Lubricating oil monitoring device

Abstract

To provide a lubricant monitor which can automatically monitor the ejecting state of a lubricant in a bearing of a power generation machine for hydroelectric power production.SOLUTION: A rotary machine 10 is supported by bearing parts 12 and 13 through a rotary shaft 11. The bearing parts 12 and 13 store a lubricant and the lubricant is constantly supplied in such a rotary part as the rotary shaft 11 or a bearing. The bearing parts 12 and 13 are provided with oil ejection confirmation windows 14 and 15. Cameras 21 and 22 take an image of the oil ejection confirmation windows 14 and 15. An image processor 23 detects the position of a crescent-shape ejection pattern and the position of the center point of the ejection formed by the ejected lubricant from images taken by the cameras 21 and 22, and determines the state of the lubricant from the position of the crescent-shape ejection pattern and the position of the ejection center point.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a lubricating oil monitoring device. [Background Art]

[0002] Conventionally, in the operation of rotating machinery such as generators and electric motors, lubricating oil is stored in a bearing housing to reduce wear of rotating parts such as rotating shafts and bearings and suppress temperature rise. The lubricating oil is constantly supplied to the rotating parts while being stirred along with the rotation of the rotating shaft within the bearing housing. Such lubricating oil can no longer exert sufficient effects due to deterioration over time and contaminants such as wear powder, so its condition is monitored and diagnosed by various methods.

[0003] Patent Document 1 discloses a technology for determining the additive concentration of lubricating oil using chromaticity data obtained based on measurement data from an optical sensor, and diagnosing deterioration of the lubricating oil based on the obtained concentration. [Prior Art Literature] [Patent Literature]

[0004] [Patent Document 1] Japanese Unexamined Patent Application Publication No. 2019-078718 [Summary of the Invention] [Problem to be Solved by the Invention]

[0005] However, in Patent Document 1, although the characteristics of the lubricating oil are evaluated using temperature, oil pressure, the concentration of particles contained in the lubricating oil, and the like in addition to the chromaticity information of the lubricating oil obtained by the optical sensor, whether the lubricating oil is sufficiently supplied to the rotating parts has not been monitored. Therefore, conventionally, the only way to check the condition of the lubricating oil is to visually observe the ejection state to an oil ejection confirmation window (described later), and there has been a demand for an automated technology.

[0006] Therefore, the object of the present invention is to provide a lubrication oil monitoring device that can automatically monitor the lubrication oil ejection state of the bearings of a hydroelectric power generator. [Means for solving the problem]

[0007] To solve the above problems, the present invention employs the following means. In other words, the lubricating oil monitoring device of the present invention is a lubricating oil monitoring device that monitors the center point of lubricating oil ejection and the crescent-shaped ejection form that is formed in an arc upward from the center point of ejection, as observed in an oil ejection confirmation window from which lubricating oil is ejected from the bearings of a hydroelectric generator, and comprises a camera that photographs the oil ejection confirmation window and an image processing device, wherein the image processing device includes an ejection shape recognition unit that detects the position of the crescent-shaped ejection form from the image of the camera, an ejection center point recognition unit that detects the position of the center point of ejection from the image of the camera, and the position of the crescent-shaped ejection form and the position of the center point of ejection The system includes a state determination unit that determines the state of the lubricating oil from the location, and the ejection shape recognition unit uses a crescent-shaped template image to identify the position of the crescent-shaped ejection shape and acquires an ejection shape image, then binarizes the ejection shape image and detects the arc of the upper contour of the crescent-shaped ejection shape by circle detection processing using Hough transform, and the ejection center point recognition unit uses a circular ejection center point template image to identify the position of the ejection center point and acquires an ejection center point image, then binarizes the ejection center point image and detects the position of the ejection center point by circle detection processing using Hough transform. [Effects of the Invention]

[0008] According to this invention, the state of lubricating oil ejection from the bearings of a hydroelectric generator can be automatically monitored. [Brief explanation of the drawing]

