location identification unit 15 identifies the first intersection points 33 where the lower edge 212 of the sliding plate end face 210 intersects with the first perpendiculars 32 (step S110).

[0031] The measurement location identification unit 15 identifies the contour of the sliding plate end face 210 that is substantially parallel to the lower edge 212 as the upper edge 211. The measurement location identification unit 15 identifies the width H of the lower edge 212 and the upper edge 211 which is substantially parallel to it in the contour of the sliding plate end face 210, and identifies the first intersection 33 located at a width corresponding to the width H of the upper edge 211 (step S111). Now, in Figure 9, seven second perpendicular lines 34 are shown in the region of the central sliding plate end face 210 of the three trapezoidal sliding plate end faces 210 arranged horizontally. Of these seven second perpendiculars 34, the leftmost and rightmost second perpendiculars 34 are derived from a first intersection point 33 that is not located within the width H range where the upper edge 211 is approximately parallel to the lower edge 212 in the trapezoidal shape of the sliding plate end face 210 (the second perpendiculars 34 marked with an "x" in Figure 9). Therefore, these second perpendiculars 34 are excluded from processing.

[0032] The measurement location identification unit 15 sets a second perpendicular line 34 (the second perpendicular line 34 marked with a "○" in Figure 9) extending from the identified first intersection point 33 to the vicinity of the upper edge 211 of the sliding plate end face 210 (step S112). Here, the upper edge 211 of the sliding plate end face 210 is calculated by linear approximation using the contour of the sliding plate end face 210 and the identified multiple coordinates. However, in this process, the linear data of the linearly approximated upper edge 211 is not used, and instead, the coordinate tables of the contour of the sliding plate end face 210 used to calculate that linear data are used.

[0033] Figure 10 shows an image in which the contours of the sliding plate end face 210 and the boat-shaped end face 220, the lower edge 212 of the sliding plate end face 210, the upper edge 221 and lower edge 222 of the boat-shaped end face 220, the first perpendicular line 32, and the second perpendicular line 34 are superimposed on the image.

[0034] Now, as shown in Figure 11, assume that there are multiple coordinates 35 of the contour used to calculate the upper edge 211 of the sliding plate end face 210 near the perpendicular line 34. The measurement location identification unit 15 identifies two of these coordinates 35 that are close to the second perpendicular line 34 (step S113). The measurement location identification unit 15 identifies the intersection point of the line 36 connecting these two coordinates and the second perpendicular line 34 as the second intersection point 37, which indicates the upper contour of the sliding plate end face 210 (step S114). For convenience, Figure 11 shows an approximate line of the upper edge 211 of the sliding plate end face 210. The position of the intersection point of the approximate line of the upper edge 211 and the second perpendicular line 34 is slightly different from the identified second intersection point 37. This disclosure does not use an approximate straight line of the upper edge 211, but instead uses the coordinates of the contour of the upper edge 211 of the sluice plate end face 210 directly identified from the image to identify the intersection with the second perpendicular line 34, thereby enabling the identification of points indicating the contour on the second perpendicular line 34 with greater accuracy. The measurement location identification unit 15 outputs the combination of the identified second intersection point 37 and the first intersection point 33 corresponding to the second intersection point 37 to the thickness calculation unit 16.

[0035] The second perpendicular 34 mentioned above may be replaced with the first perpendicular 32. In other words, the first perpendicular 32 located within the width H range may be identified, and the intersection point of that first perpendicular 32 and the line connecting two neighboring coordinates indicating the upper contour of the sliding plate end face 210 may be identified as the second intersection point 37. Also, although the perpendicular derivation point 31 was set at the upper edge 221 of the boat end face 220, it may also be set at the lower edge 222 of the boat end face 220.

