Roller abnormality detection system

Abstract

To provide a technique capable of determining which of a plurality of rollers has an abnormality.SOLUTION: An anomaly detection system of the present invention comprises: an imaging device for imaging multiple rollers; an acquisition unit for acquiring identification information indicating an identifier of a roller attached in advance by using a picked-up image taken by an imaging apparatus, and coordinate information indicating coordinates of a roller detected by using the picked-up image and associated with the identification information; and a determination unit for determining roller anomalies using the coordinate information.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to a roller abnormality determination system. [Background technology]

[0002] There are known technologies that can determine whether an abnormality has occurred in a conveyor. For example, Patent Document 1 discloses a technology that photographs the side of a roller when the drive rollers of multiple connected, rotating steps of a passenger conveyor come close to each other, and diagnoses whether the roller shaft is in good condition. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2010-265078 Summary of the Invention [Problem to be solved by the invention]

[0004] However, since it is not possible to distinguish between the multiple rollers of the conveyor, there is a problem in that it is not possible to determine which roller is experiencing an abnormality. [Means for solving the problem]

[0005] The present disclosure has been made to solve the above-mentioned problems, and can be realized in the following forms.

[0006] (1) According to an embodiment of the present disclosure, there is provided a roller abnormality determination system including: an imaging device that images a plurality of rollers; an acquisition unit that acquires identification information indicating identifiers of the rollers that have been assigned in advance using the images captured by the imaging device; coordinate information indicating coordinates of the rollers detected using the images and associated with the identification information; and a determination unit that determines an abnormality in the rollers using the coordinate information. According to this type of abnormality determination system, the coordinate information associated with the identification information is used to determine whether a roller is abnormal, so it is possible to determine which of the multiple rollers has an identifier associated with the roller that is experiencing the abnormality. (2) In the abnormality determination system of the above aspect, the determination unit may determine that the roller is abnormal when the size of the roller in the axial direction is equal to or smaller than a predetermined threshold value. According to this type of abnormality determination system, the determination unit determines that an abnormality has occurred when the axial size of the roller is equal to or greater than a threshold value. Therefore, if at least one wheel of the roller is missing, it can be determined that an abnormality has occurred. (3) In the abnormality determination system of the above form, the determination unit may determine that a roller between a first roller and a second roller different from the first roller is abnormal when the distance between the first roller and the second roller is equal to or greater than a predetermined threshold distance. According to this type of abnormality determination system, the determination unit determines that the roller between the first roller and the second roller is abnormal if the distance between the first roller and the second roller is equal to or greater than the threshold distance. Therefore, it is possible to detect abnormalities in rollers that are missing both wheels and are not recognized as rollers, and that have not been assigned an identifier in advance. (4) In the abnormality determination system of the above type, the determination unit may determine an abnormality in the roller by pattern matching using a determination image obtained by cutting out an area including the roller from the captured image, or by using a learning model that has undergone machine learning to determine an abnormality in the roller using the determination image. According to this type of abnormality determination system, the determination unit determines abnormalities in the roller by pattern matching using a determination image cut out from the area including the roller in the captured image, or by using the determination image and a learning model, so that abnormalities in the roller can be determined regardless of the position of the roller in the captured image or the background of the roller. (5) In the abnormality determination system of the above aspect, the determination unit may calculate an abnormality value indicating the degree of abnormality, and instruct replacement of the roller when the abnormality value is equal to or greater than a predetermined threshold abnormality value. In this type of abnormality determination system, the determination unit calculates an abnormality value that indicates the degree of abnormality, and if the abnormality value is equal to or greater than the threshold abnormality value, issues a command to replace the roller. Therefore, when roller replacement is not urgently required, excessive replacement commands can be suppressed.