[0009] [Figure 1] This is a schematic diagram showing the configuration of the rotation mechanism 1 of a hydroelectric power generator using the lubrication oil monitoring device 20 according to the first embodiment of the present invention. [Figure 2]This is a block diagram showing the configuration of the image processing apparatus 23 according to the first embodiment. [Figure 3] This is a flowchart illustrating the operation of the lubricating oil monitoring device 20 according to this first embodiment. [Figure 4] This is a schematic diagram illustrating the operation of the lubricating oil monitoring device 20 according to this first embodiment (detection of crescent-shaped ejection). [Figure 5] This is a schematic diagram illustrating the operation (circle detection process) of the lubricating oil monitoring device 20 according to this first embodiment. [Figure 6] This is a schematic diagram illustrating the operation (detection of the ejection center point) of the lubricating oil monitoring device 20 according to this first embodiment. [Figure 7] This is a schematic diagram illustrating the operation of the state determination unit 28 according to this first embodiment. [Figure 8] This schematic diagram shows the operation of the state determination unit 28 according to the first embodiment, illustrating the crescent-shaped ejection shape, ejection center point, and lubricating oil in both normal and abnormal lubrication states. [Figure 9] This is a schematic diagram showing the configuration of the rotation mechanism 1 of a hydroelectric power generator using the lubrication oil monitoring device 20 according to the second embodiment. [Figure 10] This is a block diagram showing the configuration of the image processing apparatus 23 according to this second embodiment. [Figure 11] This is a flowchart illustrating the operation of the lubricating oil monitoring device 20 according to this second embodiment. [Figure 12] This is a schematic diagram showing an image captured by the camera 22 according to this second embodiment. [Figure 13] This is a schematic diagram showing the configuration of the rotation mechanism 1 of a hydroelectric power generator using the lubrication oil monitoring device 20 according to the third embodiment. [Figure 14] This is a block diagram showing the configuration of the image processing apparatus 23 according to this third embodiment. [Figure 15] This is a flowchart illustrating the operation of the lubricating oil monitoring device 20 according to this third embodiment. [Figure 16]It is a schematic diagram showing captured images captured by cameras 21 and 22 according to the third embodiment and ellipse conversion processing.

Mode for Carrying Out the Invention

[0010] (First Embodiment) Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.

[0011] FIG. 1 is a schematic diagram showing the configuration of a rotation mechanism 1 of a hydroelectric power generator using a lubricating oil monitoring device 20 according to the first embodiment of the present invention. In FIG. 1, the rotation mechanism 1 includes a rotating machine 10, a rotating shaft 11, and bearing portions 12 and 13. The rotating machine 10 rotates on the rotating shaft 11 in a predetermined direction at a predetermined rotational speed. The rotating shaft 11 is generally supported by the bearing portions 12 and 13 via bearings (not shown). Lubricating oil is stored in the bearing portions 12 and 13. The lubricating oil is constantly supplied to rotating parts such as the rotating shaft 11 and the bearings while being agitated along with the rotation of the rotating shaft in the bearing portions 12 and 13.

[0012] In rotating mechanism equipment (such as generators and rotating machines), the bearing portion constantly ejects lubricating oil onto the shaft to reduce shaft friction. If the amount of lubricating oil decreases or leaks abnormally, the shaft may generate heat, break, and cause equipment damage. Therefore, it is very important to check whether the lubricating oil is properly ejected onto the shaft.

[0013] There are multiple types of methods for checking the state of lubricating oil. One of the methods is to make the ejected lubricating oil hit a window and monitor the state such as how the oil hits and its momentum. This monitoring method is defined as "ejection state monitoring". The present invention relates to this ejection checking method. In ejection state monitoring, the ejection shape has the following features. · A circular ejection center point is generated at the position where the lubricating oil hits. · The lubricating oil spreads on the upper surface portion, forming a crescent-shaped ejection shape.

[0014] The bearing sections 12 and 13 are provided with oil ejection confirmation windows 14 and 15 on their sides for monitoring the ejection state. The oil ejection confirmation windows 14 and 15 have a window portion made of a light-transmitting material (such as plastic or glass). A portion of the lubricating oil being agitated in the bearing sections 12 and 13 is ejected through a predetermined pipe (not shown) into the inner surface of the oil ejection confirmation windows 14 and 15. At this time, as described above, the oil ejection confirmation windows 14 and 15 can observe a circular ejection center point formed by the ejected lubricating oil and a crescent-shaped ejection form that curves upward from the ejection center point.

