Timing chain inspection device and timing chain inspection method

Abstract

To provide a timing chain inspection device and a timing chain inspection method that can easily inspect timing chain piece misalignment.SOLUTION: As an example, a timing chain inspection device captures an image of a sprocket M1 along with a timing chain M2 attached to the sprocket M1, detects each center position of the sprocket M1, the sprocket characteristic part M3, and the timing chain characteristic part M4 (first center position C1, second center position C2, third center position C3) in the captured image, and determines the presence or absence of timing chain M2 piece misalignment based on the positional relationship of each center position.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to a timing chain inspection device and a timing chain inspection method for inspecting a timing chain attached to a sprocket for misalignment. [Background technology]

[0002] In production lines for engines such as automobiles, there is an inspection process for checking whether timing chains attached to sprockets are misaligned. Conventionally, timing chain misalignment has been inspected by, for example, visually checking the alignment of the sprocket keyways and the timing chain's color points. However, the sprocket keyways and the timing chain's color points are located in positions that are invisible unless the worker peers in, which poses a risk to the work. Therefore, there has been a demand for an automated inspection method that does not require visual inspection by workers.

[0003] For example, Patent Document 1 describes a chain inspection device that uses image analysis to detect the rotation state of a chain pin. This chain inspection device acquires an image including a rivet pin with a convex area having a straight edge formed on its head, and analyzes areas of reduced brightness adjacent to the straight edge in the acquired image to detect the rotation state of the rivet pin. However, while this technology can estimate the rotation angle of each element, in order to estimate the positional relationship between two elements, it is necessary to estimate the angle for each element and calculate the difference between the two, which results in a complex system.

[0004] Furthermore, for example, Patent Document 2 describes a rotation angle detection method that detects the rotation angle by pattern matching the rotation state of an object using AI image analysis. However, as with existing AI image analysis, for example, when simply learning good and bad images, it is necessary to learn various images with 360° phase differences, which poses a problem in that a large amount of training data is required. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Publication No. 2022-52513 [Patent Document 2] Japanese Patent Application Publication No. 8-263659 Summary of the Invention [Problem to be solved by the invention]

[0006] SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned problems, and has as its object to provide a timing chain inspection device and a timing chain inspection method that can easily inspect a timing chain for misalignment. [Means for solving the problem]

[0007] The timing chain inspection device of the present invention comprises an imaging means for imaging the sprocket together with the timing chain attached to the sprocket, a center detection means for detecting the center position of the sprocket as a first center position, the center position of a characteristic portion of the sprocket as a second center position, and the center position of a characteristic portion of the timing chain as a third center position in the image obtained by the imaging means, and a discrimination means for determining whether or not there is a misalignment of the timing chain based on the positional relationship between the first center position, the second center position, and the third center position.

[0008] The timing chain inspection method of the present invention includes an imaging step of imaging the sprocket together with the timing chain attached to the sprocket; a center detection step of detecting, in the image obtained by the imaging step, the center position of the sprocket as a first center position, the center position of the sprocket's characteristic portion as a second center position, and the center position of the timing chain's characteristic portion as a third center position; and a determination step of determining whether or not there is a timing chain misalignment based on the positional relationship between the first center position, the second center position, and the third center position. [Effects of the Invention]

[0009] According to the present invention, the center positions of the sprocket, sprocket characteristic portions, and timing chain characteristic portions are detected, and the presence or absence of timing chain misalignment is determined based on their relative positions. Therefore, even if the positions of the sprocket characteristic portions and timing chain characteristic portions are misaligned depending on the rotational state of the sprocket, the presence or absence of timing chain misalignment can be easily determined from their relative positional relationships. Therefore, for example, the sprocket, sprocket characteristic portions, and timing chain characteristic portions can be detected and their center positions calculated using a learning model that has individually learned their respective characteristics. This eliminates the need to prepare a large number of training images that change the rotational state of the sprocket, thereby reducing the number of training images required. Furthermore, the timing chain inspection device can be configured simply with a photographing means such as a camera, and a computer or other device that functions as a center detection means and a discrimination means. [Brief explanation of the drawings]