[0036] This process is an example of a process in which the measurement location identification unit 15 calculates the intersection point of a straight line connecting a perpendicular line between either the upper edge 221 of the support member end face (boat end face 220) or the lower edge 222 of the support member end face (boat end face 220) and two points from the point cloud representing the upper edge 211 of the sliding plate end face 210 that are closest to the second perpendicular line 34, and then identifies the shortest distance between the upper edge 211 of the sliding plate end face 210 and the lower edge 212 of the sliding plate end face 220 based on that intersection point.

[0037] Here, the measurement location identification unit 15 may set the first perpendicular line 32 and the second perpendicular line 34 in the image using other methods, and use the first perpendicular line 32 and the second perpendicular line 34 to identify the first intersection point 33 where the lower edge 212 of the sliding plate end face 210 intersects with the first perpendicular line 32, and the second intersection point 37 where the upper edge 211 of the sliding plate end face 210 intersects with the second perpendicular line 34. For example, the measurement location identification unit 15 displays the detection results for the region of the sliding plate end face 210 and the region of the boat end face 220 on the display. The detection results may include the upper edge 221, lower edge 222, etc., of the boat end face 220. The measurement location identification unit 15 outputs the input screen of the first perpendicular line 32 and the second perpendicular line 34 to the display of the information processing device 1. The input screen may include, for example, input fields for coordinates in the image (coordinates of the start and end points of the first perpendicular 32 and the second perpendicular 34), and input fields for the rotation angle of the first perpendicular 32 and the second perpendicular 34 in the image, with the vertical axis set to 0°. The user operates the information processing device 1 to input coordinates and rotation angles to set the display settings of the first perpendicular 32 and the second perpendicular 34 in the image, and also changes the coordinates and rotation angles to set the display settings of the first perpendicular 32 and the second perpendicular 34 perpendicular to the upper edge 221 and the lower edge 222 in the image displayed on the screen. The user also operates the information processing device 1 to similarly set the display settings of multiple first perpendiculars 32 and second perpendiculars 34 as shown in Figures 8 and 9. The measurement location identification unit 15 may then identify the first intersection 33 and the second intersection 37 according to the processing from step S110 as described above, based on the first perpendicular 32 and the second perpendicular 34 set by the user.

[0038] The thickness calculation unit 16 calculates the number of pixels for the distance between the identified first intersection 33 and second intersection 37 (step S115). Note that the closer the spacing between the perpendicular derivation points 31, the more first perpendiculars 32 and second perpendiculars 34 there are, and the more distances between the first intersection 33 and second intersection 37 can be calculated, thus ensuring a sufficient number of samples. The number of pixels for the distance between the first intersection 33 and second intersection 37 represents the number of pixels corresponding to the thickness of the shimaki plate end surface 210 in the image. The thickness calculation unit 16 calculates multiple numbers of pixels for the distance between the first intersection 33 and second intersection 37 in a trapezoidal region of one shimaki plate end surface 210, and identifies the number of pixels for the distance with the smallest value as the number of pixels representing the thickness in the image of one region indicated by the trapezoid of the current shimaki plate end surface 210 (step S116). This makes it possible to calculate the number of pixels representing the thickness in the image of one region indicated by the trapezoid of the shimaki plate end surface 210 with high accuracy.