[0015] The lubricating oil monitoring device 20 consists of cameras 21 and 22 and an image processing device 23. Cameras 21 and 22 are each installed facing the oil ejection confirmation windows 14 and 15, and have an imaging field of view approximately the same size as the oil ejection confirmation windows 14 and 15. Cameras 21 and 22 photograph the ejection center point of the lubricating oil ejected onto the inner surface of the oil ejection confirmation windows 14 and 15, and the crescent-shaped ejection form that is formed in an arc upward from the ejection center point, and send the captured images (still images or videos taken at predetermined time intervals) to the image processing device 23.

[0016] The image processing device 23 detects the position of the crescent-shaped ejection and the position of the ejection center point from the images from the cameras 21 and 22, and determines the state of the lubricating oil from the position of the crescent-shaped ejection and the position of the ejection center point. The image processing device 23 may be a so-called computer. The cameras 21 and 22 and the image processing device 23 may be connected by either a wired or wireless connection.

[0017] Figure 2 is a block diagram showing the configuration of the image processing device 23 according to this first embodiment. The image processing device 23 consists of an image input unit 24, a storage unit 25, a jet shape recognition unit 26, a jet center point recognition unit 27, and a state determination unit 28. The image input unit 24 captures images taken by cameras 21 and 22 and supplies them to the storage unit 25.

[0018] The ejection shape recognition unit 26 detects the location of the crescent-shaped ejection from the captured image. More specifically, the ejection shape recognition unit 26 uses a crescent-shaped template image to identify the location of the crescent-shaped ejection and acquire an ejection shape image. After binarizing the ejection shape image, it performs a circle detection process using Hough transform to detect the position of the arc on the upper contour of the crescent-shaped ejection. This makes it possible to remove the background outside the oil ejection confirmation window, such as the equipment wall. If the ejection shape is not detected when identifying the ejection shape location using the crescent-shaped template, it is determined that there is no lubricating oil ejection.

[0019] The above binarization process is performed to enhance the contour of the crescent-shaped ejection. In this process, Gaussian filters or median filters may be used to remove noise within the oil ejection confirmation windows 14 and 15, and histogram flattening or gamma correction may be performed to enhance the contour.

[0020] The ejection center point recognition unit 27 detects the position of the ejection center point from the captured image. More specifically, the ejection center point recognition unit 27 uses a circular ejection center point template image to identify the position of the ejection center point and acquire an ejection center point image. After binarizing the ejection center point image, it detects the position of the ejection center point by performing a circle detection process using Hough transform. This makes it possible to remove the background outside the oil ejection confirmation windows 14 and 15, such as the equipment wall. If no ejection shape is detected when identifying the ejection shape location using the template, it is determined that there is no lubricating oil ejection.

[0021] The above binarization process is performed to enhance the contour of the circular ejection center point. In this process, Gaussian filters or median filters may be used to remove noise within the oil ejection confirmation windows 14 and 15, and histogram flattening or gamma correction may be performed to enhance the contour.

[0022] Note that, depending on the lubricating oil ejection pattern, the ejection center point may not appear circular due to surrounding noise. In such cases, the position coordinates of the ejection center point should be determined from a pre-set coordinate position (for example, the center position of the template image being the coordinate of the center point).

[0023] The state determination unit 28 determines the state of the lubricating oil from the position of the crescent-shaped ejection and the position of the ejection center point. More specifically, the state determination unit 28 compares the position of the arc of the upper contour of the crescent-shaped ejection and the position of the ejection center point in a normal state with the position of the arc of the upper contour of the crescent-shaped ejection and the position of the ejection center point at the time of observation, and determines that the state of the lubricating oil is abnormal if the difference between the two exceeds a predetermined threshold.

[0024] The memory unit 25 stores images captured by cameras 21 and 22, crescent-shaped template images used by the ejection shape recognition unit 26, position information of the crescent-shaped ejection shape, circular ejection center point template images used by the ejection center point recognition unit 27, position information of the ejection center point, and the determination results (lubricating oil status) from the state determination unit 28.

[0025] Figure 3 is a flowchart illustrating the operation of the lubricating oil monitoring device 20 according to this first embodiment. Figures 4, 5, and 6 are schematic diagrams illustrating the operation of the lubricating oil monitoring device 20 according to this first embodiment.

[0026] First, the image input unit 24 receives images of the oil ejection confirmation window captured by cameras 21 and 22 (step S10). As shown in Figure 4(a), the captured image 40 shows the oil ejection confirmation window 14 (15), and within it, the crescent-shaped ejection shape 41 and the ejection center point 42.