[0010] [Figure 1] 1 is a diagram illustrating a configuration of a timing chain inspection device according to an embodiment of the present invention. [Figure 2] 1. FIG. 4 is a diagram illustrating a procedure for detecting the center positions of a sprocket, a sprocket characteristic portion, and a timing chain characteristic portion from a photographed image by the center detection means shown in FIG. [Figure 3] 1. FIG. 4 is a diagram illustrating a method for determining whether or not there is a timing chain misalignment based on the positional relationship between the first center position, the second center position, and the third center position, using the determining means shown in FIG. [Figure 4] 2 is a diagram illustrating an example of the hardware configuration of the extraction means and the determination means illustrated in FIG. 1. FIG. [Figure 5] 1 is a flowchart illustrating a timing chain inspection method according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

[0012] 1 shows the configuration of a timing chain inspection device 1 according to one embodiment of the present invention. This timing chain inspection device 1 inspects timing chains attached to sprockets for misalignment, and can be used, for example, on an automobile engine production line. In this embodiment, an example will be described in which a timing chain M2 is inspected for misalignment on a part M in an automobile engine production line, where the timing chain M2 is stretched across three sprockets M1.

[0013] The timing chain inspection device 1 includes, for example, an imaging means 10 that images the sprocket M1 together with the timing chain M2 attached to the sprocket M1, a center detection means 20 that detects the center positions of the sprocket M1, the sprocket characteristic portion M3, and the timing chain characteristic portion M4 in the image obtained by the imaging means 10, a discrimination means 30 that determines whether or not there is a misalignment of the timing chain M2 based on the positional relationship of the center positions detected by the center detection means 20, and a display means 40 that displays the discrimination results by the discrimination means 30, etc.

[0014] The photographing means 10 includes a camera 11, such as a CCD camera, and a control / recording means 12 that controls photographing by the camera 11 and records the photographed images. The camera 11 is fixedly disposed so as to photograph the side of each sprocket M1 in the rotational axis direction and the attached timing chain M2 for engine parts M transported on the production line. The photographing range of the camera 11 is preferably set so as to include, for example, each sprocket M1 and the portion of the timing chain M2 positioned around it, either individually or collectively, corresponding to each sprocket M1.

[0015] The control and recording means 12 can be configured, for example, by a computer, and is configured to function as the control and recording means 20 by executing a program. The camera control and recording means 12 preferably includes, for example, an object detection means 12A connected to the camera 11, which detects the sprocket M1 from the image captured by the camera 11, and a recording means 12B, which captures and records an image when the sprocket M1 detected by the object detection means 12A moves to a predetermined position in the image, for example, the center of the image in the movement direction. This configuration allows automatic capture of images triggered by an image. The recording means 12B may capture images individually for each sprocket M1, or may capture multiple images at once.

[0016] The object detection means 12A can use, for example, known object detection technology. For example, it is preferable to use an object detection learning model that learns features for each sprocket M1 by performing machine learning using training images that are positive examples of the sprocket M1 to be detected. The object detection learning model can be generated by an object detection learning model generation means that prepares multiple training images that are positive examples by cutting out relevant portions from photographed images of the sprocket M1 and extracts features using deep learning. For deep learning, it is preferable to use a convolutional neural network (CNN).

[0017] FIG. 2 shows the procedure by which the center detection means 20 detects the center positions of the sprocket M1, sprocket-characteristic portion M3, and timing-chain-characteristic portion M4 from a captured image. The center detection means 20 can be configured, for example, by a computer, and is configured to function as the center detection means 20 by executing a program. The center detection means 20 is preferably configured, for example, to detect the center position of the sprocket M1 as a first center position C1, the center position of the sprocket-characteristic portion M3 as a second center position C2, and the center position of the timing-chain-characteristic portion M4 as a third center position C3 in the captured image obtained by the imaging means 10. Examples of the sprocket-characteristic portion M3 include a key groove, and examples of the timing-chain-characteristic portion M4 include a color point. For ease of understanding, in FIG. 2, the sprocket-characteristic portion M3 is indicated by a diagonal line in the lower left corner, and the timing-chain-characteristic portion M4 is indicated by a shaded area.