[0039] The thickness calculation unit 16 calculates third intersections 38 (Figure 8) where the lower edge 222 of the boat-shaped end face 220 intersects with each of the first perpendiculars 32 corresponding to the first intersection 33 within the specified width H range (step S117). The thickness calculation unit 16 calculates the number of pixels indicating the distance for each combination of the specified third intersection 38 and the corresponding perpendicular derivation point 31 (step S118). The thickness calculation unit 16 calculates the average of the number of pixels for that distance as the number of pixels indicating the vertical thickness value of the boat-shaped end face 220. The thickness calculation unit 16 has the actual dimensions (millimeters) of the boat-shaped end face 220 stored in advance and obtains these actual dimensions from memory or the like. The thickness calculation unit 16 calculates the thickness value (in millimeters) of each trapezoidal region of the abrasive plate end face 210 by multiplying the ratio of the number of pixels (indicating the thickness of the abrasive plate end face 210 in the image with the smallest value identified in one trapezoidal region of the abrasive plate end face 210 to the number of pixels (indicating the thickness of the hull end face 220 in the image) by the actual size (in millimeters) of the hull end face 220 (step S119). The thickness calculation unit 16 outputs the thickness value (in millimeters) of the abrasive plate end face 210 to the output unit 17. In step S116, the thickness calculation unit 16 may also calculate information on the thickness of the worn abrasive plate 21 by other methods. For example, in step S116, the thickness calculation unit 16 may calculate multiple actual distance values ​​for the distance between the first intersection 33 and the second intersection 37 in a trapezoidal region of one abrasive plate end face 210, identify the distance between the first intersection 33 and the second intersection 37 with the smallest value as the worn thickness in the image of one region represented by the trapezoid of the current abrasive plate end face 210, and output the worn thickness of each of the multiple trapezoids, as well as their average value, to the output unit 17.

[0040] The output unit 17 may output the thickness value (in millimeters) of the abrasive plate end surface 210 for each of the multiple trapezoidal regions of the abrasive plate end surface 210 to a predetermined output destination (step S120). The output unit 17 may also output to the output destination the thickness value of the abrasive plate end surface 210 with the smallest value among the thickness values ​​(in millimeters) of the abrasive plate end surface 210 for each of the multiple trapezoidal regions of the abrasive plate end surface 210. The output destination may be the monitor of the information processing device 1. The output unit 17 may also output to the output destination the difference between the thickness value (in millimeters) of the abrasive plate end surface 210 and a reference value of the thickness of the abrasive plate end surface as the amount of wear.

[0041] According to the processing of the information processing device 1 described above, the wear condition of the contact strip end surface 210 of the pantograph 2 can be determined with high accuracy.

[0042] Figure 12 is a second figure showing an image of a pantograph according to one embodiment of the present disclosure. Figure 13 is a seventh figure showing an overview of the processing of an information processing device according to one embodiment of the present disclosure. Figure 4 shows an example where the contact strip 21 constituting the pantograph 2 is a contact strip 21 called a segmented contact strip. The structure of the contact strip 21 can vary; for example, in addition to segmented contact strips, there are integrated contact strips as shown in Figure 12. A segmented contact strip (Figure 4) consists of plates with a trapezoidal cross-section in the direction of vehicle travel, connected in a direction perpendicular to the direction of vehicle travel. On the other hand, an integrated contact strip, as shown in Figure 13, has a rectangular cross-section that is long in a direction perpendicular to the direction of vehicle travel. The information processing device 1 may set perpendicular derivation points 31 at predetermined intervals on the upper edge of the shoe end face 220 that supports the integrated contact strip from below, and similarly set the first perpendicular 32, the first intersection 33, the second perpendicular 34, and the second intersection 37 by the same processing as described above. The information processing device may then determine the shortest distance between the first intersection 33 and the second intersection 37 of the end face 210 of the integrated contact strip as the thickness of the contact strip 21 after wear.

[0043] Figure 14 is a functional block diagram of another example of an information processing device. Figure 15 is a diagram showing the processing flow of another example of an information processing device. The information processing device 1 only needs to include a region detection means 41 and a thickness calculation means 42. The region detection means 41 detects the region of the pantograph contact strip end face in the direction of extension of the overhead wire and the region of a predetermined member located near the contact strip in an image (step S201). The thickness calculation means 42 calculates the vertical thickness of the contact strip based on the ratio of the distance in the vertical image of the contact strip end face to the distance in the image in a predetermined direction of the region of the predetermined member and the actual dimensions in a predetermined direction of the region of the predetermined member (step S202).

[0044] Although the present disclosure has been described above with reference to the embodiments, the present disclosure is not limited to the above-described embodiments. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure.

[0045] Some or all of the above embodiments can be described as follows, but are not limited thereto.