[0027] Next, the ejection shape recognition unit 26 recognizes the ejection shape from the captured image (step S12). More specifically, the ejection shape recognition unit 26 uses a crescent-shaped template image to identify the position of the crescent-shaped ejection shape 41 and acquires an ejection shape image. After binarizing the ejection shape image, it detects the arc of the upper contour of the crescent-shaped ejection shape by performing a circle detection process using the Hough transform.

[0028] As shown in Figure 4(b), the position 43 of the crescent-shaped ejection 41 is identified from the captured image 40, and as shown in Figure 4(c), an ejection shape image 45 is obtained. Furthermore, as shown in Figure 5(a), the position of the arc 46 of the upper contour of the crescent-shaped ejection 41 is detected by binarizing the ejection shape image 45 and then performing circle detection processing using the Hough transform (dashed line).

[0029] Depending on the lubricating oil ejection conditions, the arc 46 of the upper contour may take on a separated shape, as shown in Figure 5(b). However, by using circle detection processing, even if the circular shape is broken off from the most circular information, it is possible to detect its position as the arc 46 of the upper contour of the crescent-shaped ejection shape 41.

[0030] Next, the ejection center point recognition unit 27 recognizes the position of the ejection center point from the captured image (step S14). More specifically, the ejection center point recognition unit 27 uses a circular ejection center point template image to identify the position of the ejection center point and acquire an ejection center point image. After binarizing the ejection center point image, it detects the (center) position of the ejection center point by performing a circle detection process using the Hough transform.

[0031] As shown in Figure 6(a), the position 50 of the eruption center point 42 is identified from the captured image 40, and as shown in Figure 6(b), an eruption center point image 51 is obtained. Furthermore, as shown in Figure 6(c), the (center) position 52 of the eruption center point 42 is detected by binarizing the eruption center point image 51 and then performing circle detection processing using the Hough transform (dashed line).

[0032] Next, the state determination unit 28 determines the state of the lubricating oil from the position of the arc 46 of the upper contour of the crescent-shaped ejection shape 41 and the position 52 of the ejection center point (step S16). More specifically, the state determination unit 28 calculates the size of the crescent-shaped ejection shape and the position 52 of the ejection center point 42 as the ejection state using the following formula, based on the position of the arc 46 of the upper contour of the crescent-shaped ejection shape and the position 52 of the ejection center point 42.

[0033] The (vertical) size c of the crescent-shaped ejection is equal to the position of the ejection center b minus the position of the top of the crescent a. The coordinates of the ejection center point d = vertical size of the oil ejection confirmation window image (100) / 2 - ejection center position b

[0034] In this first embodiment, "position" refers to the vertical (up and down) distance from the origin, where the upper left corner of the captured image 40 is defined as the origin "0" and the lower left corner as "100". The coordinates correspond to position "50" and to the exact center of the captured image 40 (coordinate = 0).

[0035] Figure 7 is a schematic diagram illustrating the operation of the state determination unit 28 according to this first embodiment. Figure 7(a) shows the position a of the arc of the upper contour of the crescent-shaped ejection, and Figure 7(b) shows the ejection center position b. Figure 7(c) shows the (vertical) size c of the crescent-shaped ejection and the ejection center point coordinates d (=100 / 2-b). For example, if b=50, the ejection center point coordinates d become 100 / 2-50=0, indicating that it is exactly in the center of the captured image 40.

[0036] Figure 8 is a schematic diagram showing the operation of the state determination unit 28 according to this first embodiment, illustrating the crescent-shaped ejection shape, ejection center point, and lubricating oil in both normal and abnormal lubrication states. In the normal state, as shown in Figure 8(a), the coordinate d (position) of the ejection center point 42 is located approximately in the center of the oil ejection confirmation windows 14 and 15, and the size c of the crescent-shaped ejection shape 41 is relatively large, extending from above the oil ejection confirmation windows 14 and 15 to the position of the ejection center point 42. On the other hand, in the abnormal state, as shown in Figure 8(b), the coordinate d (position) of the ejection center point is located below the center of the oil ejection confirmation windows 14 and 15, the size c of the crescent-shaped ejection shape becomes smaller, and it takes on a vertically flattened shape.