[0007] The present disclosure can be realized in various forms, for example, as a determination device, a roller abnormality determination method, or the like. [Brief explanation of the drawings]

[0008] [Figure 1] FIG. 1 is an explanatory diagram illustrating a configuration of an abnormality determination system. [Figure 2] 10 is a flowchart illustrating an example of an abnormality determination process. [Figure 3] 10 is an example of a captured image. [Figure 4] 10 is another example of a captured image. DETAILED DESCRIPTION OF THE INVENTION

[0009] A. First embodiment: 1 is an explanatory diagram showing the configuration of an abnormality determination system 10 in this embodiment. The abnormality determination system 10 determines abnormalities in rollers 21 of a conveyor 20. The abnormality determination system 10 includes an imaging device 100, an identification device 200, and a determination device 300. The conveyor 20 has a plurality of rollers 21.

[0010] The imaging device 100 is capable of capturing images at a frame rate at which the same roller 21 appears in multiple frames. The imaging device 100 can capture images at, for example, 24 fps. The frame rate is a value greater than the value obtained by dividing the movement distance of the roller 21 in one second by the length of the roller 21 that can be captured by the imaging device 100 in the movement direction.

[0011] The identification device 200 is a device that identifies the roller 21 using the image captured by the imaging device 100. The identification device 200 is configured by a computer that includes an input / output interface 210, a storage unit 220, and a CPU 230. The input / output interface 210, the storage unit 220, and the CPU 230 are connected to each other so that they can communicate bidirectionally. The storage unit 220 is configured by a ROM and a RAM.

[0012] The CPU 230 executes a program pre-installed in the storage unit 220 to implement the functions of the recognition unit 231 and the coordinate detection unit 232. However, some or all of the functions of these units may be implemented by hardware circuits.

[0013] The identification unit 231 uses an image acquired from the imaging device 100 via the input / output interface 210 to identify an identifier previously assigned to each roller 21 by artificial intelligence (AI). The identification unit 231 detects the rollers 21 from the image acquired from the imaging device 100, for example, by using the artificial intelligence. Thereafter, the identification unit 231 identifies the identifier of the detected roller 21 by using the feature amount of each roller 21 stored in the storage unit 220 and identification information indicating the identifier previously assigned to each roller 21 that is associated with the feature amount of each roller 21.

[0014] The coordinate detection unit 232 uses the image acquired from the imaging device 100 via the input / output interface 210 to detect the coordinates of the roller 21 identified by the identification unit 231 in the image.

[0015] The determination device 300 is a device that determines whether there is an abnormality in the roller 21. The determination device 300 is configured by a computer that includes an input / output interface 310, a storage unit 320, and a CPU 330. The input / output interface 310, the storage unit 320, and the CPU 330 are connected to each other so as to be able to communicate bidirectionally.

[0016] The storage unit 320 is composed of a ROM and a RAM. The storage unit 320 stores a learning model 321. The learning model 321 is a learning model generated by performing machine learning using an image of a normal roller and an image of a roller in which an abnormality has occurred. The learning model 321 can determine an abnormality in the roller 21 contained in the determination image, using a determination image obtained by cutting out an area including the roller 21 from an input image captured by the imaging device 100. The learning model 321 may be stored in an external storage device such as an external hard disk or a server on the Internet.

[0017] The CPU 330 executes a program pre-installed in the storage unit 320 to implement the functions of the acquisition unit 331 and the determination unit 332. However, some or all of the functions of these units may be implemented by a hardware circuit.

[0018] The acquiring unit 331 acquires identification information and coordinate information associated with the identification information from the identification device 200 via the input / output interface 310. The identification information is information indicating an identifier of the roller 21 assigned in advance by the identification unit 231. The coordinate information is information indicating the coordinates of the roller 21 detected by the coordinate detection unit 232. In this embodiment, the acquiring unit 331 also acquires from the identification device 200 the captured image used by the identification unit 231 to assign an identifier to each roller 21.

[0019] The determination unit 332 uses the coordinate information acquired by the acquisition unit 331 to determine whether or not an abnormality has occurred in the roller 21 included in the video. The determination unit 332 makes the determination using, for example, an output result indicating whether or not an abnormality has occurred in the roller 21, output by the learning model 321 to which a determination image cut out from the captured image using the coordinate information has been input. Details of the abnormality determination will be described later. In addition, in this embodiment, the determination unit 332 calculates an abnormality value indicating the degree of abnormality. For example, the determination unit 332 assigns a weight to each roller 21 according to the degree of abnormality, and calculates the abnormality value by summing the weights of all rollers 21 in the conveyor 20. The abnormality value increases as the degree of abnormality increases. The weight according to the degree of abnormality is, for example, 1 if both wheels of the roller 21 are missing, 0.5 if one wheel of the roller 21 is missing, and 0.2 if the roller 21 has a crack or an attachment.