[0018] Specifically, the center detection means 20 preferably includes, for example, an image cutting means 21 that cuts out a necessary portion from the captured image obtained by the photographing means 10 as a cut-out image, a characteristic position detection means 22 that detects the sprocket M1, the sprocket characteristic portion M3, and the timing chain characteristic portion M4 from the cut-out image cut out by the image cutting means 21, and a center calculation means 23 that calculates the respective center positions of the sprocket M1, the sprocket characteristic portion M3, and the timing chain characteristic portion M4 detected by the characteristic position detection means 22, i.e., the first center position C1, the second center position C2, and the third center position C3. Note that Figure 2(A) is a conceptual diagram showing an example of a cropped image cropped by the image cropping means 21, Figure 2(B) is a conceptual diagram in which the area detected as the sprocket M1 in the cropped image is indicated by a dashed line frame F1, the area detected as the sprocket characteristic portion M3 is indicated by a one-dot dashed line frame F2, and the area detected as the timing chain characteristic portion M4 is indicated by a two-dot dashed line F3, and Figure 2(C) is a conceptual diagram showing the center positions (first center position C1, second center position C2, third center position C3) of the detected sprocket M1, sprocket characteristic portion M3, and timing chain characteristic portion M4 calculated from these.

[0019] The feature position detection means 22 can use known object detection technology. For example, it is preferable to use a feature position detection learning model in which individual features are learned by performing machine learning using training images that serve as positive examples of the sprocket M1, sprocket characteristic portion M3, and timing chain characteristic portion M4 to be detected. The feature position detection learning model can be generated by a feature position detection learning model generation means that extracts features through deep learning by preparing multiple training images that serve as positive examples by cutting out relevant portions from photographed images of the sprocket M1, sprocket characteristic portion M2, or timing chain characteristic portion M3. A convolutional neural network is preferably used for deep learning. The center calculation means 23 can use known processing technology and is configured, for example, to calculate center coordinates as the center position. Specifically, for example, it is preferable to configure the sprocket M1, sprocket characteristic portion M2, and timing chain characteristic portion M3 detected by the feature position detection means 22 to be individually surrounded by frames such as squares, and to calculate the center coordinates of the portions surrounded by those frames.

[0020] That is, the center detection means 20 is preferably configured to detect, for example, the sprocket M1, the sprocket characteristic portion M3, and the timing chain characteristic portion M4 using a learning model that has learned the characteristics of each individually, and to calculate the first center position C1, the second center position C2, and the third center position C3.

[0021] The determination means 30 can be configured, for example, by a computer, and is configured to function as the determination means 30 by executing a program. The determination means 30 is preferably configured to determine whether or not there is misalignment of the timing chain M2 based on, for example, the positional relationship of the first center position C1, the second center position C2, and the third center position C3 relative to the first center position C1. This is because even if the positions of the sprocket characteristic portion M3 and the timing chain characteristic portion M4 are misaligned depending on the rotation state of the sprocket M1, it is possible to easily determine whether or not there is misalignment from the relative positional relationship.

[0022] The discrimination means 30 is preferably configured to calculate, for example, as shown in FIG. 3, the angle θ1 between a first line connecting the first center position C1 and the second center position C2 and a second line connecting the first center position C1 and the third center position C3, or the angle θ2 between a first vector extending from the first center position C1 to the second center position C2 and a second vector extending from the first center position C1 to the third center position C3, and to determine whether or not there is a misalignment of the timing chain M2 from the angle θ1 between the two obtained lines or the angle θ2 between the two vectors. Specifically, it is preferable that the discrimination means 30 has, for example, an angle calculation means 31 that calculates the angle θ1 between the first line and the second line or the angle θ2 between the first vector and the second vector as the positional relationship between the first center position C1, the second center position C2, and the third center position C3, and a threshold comparison means 32 that compares the angle calculated by the angle calculation means 31, i.e., the angle θ1 between the first line and the second line or the angle θ2 between the first vector and the second vector, with a threshold to determine whether or not there is a frame misalignment in the timing chain M2.

[0023] The threshold value varies depending on the sprocket M1 and can be set arbitrarily. For example, if the angle calculated by the angle calculation means 31 is angle θ1, the threshold value is set in the range of 1° to 6°, and the threshold value comparison means 32 is configured to determine that there is no piece misalignment if the angle θ1 is within the threshold value range and that there is piece misalignment if the angle is outside the threshold value range. Also, if the angle calculated by the angle calculation means 31 is angle θ2, the threshold value is set in the range of 174° to 179°, and the threshold value comparison means 32 is configured to determine that there is no piece misalignment if the angle θ2 is within the threshold value range and that there is piece misalignment if the angle is outside the threshold value range.