[0037] In this first embodiment, the method for evaluating the state of the lubricating oil is not specified. For example, the evaluation may be performed by setting threshold values ​​based on the normal state values, or by summing the differences between the normal state values ​​calculated in advance, or by weighted sums, etc.

[0038] Subsequently, the flowchart is executed repeatedly, and observation of the oil ejection confirmation windows 14 and 15 continues.

[0039] As described above, according to this first embodiment, the shape of the lubricating oil being ejected into the oil ejection confirmation windows can be recognized from the images captured by cameras 21 and 22, which are installed facing the oil ejection confirmation windows 14 and 15, and the ejection state can be understood from the shape information (its position and size).

[0040] Furthermore, according to this first embodiment, the position of the lubricating oil ejection center point and the size of the crescent-shaped ejection shape derived from the position of the arc of the upper contour of the crescent-shaped ejection shape formed in an arc upward from the ejection center point can be recognized from the captured image, and the ejection state can be grasped from the recognized position of the ejection center point and the size of the crescent-shaped ejection shape. In addition, by grasping the ejection state, it is possible to quantify the ejection state monitoring and automate the work. In this embodiment, two sets of oil ejection confirmation windows and cameras have been described, but the system is not limited to this, and one set of oil ejection confirmation windows and cameras may be used, and the above-described functions and effects can be sufficiently obtained with one set of oil ejection confirmation windows and cameras. Of course, depending on the conditions of implementation, a configuration with more than two sets may be adopted.

[0041] (Second Embodiment) The difference between the lubricating oil monitoring device 20 of this second embodiment and the first embodiment lies in the installation positions of the cameras 21 and 22, and consequently, the provision of an oil ejection confirmation window detection unit 29 on the image processing device 23. The other configurations are the same as those of the first embodiment, so their description will be omitted.

[0042] Figure 9 is a schematic diagram showing the configuration of the rotation mechanism 1 of a hydroelectric power generator using the lubrication oil monitoring device 20 according to this second embodiment. This second embodiment is characterized by the fact that cameras 21 and 22 can be installed at a distance from the oil ejection confirmation windows 14 and 15, as shown in Figure 9. Because there is a distance between the oil ejection confirmation windows 14 and 15 and the cameras 21 and 22, images other than the oil ejection confirmation windows 14 and 15 will be captured in the images taken by the cameras 21 and 22. Therefore, an oil ejection confirmation window detection unit 29, which will be described later, is provided to detect the image of the oil ejection confirmation window from the captured image.

[0043] Figure 10 is a block diagram showing the configuration of the image processing device 23 according to this second embodiment. In Figure 10, the image processing device 23 further includes an oil ejection confirmation window detection unit 29. Before processing by the ejection shape recognition unit 26 and the ejection center point recognition unit 27, the oil ejection confirmation window detection unit 29 detects an image of the oil ejection confirmation window using an oil ejection confirmation window template image from captured images taken by cameras 21 and 22 from a distance, which include backgrounds other than the oil ejection confirmation windows 14 and 15.

[0044] Figure 11 is a flowchart illustrating the operation of the lubrication oil monitoring device 20 according to this second embodiment. Figure 12 is a schematic diagram showing an image captured by the camera 22 according to this second embodiment. First, the image input unit 24 receives the image of the oil ejection confirmation window captured by cameras 21 and 22 (step S20). As shown in Figure 12, the captured image 60 includes not only the oil ejection confirmation window 15, but also the bearing section 13, the rotating shaft 11, and a part of the rotating machine (generator) 10.

[0045] Next, the oil ejection window detection unit 29 detects images of the oil ejection windows 14 and 15 from the captured images taken by cameras 21 and 22, which include backgrounds other than the oil ejection windows 14 and 15, using an oil ejection window template image (step S22). As shown in Figure 12, an image 61 of the oil ejection window 15 is detected from the image taken by camera 22. Similarly, an image of the oil ejection window 14 is detected from the image taken by camera 21.

[0046] Next, the ejection shape recognition unit 26 uses a crescent-shaped template image to identify the position of the crescent-shaped ejection shape from the images of the oil ejection confirmation windows 14 and 15 detected by the oil ejection confirmation window detection unit 29, and obtains an ejection shape image. After binarizing the ejection shape image, it detects the arc of the upper contour of the crescent-shaped ejection shape by performing a circle detection process using the Hough transform (step S24).