[0020] 2 is a flowchart showing an example of the abnormality determination process. In step S100, the identification unit 231 executes an "identification step" of identifying each roller 21 in a plurality of captured images in which a plurality of rollers 21 are captured in one frame by the imaging device 100. More specifically, the identification unit 231 uses artificial intelligence to detect the rollers 21 from the captured images obtained by dividing the video captured by the imaging device 100 into a plurality of still images, and identifies the identifier of each roller 21.

[0021] In step S110, the coordinate detection unit 232 executes a "position detection step" of detecting the coordinates of each roller 21 identified in step S100. In this embodiment, the coordinate detection unit 232 detects the coordinates of each vertex of a circumscribed rectangle of each roller 21 in the captured image. The coordinate detection unit 232 may, for example, acquire the coordinates of both ends of each roller 21 in the axial direction in the captured image.

[0022] In step S120, the acquisition unit 331 acquires identification information indicating the identifier of the roller 21 assigned by the identification unit 231 in step S100 and coordinate information associated with the identification information acquired by the coordinate detection unit 232 in step S110. In this embodiment, the acquisition unit 331 acquires the coordinates of each vertex of the circumscribed rectangle of each roller 21 in the captured image as the coordinate information.

[0023] In step S130, the determination unit 332 executes an "abnormal value calculation step" in which it determines whether or not an abnormality has occurred in the roller 21 using the coordinate information acquired in step S120, and calculates an abnormal value indicating the degree of the abnormality.

[0024] In this embodiment, the determination unit 332 determines that one wheel of the roller 21 is missing if the size of the roller 21 in the axial direction is equal to or smaller than a predetermined threshold. The size of the roller 21 in the axial direction is, for example, the difference between the maximum and minimum values ​​of the circumscribed rectangle of the roller 21 in the axial direction, as indicated by the coordinate information. The threshold may be determined experimentally or empirically, or may be determined by the determination unit 332 using the coordinate information. The determination unit 332 may, for example, determine the threshold to be equal to or smaller than 80% of the most frequent value of the sizes of all the rollers 21 in the axial direction. FIG. 3 is an example of a captured image. The captured image shown in FIG. 3 includes a first roller 21a and a second roller 21b. Hereinafter, when referring to a roller 21 without distinction, it will be referred to simply as roller 21, and when referring to a roller 21 with an identifier, it will be referred to as the n-th roller 21α (n is a natural number, and α is a symbol). In this embodiment, the size of the second roller 21b in the axial direction is equal to or smaller than the threshold. Therefore, the determining unit 332 determines that one wheel of the second roller 21b is missing.

[0025] Furthermore, if the distance between one roller 21 and another roller 21 different from the one roller 21 among the plurality of rollers 21 is equal to or greater than a predetermined threshold distance, the determination unit 332 determines that both wheels of the roller 21 between the one roller 21 and the other roller 21 are missing. The distance between the one roller 21 and the other roller 21 is, for example, the difference in the direction of travel of the roller 21 between the center of gravity of the circumscribed rectangle of the one roller 21 indicated by the coordinate information and the center of gravity of the circumscribed rectangle of the other roller 21 indicated by the coordinate information. The threshold distance may be determined experimentally or by the determination unit 332 using the coordinate information. For example, the determination unit 332 may determine the threshold distance as a distance equal to or greater than 120% of the most frequent value of the distance between adjacent rollers 21. FIG. 4 shows an example of a captured image. Below, FIG. 4 shows captured images F1 and F2. The third roller 21c is captured in the captured image F1, and the fourth roller 21d is captured in the captured image F2. Between the third roller 21c and the fourth roller 21d, there is a roller 22 that is missing both wheels. The roller 22 is not assigned an identifier by the identifier 231. In the following, an example will be described in which the coordinate information of the third roller 21c and the coordinate information of the fourth roller 21d are used.