[0046] (Supplementary Note 1) In an image including a region of the end face of the sliding plate in the pantograph that contacts the overhead line in the overhead line extending direction and a region of a predetermined member located in the vicinity of the sliding plate, region detection means for detecting the region of the end face of the sliding plate and the region of the predetermined member using a region detection model; thickness calculation means for calculating the thickness of the sliding plate in the vertical direction based on a ratio between a distance in the vertical direction in the image of the end face of the sliding plate and a distance in a predetermined direction in the image of the region of the predetermined member, and a real dimension value in the predetermined direction in the region of the predetermined member. An information processing apparatus comprising the above.

[0047] (Supplementary Note 2) The region detection means detects the end face of the support member in the overhead line extending direction as the region of the predetermined member in the image including the end face of the support member that supports the sliding plate from below; the thickness calculation means calculates the thickness of the sliding plate in the vertical direction based on a ratio between a distance in the vertical direction in the image of the end face of the sliding plate and a distance in the vertical direction in the image of the end face of the support member, and a real dimension value in the vertical direction in the region of the end face of the support member. The information processing apparatus according to Supplementary Note 1.

[0048] (Supplementary Note 3) Position specifying means for specifying, as the distance in the vertical direction in the image of the end face of the sliding plate, the shortest distance between the upper edge and the lower edge of the end face of the sliding plate; the thickness calculation means calculates the thickness of the sliding plate in the vertical direction based on a ratio between a distance in the vertical direction in the image of the end face of the sliding plate and a distance in the vertical direction in the image of the end face of the support member, and a real dimension value in the vertical direction in the region of the end face of the support member. The information processing apparatus according to Supplementary Note 2.

[0049] (Note 4) The information processing apparatus according to Note 3, wherein the region detection means detects the region of the sliding plate end surface and the region of the predetermined member using a region detection model obtained by machine learning the region of the sliding plate end surface and the region of the predetermined member in the image in which the region of the sliding plate end surface and the region of the predetermined member are captured.

[0050] (Note 5) An information processing device according to Note 3 or Note 4, comprising an overdetection area identification means for identifying an overdetection area in machine learning inference based on the results of detecting the area of ​​the edge surface of the sliding plate and the area of ​​the predetermined member using the area detection model, wherein the thickness calculation means calculates the thickness of the sliding plate in the vertical direction using the area of ​​the edge surface of the sliding plate and the area of ​​the predetermined member after removing the overdetection area.

[0051] (Note 6) The position determination means is an information processing device according to any one of Notes 3 to 5, wherein the distance between the upper edge of the sliding plate end face and the lower edge of the sliding plate end face is the shortest distance in the vertical image of the sliding plate end face, within the range in which the upper edge of the sliding plate end face is parallel to the upper edge of the support member end face.

[0052] (Note 7) The information processing apparatus according to any one of Notes 3 to 6, wherein the region detection means linearly approximates the lower edge of the end face of the sliding plate, the upper edge of the end face of the support member, and the lower edge of the end face of the support member, and the thickness calculation means calculates the thickness of the sliding plate in the vertical direction based on the ratio of the distance in the vertical direction image of the end face of the sliding plate and the distance in the vertical direction image of the end face of the support member, and the actual vertical dimension in the region of the end face of the support member.

[0053] (Note 8) The position determination means calculates the intersection of a perpendicular line between either the upper edge of the support member end face or the lower edge of the support member end face and a straight line connecting two points from the point cloud representing the upper edge of the sliding plate end face that are close to the perpendicular line, and based on that intersection, determines the shortest distance between the upper edge of the sliding plate end face and the lower edge of the sliding plate end face as described in Note 6 or Note 7.

[0054] (Note 9) An information processing method for detecting the region of the pantograph contact strip end face in the direction of extension of the overhead wire and the region of a predetermined member located near the contact strip in an image including the region of the pantograph contact strip end face and the region of the predetermined member using a region detection model, and calculating the thickness of the contact strip in the vertical direction based on the ratio of the distance in the image of the pantograph contact strip end face in the vertical direction and the distance in the image of the region of the predetermined member in a predetermined direction and the actual dimensions of the region of the predetermined member in the predetermined direction.