[0047] Next, the ejection center point recognition unit 27 uses a circular ejection center point template image to identify the position of the ejection center point from the image detected by the oil ejection confirmation window detection unit 29 and acquires an ejection center point image. After binarizing the ejection center point image, it detects the (center) position of the ejection center point by performing a circle detection process using the Hough transform (step S26).

[0048] Next, the state determination unit 28 calculates the size of the crescent-shaped ejection shape and the coordinates (position) of the ejection center point from the position of the arc of the upper contour of the crescent-shaped ejection shape and the position of the ejection center point using the formula described above, and determines the state of the lubricating oil (step S28).

[0049] In this second embodiment, the method for evaluating the state of the lubricating oil is the same as in the first embodiment, so a description will be omitted. Subsequently, the flowchart is repeatedly executed, and observation of the oil ejection confirmation windows 14 and 15 is continued.

[0050] According to this second embodiment, the shape of the lubricating oil being ejected into the oil ejection confirmation windows is recognized from images captured by cameras 21 and 22 installed at a distance from the oil ejection confirmation windows 14 and 15, and the ejection state is understood from the shape information (its position and size). As a result, cameras 21 and 22 can be installed at a distance from the oil ejection confirmation windows 14 and 15, and even if there is no space to install them near the rotating mechanism 1, there is no constraint on the distance between the oil ejection confirmation windows 14 and 15 and the cameras 21 and 22, which has the advantage of making it easy to secure a place to install the cameras 21 and 22.

[0051] Furthermore, in this second embodiment, since the images are captured by cameras 21 and 22 installed at a distance from the oil ejection confirmation windows 14 and 15, images other than the oil ejection confirmation windows may appear in the captured images. However, the oil ejection confirmation window detection unit 29 detects the area of ​​the oil ejection confirmation windows 14 and 15 from the captured images using an oil ejection confirmation window template image, so the state of the lubricating oil can be understood in the same way as in the first embodiment. In addition, by understanding the ejection state, it becomes possible to quantify the ejection state monitoring and automate the work. It also becomes easier to secure a place to install the cameras 21 and 22. In this embodiment, two sets of oil ejection confirmation windows and cameras have been described, but this is not limited to this, and one set of oil ejection confirmation windows and cameras may be used, and the above-described functions and effects can be sufficiently obtained with one set of oil ejection confirmation windows and cameras. Of course, depending on the conditions of implementation, a configuration with more than two sets may be adopted.

[0052] (Third embodiment) The lubrication oil monitoring device 20 of this third embodiment differs from the first embodiment in that it can capture images using cameras 21 and 22 that are at different distances from the oil ejection confirmation windows 14 and 15, and whose shooting positions (shooting directions) vary each time. Accordingly, the image processing device 23 is further equipped with an oil ejection confirmation window detection unit 29 and an elliptic transformation unit 30. The other configurations are the same as those of the first embodiment, so their description will be omitted.

[0053] Figure 13 is a schematic diagram showing the configuration of the rotation mechanism 1 of a hydroelectric power generator using the lubrication oil monitoring device 20 according to this third embodiment. In this third embodiment, cameras 21 and 22 photograph the oil ejection confirmation windows 14 and 15 from a distance and shooting position (shooting direction) that varies each time. In this way, when the distance from the oil ejection confirmation windows 14 and 15 and the shooting position (shooting direction) vary each time, images taken by cameras 21 and 22 will include elements other than the oil ejection confirmation windows 14 and 15, and if photographed from an oblique direction, the oil ejection confirmation windows 14 and 15 will appear as an elliptical image, requiring appropriate image processing. However, this has the advantage of allowing photography (observation) from moving objects, such as rail-mounted cameras or cameras mounted on mobile robots or drones, which increases flexibility and improves convenience.

[0054] Figure 14 is a block diagram showing the configuration of the image processing device 23 according to this third embodiment. In Figure 14, the image processing device 23 further comprises an oil ejection confirmation window detection unit 29 and an elliptic transformation unit 30. Before processing by the ejection shape recognition unit 26 and the ejection center point recognition unit 27, the oil ejection confirmation window detection unit 29 detects images of the oil ejection confirmation windows 14 and 15 using an oil ejection confirmation window template image from captured images taken by cameras 21 and 22 from a position away from the oil ejection confirmation windows 14 and 15, which include backgrounds other than the oil ejection confirmation windows 14 and 15. The elliptic transformation unit 30 performs an elliptic transformation process on the images of the oil ejection confirmation windows 14 and 15 detected by the oil ejection confirmation window detection unit 29, thereby converting the images of the oil ejection confirmation windows 14 and 15, which have been deformed into an elliptical shape, into a circular shape.