[0024] The display means 40 is configured to display, for example, the presence or absence of timing chain misalignment determined by the determination means 30. Furthermore, the display means 40 may be configured to display the positional relationship between the first center position C1, the second center position C2, and the third center position C3, which are the basis for the determination.

[0025] 4 shows an example of the hardware configuration of the control and recording means 12, the center detection means 20, and the discrimination means 30. The control and recording means 12, the center detection means 20, and the discrimination means 30 each include, for example, a CPU (Center Processing Unit) 51, a ROM (Read Only Memory) 52, a RAM (Random Access Memory) 53, a HDD (Hard Disk Drive) 54, and an operation interface (operation I / F) 55. The CPU 51 executes various processes in accordance with various programs stored in the ROM 52 or various programs loaded from the HDD 54 to the RAM 53. The RAM 53 also stores data necessary for the CPU 51 to execute various processes, as appropriate. The HDD 54 stores various data.

[0026] This timing chain inspection device 1 is used, for example, as follows: Figure 5 shows the flow of a timing chain inspection method using the timing chain inspection device 1. First, as a preparatory step, for example, an object detection learning model is generated by the object detection learning model generation means, and a feature position detection learning model is generated by the feature position detection learning model generation means (preparatory step; step S110).

[0027] Specifically, for example, in an automobile engine production line, the side of a sprocket M1, which is the detection target, is photographed by the photographing means 10, and multiple training images are prepared as positive examples by cutting out images of the sprocket M1 from the obtained photographed images, and an object detection training model is generated by deep learning.Also, for example, multiple training images are prepared as positive examples by cutting out portions of the sprocket M1, sprocket characteristic portion M3, and timing chain characteristic portion M4 from the photographed images photographed by the photographing means 10, and a feature position detection training model is generated by learning the features individually by deep learning.

[0028] Next, for example, an engine part M conveyed on a production line is inspected for misalignment of the timing chain M2. First, for example, the imaging means 10 images the sprocket M1 and the timing chain M2 attached to the sprocket M1 (imaging procedure; step S120). Specifically, for example, the engine part M conveyed on the production line is captured by the camera 11, and the object detection means 12A uses an object detection learning model to detect the sprocket M1 from the image captured by the camera 11 (object detection procedure; step S121). When the detected sprocket M1 moves to a predetermined position in the image, the recording means 12 captures and records an image (recording procedure; step S122). At this time, each sprocket M1 may be photographed individually and sequentially, or multiple sprockets M1 may be photographed together. For example, as shown in FIG. 1, if two sprockets M1 are located close to each other, the two nearby sprockets M1 may be included in a single captured image.

[0029] Next, for example, the center detection means 20 detects the center positions of the sprocket M1, sprocket characteristic portion M3, and timing chain characteristic portion M4 in the captured image obtained in the photographing procedure (center detection procedure; step S130). Specifically, for example, first, the image cropping means 21 crops a necessary portion from the captured image obtained in the photographing procedure as a cropped image (image cropping procedure; step S131 (see FIG. 2(A))). Next, for example, the feature position detection means 22 detects the sprocket M1, sprocket characteristic portion M3, and timing chain characteristic portion M4 from the cropped image cropped in the image cropping procedure using a feature position detection learning model (feature position detection procedure; step S132 (see FIG. 2(B))). The detected locations are indicated by, for example, a rectangular frame or the like. Next, for example, the center calculation means 23 calculates the coordinates of the center positions of the sprocket M1, sprocket characteristic portion M3, and timing chain characteristic portion M4 detected in the characteristic position detection procedure, i.e., the first center position C1, the second center position C2, and the third center position C3 (center calculation procedure; step S133 (see Figure 2(C))).

[0030] Next, for example, the determination means 30 determines whether or not there is a frame misalignment of the timing chain M2 based on the positional relationship between the first center position C1, the second center position C2, and the third center position C3 calculated in the center detection procedure (determination procedure; step S140). Specifically, the angle calculation means 31 calculates the angle θ1 between a first line connecting the first center position C1 and the second center position C2 and a second line connecting the first center position C1 and the third center position C3, or the angle θ2 between a first vector extending from the first center position C1 to the second center position C2 and a second vector extending from the first center position C1 to the third center position C3 (angle calculation procedure; step S141 (see FIG. 3)). The threshold comparison means 32 then compares the obtained angle θ1 between the first line and the second line or the angle θ2 between the first vector and the second vector with a threshold to determine whether or not there is a frame misalignment of the timing chain (threshold comparison procedure; step S142). Thereafter, the display means 40 displays the results of the discrimination according to the discrimination procedure (display procedure; step S150).