[0055] (Note 10) The information processing method according to Note 9, wherein in the image including the end face of the support member in the direction of extension of the overhead wire of the support member that supports the sliding plate from below, the end face of the support member is detected as the region of the predetermined member, and the thickness of the sliding plate in the vertical direction is calculated based on the ratio of the distance in the image in the vertical direction of the sliding plate end face to the distance in the image in the vertical direction of the support member end face and the actual vertical dimensions in the region of the support member end face.

[0056] (Note 11) The information processing method according to Note 10, wherein the distance between the upper edge and the lower edge of the sliding plate end face is identified as the distance in the vertical direction of the image of the sliding plate end face, and the thickness of the sliding plate in the vertical direction is calculated based on the ratio of the distance in the vertical direction of the image of the sliding plate end face to the distance in the vertical direction of the image of the support member end face, and the actual vertical dimension in the region of the support member end face.

[0057] (Note 12) The information processing method according to Note 11, wherein the region of the sliding plate end surface and the region of the predetermined member are detected using a region detection model obtained by machine learning the region of the sliding plate end surface and the region of the predetermined member in the image in which the region of the sliding plate end surface and the region of the predetermined member are captured.

[0058] (Note 13) The information processing method according to Note 11 or Note 12, which involves identifying an over-detection region in machine learning inference based on the results of detecting the region of the sliding plate end face and the region of the predetermined member using the region detection model, and calculating the vertical thickness of the sliding plate using the region of the sliding plate end face and the region of the predetermined member after removing the over-detection region.

[0059] (Note 14) The information processing method according to any one of Notes 11 to 13, wherein, in the range where the upper edge of the sliding plate end face is parallel to the upper edge of the support member end face, the shortest distance between the upper edge of the sliding plate end face and the lower edge of the sliding plate end face is calculated as the distance in the vertical direction of the image of the sliding plate end face.

[0060] (Note 15) An information processing method according to any one of Notes 11 to 14, wherein the lower edge of the end face of the sliding plate, the upper edge of the end face of the support member, and the lower edge of the end face of the support member are approximated by a straight line, and after the straight line approximation, the thickness of the sliding plate in the vertical direction is calculated based on the ratio of the distance in the vertical direction of the image of the end face of the sliding plate and the distance in the vertical direction of the image of the end face of the support member, and the actual vertical dimension in the region of the end face of the support member.

[0061] (Note 16) The information processing method according to Note 14 or Note 15, which calculates the intersection point of a straight line connecting two points from the point cloud representing the upper edge of the end face of the sliding plate that are close to the perpendicular line between the upper edge of the end face of the support member and the lower edge of the end face of the support member and calculates the intersection point of the straight line connecting two points close to the perpendicular line, and based on that intersection point, identifies the shortest distance between the upper edge of the end face of the sliding plate and the lower edge of the end face of the sliding plate.

[0062] (Note 17) A program that causes the computer of the information processing device to function as: an area detection means that detects the area of ​​the pantograph contact strip end face in the direction of extension of the overhead wire and the area of ​​a predetermined member located near the contact strip in an image including the area of ​​the pantograph contact strip end face in the direction of extension of the overhead wire and the area of ​​a predetermined member located near the contact strip, using an area detection model; and a thickness calculation means that calculates the thickness of the contact strip in the vertical direction based on the ratio of the distance in the image of the pantograph contact strip end face in the vertical direction and the distance in the image of the area of ​​the predetermined member in a predetermined direction and the actual dimensions of the area of ​​the predetermined member in the predetermined direction.