[0055] Figure 15 is a flowchart illustrating the operation of the lubricating oil monitoring device 20 according to this third embodiment. Figure 16 is a schematic diagram showing the captured images and elliptic transformation processing performed by the cameras 21 and 22 according to this third embodiment. First, the image input unit 24 receives images of the oil ejection confirmation windows 14 and 15 captured by the cameras 21 and 22 (step S30). Because the captured images are taken from a relatively distant position, the background other than the oil ejection confirmation windows 15 is visible in the images.

[0056] Next, the oil ejection window detection unit 29 detects images of the oil ejection windows 14 and 15 from the captured images taken by cameras 21 and 22, which include backgrounds other than the oil ejection windows 14 and 15, using an oil ejection window template image (step S32). As shown in Figure 16, an image 70 of the oil ejection window 14 (15) is detected. If the image is taken from an oblique angle, the oil ejection window 14 (15) will appear deformed into an elliptical shape in the image, as shown in Figure 16.

[0057] Although the operation is the same as in Embodiment 1, the positions of cameras 21 and 22 change each time, so there is a possibility that the oil ejection confirmation windows 14 and 15 may not be detected in a single oil ejection confirmation window template image. For this reason, multiple oil ejection confirmation window template images may be prepared in advance, and the process of detecting the oil ejection confirmation windows 14 and 15 in the same way with the next template may be continued each time a failure occurs. This makes it possible to detect images of the oil ejection confirmation windows 14 and 15.

[0058] Next, the elliptic transformation unit 30 performs an elliptic transformation on the images 70 of the oil ejection confirmation windows 14 and 15 detected by the oil ejection confirmation window detection unit 29, thereby converting the images 70 of the oil ejection confirmation windows 14 and 15, which have been deformed into an elliptic shape, into a circular shape (step S34).

[0059] Here, we will explain elliptic transforms in detail. When the identified oil ejection confirmation windows 14 and 15 are photographed from an oblique angle by cameras 21 and 22, they are captured in an elliptical shape, as shown in Figure 16. At this time, the elliptical shape (edge ​​of the oil ejection confirmation window) is converted to a circular shape from the image 70 of the elliptical oil ejection confirmation window 14 (15) by performing an image transformation at a magnification that makes the ratio of the vertical and horizontal directions of the ellipse the same, based on the parameters that constitute the ellipse (edge ​​of the oil ejection confirmation window) (center coordinates: X, Y, vertical length of the ellipse: H, horizontal length of the ellipse: W, tilt angle: DEG). This makes it possible to process the image as the same regardless of the direction (viewpoint) from which it is photographed.

[0060] Next, the ejection shape recognition unit 26 uses a crescent-shaped template image to identify the position of the crescent-shaped ejection shape from the images of the oil ejection confirmation windows 14 and 15, which have been converted to a circular shape by the elliptic transformation unit 30, and obtains an ejection shape image. After binarizing the ejection shape image, it detects the arc of the upper contour of the crescent-shaped ejection shape by performing a circle detection process using the Hough transform (step S36).

[0061] Next, the ejection center point recognition unit 27 uses a circular ejection center point template image to identify the position of the ejection center point from the images of the oil ejection confirmation windows 14 and 15 that have been converted into a circular shape by the elliptic transformation unit 30, and obtains an ejection center point image. After binarizing the ejection center point image, it detects the (center) position of the ejection center point by performing a circle detection process using the Hough transform (step S38).

[0062] Next, the state determination unit 28 calculates the size of the crescent-shaped ejection shape and the coordinates (position) of the ejection center point from the position of the arc 46 of the upper contour of the crescent-shaped ejection shape 41 and the position of the ejection center point using the formula described above, and determines the state of the lubricating oil (step S40).

[0063] In this third embodiment, the method for evaluating the state of the lubricating oil is the same as in the first and second embodiments, so a detailed explanation will be omitted. Subsequently, the flowchart is repeatedly executed, and observation of the oil ejection confirmation windows 14 and 15 is continued.

