Water turbine large shaft creep detection method based on improved three-frame difference method

Abstract

The invention belongs to the field of water turbine large shaft creeping detection, and particularly provides a water turbine large shaft creeping detection method based on an improved three-frame difference method, and the method comprises the steps: collecting a large shaft image after a unit is shut down; carrying out graying processing on the acquired image; performing region matching on the three frames of grayscale images to obtain a compensation image; performing difference calculation on the two adjacent frames of compensation images to obtain two difference images; performing threshold segmentation processing on the two difference images to obtain a binary image; performing morphological expansion and corrosion processing on the binarized image, and then performing difference calculation; performing threshold value processing on the difference image, and regarding pixel points with difference values exceeding a threshold value as a wriggle area; and performing connected region analysis on the creeping region, and judging whether each connected region meets a creeping condition or not. The method provided by the invention can effectively detect the wriggling state of the large shaft of the water turbine, has high detection precision and real-time performance, and provides important guarantee for safe operation of a hydroelectric generating set.

Description

Technical Field

[0001] The invention belongs to the field of water turbine main shaft creep detection, and in particular relates to a water turbine main shaft creep detection method based on an improved three-frame difference method. Background Art

[0002] In the field of hydropower generation, unit creep is a complex issue that has attracted much attention. The root cause is the water leakage of the guide vanes that may occur when the unit is shut down, which may cause the unit rotor to rotate slowly, thus causing the risk of unit creep. For vertical units, creep mainly harms the thrust bearing; for horizontal units, it may affect the guide bearing and thrust bearing, and even shorten the bearing life, causing serious accidents such as dry friction and burning of the bearing.

[0003] In order to timely discover and solve the unit creep problem, an effective creep detection device is urgently needed. There are two types of creep detection devices on the market: contact and non-contact, but they both have certain limitations. Contact devices are usually located near the main shaft and are easily damaged by long-term vibration. They may even be mistakenly put into use, causing friction with the main shaft and damaging the equipment. Traditional non-contact devices use a combination of iron gear discs and probes, which have low detection accuracy. Especially when the main shaft creeps at a low speed, it cannot accurately reflect the creep of the unit, delaying the identification and treatment of the problem.

[0004] In recent years, computer vision has developed rapidly as an important field of artificial intelligence. As a commonly used target detection technology in the field of computer vision, the three-frame difference method is widely used in video surveillance systems. The traditional three-frame difference method is prone to the disadvantage of "hole" phenomenon. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a method for detecting creep of a main shaft of a turbine based on an improved three-frame differential method, which has relatively high detection accuracy, effectively guarantees the safe operation of a hydroelectric generating set, and ensures the safety and reliability of power production.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: a method for detecting creep of a main shaft of a turbine based on an improved three-frame difference method, comprising the following steps: S1. Obtaining the main axis image data: After the unit is shut down, collect the main axis image and obtain the latest three consecutive frames of original images in real time. , , ; S2. Grayscale the collected image to obtain three frames of grayscale images , , ; S3, performing region matching on the three grayscale images after grayscale processing to obtain a compensated image; S4. After obtaining the compensated image, the compensated images of two adjacent frames are respectively subjected to differential calculation to obtain two differential images. and ; S5. Perform threshold segmentation on the two difference images to obtain binary images. and ; S6, performing morphological dilation and corrosion processing on the binary image; S7, performing pixel-level difference calculation on the binary image processed in step S6 to obtain a difference image; S8, performing threshold processing on the difference image, and considering the pixel points whose difference values ​​exceed the threshold as the creeping area; S9, performing connected region analysis on the peristaltic region to find the boundary or center point of the connected region; S10. According to the characteristics of the connected regions and in combination with the actual creep characteristics, whether each connected region meets the creep condition is determined by analyzing whether the area, shape, and position of the connected regions meet the predetermined creep characteristics, thereby obtaining the creep information of the main shaft of the turbine.

[0007] In the preferred solution, in step S1, three consecutive frames of original images are obtained. , , In the image, each frame consists of three channels: R, G, and B. The mathematical expression is: ; in, represents the pixel in the i-th row and j-th column of the image, Contains the values ​​of the three channels R, G, and B: , , are the pixel values ​​of the R channel, G channel, and B channel in the i-th row and j-th column respectively.

[0008] In the preferred solution, in step S2, the grayscale processing is: each channel of each frame image has a certain weight ratio, and the three-dimensional original image is merged into a one-dimensional grayscale image, and the expression is: ; in, represents a grayscale image, , and Represents the pixel values ​​of the R channel, G channel, and B channel.

[0009] In a preferred solution, in step S3, the processed three frames of video images are subjected to region matching, comprising the following steps: S3-1, in the previous frame image Select a module area and set the size to 32*32; S3-2, in the current frame image Search for the optimal area that can match the reference model in the previous frame image, and calculate the optimal matching point by the formula: ; in, and are the grayscale values ​​of the current frame and the previous frame respectively, m and n represent the corresponding matching pixel coordinates in images of different frames, is the minimum absolute error, 32*32 is the size of the selected area matching template, and the The point corresponding to the minimum value of is the optimal matching point.