[0063] (Note 18) The program described in Note 17 wherein the region detection means detects the end face of the support member in the direction of the extension of the overhead wire in the image of the support member that supports the sliding plate from below as the region of the predetermined member, and the thickness calculation means calculates the thickness of the sliding plate in the vertical direction based on the ratio of the distance in the image of the sliding plate end face in the vertical direction to the distance in the image of the support member end face in the vertical direction and the actual vertical dimension in the region of the support member end face.

[0064] (Note 19) The program described in Note 18, wherein the distance between the upper edge and the lower edge of the sliding plate end face is used as a position identification means to identify the distance in the vertical direction of the sliding plate end face as the distance in the vertical direction of the image, and the thickness calculation means calculates the vertical thickness of the sliding plate based on the ratio of the distance in the vertical direction of the image of the sliding plate end face to the distance in the vertical direction of the image of the support member end face and the actual vertical dimensions in the region of the support member end face.

[0065] (Note 20) The region detection means is the program described in Note 19, which detects the region of the sliding plate end surface and the region of the predetermined member using a region detection model obtained by machine learning the region of the sliding plate end surface and the region of the predetermined member in the image in which the region of the sliding plate end surface and the region of the predetermined member are captured.

[0066] (Note 21) The program described in Note 19 or Note 20, wherein the region detection model is used to detect the region of the edge surface of the sliding plate and the region of the predetermined member, and the program is used to detect the overdetected region in machine learning inference, and the thickness calculation means calculates the thickness of the sliding plate in the vertical direction using the region of the edge surface of the sliding plate and the region of the predetermined member after removing the overdetected region.

[0067] (Note 22) The position determination means is a program described in any one of Notes 19 to 21, which calculates the distance in the vertical image of the sliding plate end face of the smallest length between the upper edge of the sliding plate end face and the lower edge of the sliding plate end face in the range where the upper edge of the sliding plate end face is parallel to the upper edge of the support member end face.

[0068] (Note 23) The program described in any one of Notes 19 to 22, wherein the region detection means linearly approximates the lower edge of the end face of the sliding plate, the upper edge of the end face of the support member, and the lower edge of the end face of the support member, and the thickness calculation means calculates the thickness of the sliding plate in the vertical direction based on the ratio of the distance in the vertical direction image of the end face of the sliding plate and the distance in the vertical direction image of the end face of the support member, and the actual vertical dimensions in the region of the end face of the support member.

[0069] (Note 24) The position determination means is the program described in Note 22 or Note 23, which calculates the intersection point of a straight line connecting a perpendicular line between either the upper edge of the end face of the support member or the lower edge of the end face of the support member and two points from the point cloud representing the upper edge of the end face of the sliding plate that are close to the perpendicular line, and determines the shortest distance between the upper edge of the end face of the sliding plate and the lower edge of the end face of the sliding plate based on that intersection point.

[0070] This application claims priority based on Japanese Patent Application No. 2024-191011, filed on 30 October 2024, and incorporates all of its disclosures herein.

[0071] This disclosure may be applied to information processing devices, information processing methods, and recording media.

[0072] 1... Information processing device 2... Pantograph 3... Imaging device 4... Light 5... PoE hub 11... Control unit 12... Area detection unit 13... Linear approximation unit 14... False detection removal unit 15... Measurement location identification unit 16... Thickness calculation unit 17... Output unit

Claims

1. An information processing device comprising: an area detection means for detecting the area of ​​the pantograph contact strip end face in the direction of extension of the overhead wire and the area of ​​a predetermined member located near the contact strip in an image including the area of ​​the pantograph contact strip end face in the direction of extension of the overhead wire and the area of ​​a predetermined member located near the contact strip, using an area detection model; and a thickness calculation means for calculating the thickness of the contact strip in the vertical direction based on the ratio of the distance in the image of the pantograph contact strip end face in the vertical direction and the distance in the image of the area of ​​the predetermined member in a predetermined direction and the actual dimensions of the area of ​​the predetermined member in the predetermined direction.