[0064] According to this third embodiment, the oil ejection confirmation windows 14 and 15 are photographed by cameras 21 and 22, which are at different distances from the windows and have different shooting positions (shooting directions) each time. The oil ejection confirmation window detection unit 29 detects images of the oil ejection confirmation windows 14 and 15 from images that include the background using an oil ejection confirmation window template image. The elliptic transformation unit 30 then performs an elliptic transformation to convert the images into circular shapes. As a result, the state of the lubricating oil can be understood in the same way as in the first embodiment, but the constraints on the camera's shooting position are eliminated, providing greater flexibility. As long as the oil ejection confirmation windows 14 and 15 can be photographed, the images can be taken from anywhere. Furthermore, understanding the ejection state enables the quantification of ejection state monitoring and the automation of the work.

[0065] In the third embodiment, cameras 21 and 22 were used to capture images from different distances from the oil ejection confirmation windows 14 and 15, and from different shooting positions (shooting directions) each time. However, cameras mounted on mobile devices such as mobile robots or flying objects (drones, etc.) may be used to capture images from any position and direction. The image processing device 23 may be implemented as an information processing device mounted on the mobile device, or it may be installed separately in another location and wirelessly connected to cameras 21 and 22. In this embodiment, two sets of oil ejection confirmation windows and cameras were described, but the system is not limited to this. There may be only one set of oil ejection confirmation windows and cameras, and the above-described functions and effects can be sufficiently obtained with one set of oil ejection confirmation windows and cameras. Of course, more than two sets may be adopted, taking into account the conditions of implementation as appropriate.

[0066] In the above-described embodiment, the method for evaluating the state of the lubricating oil was not specified. However, for example, the position of the arc of the upper contour of the crescent-shaped ejection shape and the position of the ejection center point under normal conditions may be compared with the position of the arc of the upper contour of the crescent-shaped ejection shape and the position of the ejection center point at the time of observation. If the difference between the two exceeds a predetermined threshold, it may be determined that the state of the lubricating oil is abnormal. Furthermore, if the difference between the two exceeds a predetermined threshold, the abnormal state of the lubricating oil may be notified by a predetermined means (such as sounding a buzzer). [Explanation of Symbols]

[0067] 1. Rotation mechanism 10 Rotating Machines 11 Rotation axis 12, 13 Bearing section 14, 15 Oil spray confirmation window 21, 22 Camera 23 Image Processing Device 24 Image Input Section 25 Memory section 26 Spout shape recognition part 27 Ejection center point recognition unit 28 State determination unit 29 Oil ejection confirmation window detection unit 30 Elliptic Transform Section 40 captured images 41 Crescent-shaped spout shape 42 Center point of eruption 43. Location of the crescent-shaped ejection nozzle 45. Image of ejection shape 46 Arc of the upper contour 50. Location of the eruption center 51 Eruption center point image 52. The (center) position of the eruption center. 60 captured images 61 Image of the oil spray indicator window 70 Image of the oil spray indicator window

Claims

[Claim 1] A lubricating oil monitoring device that monitors the center point of lubricating oil ejection and the crescent-shaped ejection form that is formed in an arc upward from the center point of ejection, as observed in an oil ejection confirmation window from which lubricating oil is ejected from the bearings of a hydroelectric generator, A camera for photographing the oil ejection confirmation window, The system comprises an image processing device and, The aforementioned image processing device is A ejection shape recognition unit that detects the position of the crescent-shaped ejection from the image of the camera, A ejection center point recognition unit that detects the position of the ejection center point from the image of the camera, The system includes a state determination unit that determines the state of the lubricating oil based on the position of the crescent-shaped ejection and the position of the ejection center point. In the ejection shape recognition unit, Using a crescent-shaped template image, the position of the crescent-shaped ejection is identified and an ejection shape image is obtained. After the ejection shape image is binarized, the arc of the upper contour of the crescent-shaped ejection is detected by circle detection processing using the Hough transform. In the aforementioned ejection center point recognition unit, A lubricating oil monitoring device that identifies the position of the ejection center point using a circular ejection center point template image to obtain an ejection center point image, and then detects the position of the ejection center point by binarizing the ejection center point image and performing a circle detection process using a Hough transform.

Images