[0031] According to this embodiment, the center positions of the sprocket M1, the sprocket-characteristic portion M3, and the timing chain-characteristic portion M4 are detected, and the presence or absence of timing chain misalignment is determined based on their relative positions. Therefore, even if the positions of the sprocket-characteristic portion M3 and the timing chain-characteristic portion M4 are misaligned depending on the rotational state of the sprocket M1, the presence or absence of timing chain misalignment can be easily determined from their relative positions. Therefore, for example, the sprocket M1, the sprocket-characteristic portion M3, and the timing chain-characteristic portion M4 can be detected and their center positions calculated using a feature position detection learning model that individually learns the characteristics of each. This eliminates the need to prepare a large number of training images that change the rotational state of the sprocket M1, thereby reducing the number of training images required. Furthermore, the timing chain inspection device 1 can be configured simply with the imaging means 10, including the camera 11, and a computer or other device that functions as the center detection means 20 and the discrimination means 30.

[0032] Furthermore, the photographing means 10 is configured to include a camera 11 and a control and recording means 12. The control and recording means 12 detects the sprocket M1 from the image captured by the camera 11, and captures and records an image when the detected sprocket M1 moves to a predetermined position in the image, so that the image can be used as a trigger to automatically capture a photographed image. This eliminates the need for a facility-side trigger such as a limit switch, simplifying the device.

[0033] Although the present invention has been described above with reference to the embodiments, the present invention is not limited to the above embodiments and can be modified in various ways. For example, although each component has been specifically described in the above embodiments, the specific structure or shape of each component may be different, and the device may not include all of the above-described components, and may include other components. [Explanation of symbols]

[0034] 1...timing chain inspection device, 10...photographing means, 11...camera, 12...control and recording means, 12A...object detection means, 12B...recording means, 20...center detection means, 21...image cutting means, 22...characteristic position detection means, 23...center calculation means, 30...discrimination means, 31...angle calculation means, 32...threshold comparison means, 40...display means, 51...CPU, 52...ROM, 53...RAM, 54...HDD, 55...operation interface, M...part, M1...sprocket, M2...timing chain, M3...sprocket characteristic part, M4...timing chain characteristic part

Claims

1. an imaging means for imaging the sprocket and the timing chain attached to the sprocket; a center detection means for detecting the sprocket, sprocket characteristic portions, and timing chain characteristic portions from the photographed image obtained by the photographing means using a learning model in which the characteristics of each of the sprockets are individually learned, and for calculating the respective center positions of the detected sprockets, sprocket characteristic portions, and timing chain characteristic portions, with the center position of the sprocket being a first center position, the center position of the sprocket characteristic portion being a second center position, and the center position of the timing chain characteristic portion being a third center position; a determining means for determining whether or not there is a misalignment of the timing chain based on the positional relationship between the first center position, the second center position, and the third center position; A timing chain inspection device comprising:

2. 2. The timing chain inspection device according to claim 1, wherein the discrimination means calculates an angle between a first line connecting the first center position and the second center position and a second line connecting the first center position and the third center position, or an angle between a first vector extending from the first center position to the second center position and a second vector extending from the first center position to the third center position, and determines whether or not there is a misalignment of the timing chain from the angle between the two obtained lines or the angle between the two vectors.

3. a photographing step of photographing a sprocket and a timing chain attached to the sprocket; a center detection step of detecting the sprocket, sprocket characteristic portions, and timing chain characteristic portions from the photographed images obtained by the photographing step using a learning model in which the characteristics of each of the sprockets are individually learned, and calculating the center positions of the detected sprockets, sprocket characteristic portions, and timing chain characteristic portions, with the center position of the sprocket being a first center position, the center position of the sprocket characteristic portions being a second center position, and the center position of the timing chain characteristic portions being a third center position; a determination step of determining whether or not there is a misalignment of the timing chain based on a positional relationship between the first center position, the second center position, and the third center position; A timing chain inspection method comprising:

Images