2. The information processing apparatus according to claim 1, wherein the area detection means detects the end face of the support member in the direction of the extension of the overhead wire in the image of the support member that supports the sliding plate from below as the area of ​​the predetermined member, and the thickness calculation means calculates the thickness of the sliding plate in the vertical direction based on the ratio of the distance in the image of the sliding plate end face in the vertical direction to the distance in the image of the support member end face in the vertical direction and the actual vertical dimension in the area of ​​the support member end face.

3. The information processing apparatus according to claim 2, comprising: position identification means for identifying the shortest distance between the upper edge and the lower edge of the end face of the sliding plate as the distance in the vertical direction of the end face of the sliding plate, wherein the thickness calculation means calculates the vertical thickness of the sliding plate based on the ratio of the distance in the vertical direction of the end face of the sliding plate in the vertical direction of the image and the distance in the vertical direction of the end face of the support member in the image and the actual vertical dimensions in the region of the end face of the support member.

4. The information processing apparatus according to claim 1, wherein the region detection means detects the region of the sliding plate end surface and the region of the predetermined member using a region detection model obtained by machine learning the region of the sliding plate end surface and the region of the predetermined member in the image in which the region of the sliding plate end surface and the region of the predetermined member are captured.

5. An information processing apparatus according to claim 3 or 4, comprising an overdetection region identification means for identifying an overdetection region in machine learning inference based on the results of detecting the region of the edge surface of the sliding plate and the region of the predetermined member using the region detection model, wherein the thickness calculation means calculates the thickness of the sliding plate in the vertical direction using the region of the edge surface of the sliding plate and the region of the predetermined member after removing the overdetection region.

6. The information processing apparatus according to any one of claims 3 to 5, wherein the position determination means calculates the distance in the vertical image of the sliding plate end face of the smallest length between the upper edge of the sliding plate end face and the lower edge of the sliding plate end face in a range where the upper edge of the sliding plate end face is parallel to the upper edge of the support member end face.

7. The information processing apparatus according to any one of claims 3 to 6, wherein the region detection means linearly approximates the lower edge of the end face of the sliding plate, the upper edge of the end face of the support member, and the lower edge of the end face of the support member, and the thickness calculation means calculates the thickness of the sliding plate in the vertical direction after the linear approximation based on the ratio of the distance in the vertical direction image of the end face of the sliding plate and the distance in the vertical direction image of the end face of the support member, and the actual vertical dimension in the region of the end face of the support member.

8. The information processing apparatus according to claim 6 or 7, wherein the positioning means calculates the intersection point of a straight line connecting two points from the point cloud representing the upper edge of the end face of the support member and the lower edge of the end face of the support member and the perpendicular line connecting the perpendicular line, and based on that intersection point, determines the shortest distance between the upper edge of the end face support member, the perpendicular line connecting the perpendicular line connecting two points from the point cloud representing the upper edge of the end face of the end face of the sliding plate and the shortest distance between the upper edge of the end face of the end face of the sliding plate and the lower edge of the end face of the end face of the sliding plate.

9. An information processing method for detecting the region of the pantograph contact strip end face in the direction of extension of the overhead wire and the region of a predetermined member located near the contact strip in an image, using a region detection model, and calculating the thickness of the contact strip in the vertical direction based on the ratio of the distance in the image of the contact strip end face in the vertical direction and the distance in the image of the region of the predetermined member in a predetermined direction and the actual dimensions of the region of the predetermined member in the predetermined direction.

10. The information processing method according to claim 9, wherein in the image including the end face of the support member in the direction of extension of the overhead wire of the support member that supports the sliding plate from below, the end face of the support member is detected as the region of the predetermined member, and the thickness of the sliding plate in the vertical direction is calculated based on the ratio of the distance in the image in the vertical direction of the sliding plate end face to the distance in the image in the vertical direction of the support member end face and the actual vertical dimensions in the region of the support member end face.