[0026] The determination unit 332 calculates the movement speed of the roller 21 using the movement amount of the coordinates of the roller 21 and the frame rate of the identification device 200. The determination unit 332 calculates the movement speed of the roller 21 using, for example, the most frequent value of the movement amount of the coordinates of all rollers 21, or the average value of the top 10% of movement amounts having the highest frequency from the most frequent value, and the frame rate of the identification device 200.

[0027] The determination unit 332 calculates the distance between the third roller 21c and the fourth roller 21d by adding the difference between the coordinates of the third roller 21c in the captured image F1 and the coordinates of the fourth roller 21d in the captured image F2 to a value obtained by multiplying the movement speed of the roller 21 by the difference in the number of frames between the captured image F1 and the captured image F2. In this embodiment, the distance between the third roller 21c and the fourth roller 21d is equal to or less than the threshold distance. Therefore, the determination unit 332 determines that both wheels of the roller 22 between the third roller 21c and the fourth roller 21d are missing.

[0028] Furthermore, the determination unit 332 determines whether there is an abnormality in the roller 21 by using a determination image obtained by cutting out an area including the roller 21 from the captured image and the learning model 321. In this embodiment, the determination unit 332 determines whether both wheels of the roller 21 are missing, whether one wheel is missing, whether there is a crack or foreign matter attached to the roller 21, or whether there is no abnormality, by using the determination image and the learning model 321. The determination image is, for example, an image of an area obtained by expanding the circumscribed rectangle of the roller 21 in the captured image on all sides by a predetermined width.

[0029] In step S140 (shown in FIG. 2), the determination unit 332 determines whether the abnormal value calculated in step S130 is equal to or greater than a predetermined threshold abnormal value. The threshold abnormal value is, for example, a weighting value when both wheels of a roller 21 are missing. In this embodiment, the threshold abnormal value is 1. If the abnormal value is smaller than the threshold abnormal value, the determination unit 332 ends the abnormality determination process. On the other hand, if the abnormal value is equal to or greater than the threshold abnormal value, the determination unit 332 executes a "notification step" of instructing replacement of the rollers 21 via a notification device (not shown). In the notification step, the determination device 300 may also notify the number of rollers 21 that are abnormal and the details of the abnormality.

[0030] According to the abnormality determination system 10 of the present embodiment described above, the coordinate information associated with the identification information is used to determine whether or not an abnormality has occurred in a roller 21, and therefore it is possible to determine which of the rollers 21 has an identifier assigned to it among the plurality of rollers 21. This makes it possible to determine not only whether or not an abnormality has occurred in a roller 21 on the conveyor 20, but also the number of rollers 21 in which abnormalities have occurred.

[0031] Furthermore, the determining unit 332 determines that there is an abnormality when the size of the roller 21 in the axial direction is equal to or greater than a threshold value. Therefore, it can be determined that at least one wheel of the roller 21 is missing.

[0032] Furthermore, when the distance between the third roller 21c and the fourth roller 21d is equal to or greater than the threshold distance, the determination unit 332 determines that the roller 22 between the third roller 21c and the fourth roller 21d is abnormal. Therefore, it is possible to detect abnormalities in rollers 22 that are missing both wheels and are not recognized as rollers, and that are not assigned an identifier in advance.

[0033] Furthermore, the judgment unit 332 judges abnormalities in the roller 21 using a judgment image obtained by cutting out an area including the roller 21 in the captured image and the learning model 321, and therefore can judge abnormalities in the roller 21 regardless of the position of the roller 21 in the captured image or the background of the roller 21.

[0034] Furthermore, the determination unit 332 calculates an abnormality value indicating the degree of abnormality, and when the abnormality value is equal to or greater than the threshold abnormality value, issues an instruction to replace the roller 21. Therefore, when replacement of the roller 21 is not urgently required, excessive replacement instructions can be suppressed.