[0010] S3-3. Calculate the relative displacement by comparing the position difference of the matching area in the current frame and the previous frame image. According to the obtained relative displacement, translate all the pixels in the current frame image by the relative displacement amount, and then recalculate the matching area between the translated image and the previous frame image to compensate and correct the current frame image.

[0011] In the preferred solution, in step S4, the compensation images of two adjacent frames are respectively subjected to differential calculation to obtain two differential images. and , the calculation formula is: .

[0012] In the preferred solution, in step S5, the two difference images are and Perform threshold segmentation processing, the calculation formula is: ; in, is the binary image obtained after threshold segmentation. for and , represents the difference map calculated in step 5, and T is the set threshold.

[0013] In the preferred solution, in step S6, the segmented binary image and Perform morphological expansion and corrosion processing, including the following steps: S6-1. First, perform a morphological dilation operation on the binary image. The calculation formula is: ; Among them, dilate() is a morphological expansion operation. and is the binary image after the dilation operation; S6-2, and then carry out corrosion treatment, the calculation formula is: ; Among them, erode() is a morphological erosion operation. and It is the binary image after the erosion operation.

[0014] The invention provides a method for detecting creep of a main shaft of a turbine based on an improved three-frame difference method, which has the following beneficial effects: 1. Compared with the traditional non-contact electrical pulse detection and contact mechanical friction detection methods, the intelligent judgment of the main shaft creep of the present invention can intelligently track the main shaft creep after the unit is shut down. It has high detection accuracy and strong reliability, and can send out the unit creep signal more timely, which improves the efficiency of the hydropower plant operation and maintenance personnel in handling unit creep-related problems to a certain extent.

[0015] 2. In step S6 of the present invention, the problem of "hole" phenomenon that is prone to occur in the traditional three-frame difference method is optimized. The holes in the image are filled by a morphological dilation operation, and then the motion area is changed to its original size by an erosion operation. In this way, a more accurate motion area is obtained, thereby increasing the accuracy and reliability of large-axis creep detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The present invention will be further described below in conjunction with the accompanying drawings and embodiments: Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0018] like Figure 1 As shown, a method for detecting the creep of the main shaft of a turbine based on an improved three-frame difference method is described. The specific operation steps are as follows: S1. After the unit is shut down, start collecting the main axis image. Use an industrial camera with a resolution of 1920x1080 and a collection frequency of 30 frames per second to obtain the latest three consecutive original frames in real time. , , .

[0019] S2: Grayscale the collected image to obtain a grayscale image , , , the grayscale processing formula is: ; ; ; S3, performing region matching on the processed three frames of video images, the matching process is as follows: S3-1, in the previous frame image A module area with a size of 32*32 is selected as the reference model.

[0020] S3-2, in the current frame image Search for the most similar area that can match the reference model in the previous frame image, calculate the optimal matching point through the formula, and calculate the relative displacement for compensation correction. The specific formula is:

[0021] in, and are the grayscale values ​​of the current frame and the previous frame respectively, is the minimum absolute error, 32*32 is the size of the selected area matching template, and the The point corresponding to the minimum value of is the optimal matching point.

[0022] S3-3. Calculate the relative displacement by comparing the position difference of the matching area in the current frame and the previous frame image. According to the obtained relative displacement, translate all the pixels in the current frame image by the relative displacement amount, and then recalculate the matching area between the translated image and the previous frame image to compensate and correct the current frame image.

[0023] Calculate the relative displacement between the current frame and the previous frame, and perform compensation correction on the current frame. The compensation calculation of the current frame is the same as that of the next frame.

[0024] S4. After obtaining the compensated image, the compensated images of two adjacent frames are respectively subjected to differential calculation to obtain two differential images. The calculation formula is as follows: ; S5, the two difference images and Perform threshold segmentation processing, the formula is as follows; ; ; The thresholds T1 and T2 are set to 15 respectively.

[0025] Through the above formula, the two images are subjected to threshold segmentation processing to obtain and Binarized image.

[0026] S6. Binarize the above image and Perform morphological expansion and corrosion processing: S6-1. First, perform a morphological dilation operation on the binary image. The specific formula is as follows: ; Among them, dilate() is a morphological expansion operation.

[0027] S6-1, then carry out corrosion treatment, the specific formula is as follows: ; Among them, erode() is a morphological erosion operation.

[0028] S7, performing pixel-level difference calculation on the processed binary image to obtain a difference image: ; S8. Difference image Threshold processing is performed, and the pixels whose difference values ​​exceed the threshold are regarded as crawling areas.

[0029] In this embodiment, the threshold is set to 5. These regions may represent small changes in the surface of the cylinder, such as small displacements or deformations caused by rotation.

[0030] S9, perform connected area analysis on the peristaltic area to find the boundary or center point of the connected area. The characteristics of the connected area include area, perimeter, center position, etc. Here, in the connected area analysis, the area threshold can be set to 100 pixels, the perimeter threshold can be set to 30 pixels, and the offset range of the center position can be set to within 5 pixels. When these three preset conditions are met at the same time, the system determines that the area has peristalsis.