11. The information processing method according to claim 10, wherein the distance between the upper edge and the lower edge of the sliding plate end face is identified as the distance in the vertical direction of the image of the sliding plate end face, and the vertical thickness of the sliding plate is calculated based on the ratio of the distance in the vertical direction of the image of the sliding plate end face to the distance in the vertical direction of the image of the support member end face, and the actual vertical dimension in the region of the support member end face.

12. The information processing method according to claim 11, wherein the region of the sliding plate end surface and the region of the predetermined member are detected using a region detection model obtained by machine learning the region of the sliding plate end surface and the region of the predetermined member in the image in which the region of the sliding plate end surface and the region of the predetermined member are captured.

13. The information processing method according to claim 11 or claim 12, which involves identifying an over-detection region in machine learning inference based on the results of detecting the region of the sliding plate end face and the region of the predetermined member using the region detection model, and calculating the thickness of the sliding plate in the vertical direction using the region of the sliding plate end face and the region of the predetermined member after removing the over-detection region.

14. The information processing method according to any one of claims 11 to 13, wherein, in the range where the upper edge of the sliding plate end face is parallel to the upper edge of the support member end face, the distance between the upper edge of the sliding plate end face and the lower edge of the sliding plate end face is calculated as the distance in the vertical image of the sliding plate end face.

15. An information processing method according to any one of claims 11 to 14, wherein the lower edge of the end face of the sliding plate, the upper edge of the end face of the support member, and the lower edge of the end face of the support member are approximated by a straight line, and after the straight line approximation, the thickness of the sliding plate in the vertical direction is calculated based on the ratio of the distance in the vertical direction of the image of the end face of the sliding plate and the distance in the vertical direction of the image of the end face of the support member, and the actual vertical dimension in the region of the end face of the support member.

16. The information processing method according to claim 14 or 15, which involves calculating the intersection point of a straight line connecting two points from the point cloud representing the upper edge of the end face of the support member and the lower edge of the end face of the support member with the perpendicular line, and determining the shortest distance between the upper edge of the end face of the end face of the sliding plate and the lower edge of the end face of the sliding plate based on that intersection point.

17. A recording medium storing a program that causes the computer of an information processing device to function as: an area detection means that detects the area of ​​the pantograph contact strip end face in the direction of extension of the overhead wire and the area of ​​a predetermined member located near the contact strip using an area detection model in an image including the area of ​​the pantograph contact strip end face in the direction of extension of the overhead wire and the area of ​​a predetermined member located near the contact strip; and a thickness calculation means that calculates the thickness of the contact strip in the vertical direction based on the ratio of the distance in the image of the pantograph contact strip end face in the vertical direction and the distance in the image of the area of ​​the predetermined member in a predetermined direction and the actual dimensions of the area of ​​the predetermined member in the predetermined direction.

18. The recording medium according to claim 17, wherein the region detection means detects the end face of the support member in the direction of the extension of the overhead wire in the image of the support member that supports the sliding plate from below as the region of the predetermined member, and the thickness calculation means calculates the thickness of the sliding plate in the vertical direction based on the ratio of the distance in the image of the sliding plate end face in the vertical direction to the distance in the image of the support member end face in the vertical direction and the actual vertical dimension in the region of the support member end face.

19. The recording medium according to claim 18, wherein the shortest distance between the upper edge and the lower edge of the sliding plate end face functions as a position identification means for identifying the distance in the vertical direction of the sliding plate end face as the distance in the vertical direction of the image, and the thickness calculation means calculates the vertical thickness of the sliding plate based on the ratio of the distance in the vertical direction of the image of the sliding plate end face to the distance in the vertical direction of the image of the support member end face and the actual vertical dimensions in the region of the support member end face.

20. The recording medium according to claim 19, wherein the region detection means detects the region of the sliding plate end surface and the region of the predetermined member using a region detection model obtained by machine learning the region of the sliding plate end surface and the region of the predetermined member in the image in which the region of the sliding plate end surface and the region of the predetermined member are captured.

Images