[0035] B. Other Embodiments: (B1) In the above-described embodiment, the acquisition unit 331 acquires the identification information and coordinate information from the identification device 200. That is, the identification device 200 realizes the functions of the identification unit 231 and the coordinate detection unit 232. However, this is not limiting, and the determination device 300 may realize at least some of the functions of the identification unit 231 and the coordinate detection unit 232.

[0036] (B2) In the above-described embodiment, the identification unit 231 identifies the identifier of each roller 21 using identification information stored in advance in the CPU 230. Without being limited to this, the identification unit 231 may identify the identifier of each roller 21 based on, for example, an identification number or a two-dimensional information code printed in advance on each roller 21. The identification unit 231 may also assign an identifier to each roller 21. The identification unit 231 may assign an identifier to each roller 21 based on, for example, a feature amount of the roller 21 detected using a learning model generated by performing machine learning using a roller image to detect the roller 21 from a video acquired from the imaging device 100.

[0037] (B3) In the above-described embodiment, the determination unit 332 determines whether there is an abnormality in the roller 21 using the learning model 321. Alternatively, the determination unit 332 may determine whether there is an abnormality in the roller 21 by pattern matching using the determination image. For example, the determination unit 332 can determine whether there is an abnormality in the roller 21 by pattern matching using a normal image when there is no abnormality stored in the storage unit 320 and the determination image.

[0038] (B4) In the above-described embodiment, the determination unit 332 calculates an abnormality value indicating the degree of abnormality, and instructs the replacement of the roller 21 when the abnormality value is equal to or greater than the threshold abnormality value. This is not limiting, and the determination unit 332 may instruct the replacement of the roller 21 when it determines that there is an abnormality in the roller 21, regardless of the magnitude of the abnormality value. Furthermore, the determination unit 332 may simply output that there is an abnormality in the roller 21.

[0039] The present disclosure is not limited to the above-described embodiments and can be realized in various configurations without departing from the spirit thereof. For example, the technical features in the embodiments corresponding to the technical features in each aspect described in the Summary of the Invention section can be appropriately replaced or combined to solve the above-described problems or achieve some or all of the above-described effects. Furthermore, if a technical feature is not described as essential in this specification, it can be appropriately deleted. [Explanation of symbols]

[0040] 10...abnormality determination system, 20...conveyor, 21, 22...roller, 21a...first roller, 21b...second roller, 21c...third roller, 21d...fourth roller, 100...imaging device, 200...identification device, 210...input / output interface, 220...storage unit, 230...CPU, 231...identification unit, 232...coordinate detection unit, 300...determination device, 310...input / output interface, 320...storage unit, 321...learning model, 330...CPU, 331...acquisition unit, 332...determination unit

Claims

1. A roller abnormality determination system, an imaging device that images the plurality of rollers; an acquisition unit that acquires identification information indicating an identifier of the roller that has been assigned in advance using an image captured by the imaging device, and coordinate information indicating coordinates of the roller detected using the captured image, the coordinate information being associated with the identification information; a determination unit that determines an abnormality in the roller using the coordinate information.

2. The abnormality determination system according to claim 1, The abnormality determination system, wherein the determination unit determines that the roller is abnormal when the size of the roller in the axial direction is equal to or smaller than a predetermined threshold value.

3. 3. The abnormality determination system according to claim 1, The abnormality determination system, wherein the determination unit determines that a roller between a first roller and a second roller different from the first roller is abnormal when the distance between the first roller and the second roller is equal to or greater than a predetermined threshold distance.

4. 3. The abnormality determination system according to claim 1, The acquisition unit acquires the captured image, The determination unit determines whether there is an abnormality in the roller by pattern matching using a determination image obtained by cutting out an area including the roller from the captured image, or by using a learning model that has undergone machine learning to determine whether there is an abnormality in the roller using the determination image.

5. 3. The abnormality determination system according to claim 1, The determination unit calculates an abnormality value indicating the degree of abnormality, and instructs replacement of the roller when the abnormality value is equal to or greater than a predetermined threshold abnormality value.

Images