[0031] S10. According to the characteristics of the connected regions and in combination with the actual creep characteristics, whether each connected region meets the creep condition is determined by analyzing whether the area, shape, and position of the connected regions meet the predetermined creep characteristics, thereby obtaining the creep information of the main shaft of the turbine.

[0032] Assume that a connected region with an area of ​​120 pixels and a perimeter of 40 pixels is found in the connected region analysis, and the offset range of the center position is within 6 pixels. According to the preset conditions, the connected region meets the conditions of creeping, and an alarm message is immediately generated. At the same time, the first image obtained and the image after creeping are displayed on the creeping alarm screen for the operation and maintenance personnel to make judgments.

[0033] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for detecting creep of a main shaft of a hydraulic turbine based on an improved three-frame difference method, characterized in that: The following steps are involved: S1. Obtaining the main axis image data: After the unit is shut down, collect the main axis image and obtain the latest three consecutive frames of original images in real time. , , ; S2. Grayscale the collected image to obtain three frames of grayscale images , , ; S3, performing region matching on the three grayscale images after grayscale processing to obtain a compensated image; S4. After obtaining the compensated image, the compensated images of two adjacent frames are respectively subjected to differential calculation to obtain two differential images. and ; S5. Perform threshold segmentation on the two difference images to obtain binary images. and ; S6, performing morphological dilation and corrosion processing on the binary image; S7, performing pixel-level difference calculation on the binary image processed in step S6 to obtain a difference image; S8, performing threshold processing on the difference image, and considering the pixel points whose difference values ​​exceed the threshold as the creeping area; S9, performing connected region analysis on the peristaltic region to find the boundary or center point of the connected region; S10. According to the characteristics of the connected regions and in combination with the actual creep characteristics, whether each connected region meets the creep condition is determined by analyzing whether the area, shape, and position of the connected regions meet the predetermined creep characteristics, thereby obtaining the creep information of the main shaft of the turbine.

2. The method for detecting creep of a main shaft of a hydraulic turbine based on an improved three-frame difference method according to claim 1 is characterized in that: In step S1, three consecutive frames of original images are obtained. , , In the image, each frame consists of three channels: R, G, and B. The mathematical expression is: ； in, represents the pixel in the i-th row and j-th column of the image, Contains the values ​​of the three channels R, G, and B: , , are the pixel values ​​of the R channel, G channel, and B channel in the i-th row and j-th column respectively.

3. The method for detecting creep of a main shaft of a hydraulic turbine based on an improved three-frame difference method according to claim 1 is characterized in that: In step S2, the grayscale processing is as follows: each channel of each frame image has a certain weight ratio, and the three-dimensional original image is merged into a one-dimensional grayscale image, and the expression is: ； in, represents a grayscale image, , and Represents the pixel values ​​of the R channel, G channel, and B channel.

4. The method for detecting creep of a main shaft of a hydraulic turbine based on an improved three-frame difference method according to claim 1 is characterized in that: In step S3, the processed three frames of video images are subjected to region matching, which includes the following steps: S3-1, in the previous frame image Select a module area and set the size to 32*32; S3-2, in the current frame image Search for the optimal area that can match the reference model in the previous frame image, and calculate the optimal matching point by the formula: ； in, and are the grayscale values ​​of the current frame and the previous frame respectively, m and n represent the corresponding matching pixel coordinates in images of different frames, is the minimum absolute error, 32*32 is the size of the selected area matching template, and The point corresponding to the minimum value of is the optimal matching point. S3-3. Calculate the relative displacement by comparing the position difference of the matching area in the current frame and the previous frame image. According to the obtained relative displacement, translate all the pixels in the current frame image by the relative displacement amount, and then recalculate the matching area between the translated image and the previous frame image to compensate and correct the current frame image.

5. The method for detecting creep of a main shaft of a hydraulic turbine based on an improved three-frame difference method according to claim 1, characterized in that: In step S4, the compensation images of two adjacent frames are respectively subjected to differential calculation to obtain two differential images. and , the calculation formula is: 。 6. The method for detecting creep of a main shaft of a hydraulic turbine based on an improved three-frame difference method according to claim 1, characterized in that: In step S5, the two difference images are and Perform threshold segmentation processing, the calculation formula is: ； in, is the binary image obtained after threshold segmentation. for and , represents the difference map calculated in step 5, and T is the set threshold.

7. The method for detecting creep of a main shaft of a hydraulic turbine based on an improved three-frame difference method according to claim 1, characterized in that: In step S6, the segmented binary image and Perform morphological expansion and corrosion processing, including the following steps: S6-1. First, perform a morphological dilation operation on the binary image. The calculation formula is: ； Among them, dilate() is a morphological expansion operation. and is the binary image after the dilation operation; S6-2, and then carry out corrosion treatment, the calculation formula is: ； Among them, erode() is a morphological erosion operation. and It is the binary image after the erosion operation.

Images