Mechanical seal leakage online monitoring and failure early warning integrated system

Abstract

The invention belongs to the technical field of mechanical seal leakage monitoring, and particularly relates to a mechanical seal leakage online monitoring and failure early warning integrated method, which comprises the following steps: closing a to-be-detected valve, putting the to-be-detected valve into water, and ventilating the to-be-detected valve through an inflation source to obtain a video image of the to-be-detected valve in water; constructing a dynamic background model based on the acquired video image; extracting a foreground image based on the obtained dynamic background model; inputting the extracted foreground image into a deep neural network to determine bubbles, and outputting a determination result; according to the method, the video image is acquired, the foreground image of the bubbles is extracted, and the extracted foreground image is identified by using the neural network, so that misjudgment and missed judgment caused by manual work can be avoided, and the accuracy of sealing monitoring is improved.

Description

Technical Field

[0001] This application belongs to the field of sealing leakage monitoring technology, specifically relating to an integrated system for online monitoring and failure early warning of mechanical seal leakage. Background Technology

[0002] Air tightness refers to the ability of an engineering structure or equipment to effectively prevent working fluids such as gases or liquids from passing through its surface or interior. Air tightness testing is an extremely important quality control process in modern industry. By studying air tightness testing technology, the safety, reliability, and performance of products can be improved, while reducing enterprise costs and enhancing market competitiveness. With the widespread use of high-tech products and precision instruments, air tightness testing is indispensable in ensuring product quality.

[0003] The water immersion bubble method is a low-cost method for testing airtightness, and it is effective in detecting leaks and pinpointing their location, making it the method used by the vast majority of companies. However, this traditional method has a relatively low degree of automation, requiring manual observation for bubbles. Manual inspection is subject to many unstable factors, such as eye fatigue, differences in vision among workers, and small bubbles appearing when the leak is small, leading to misjudgments and missed detections of airtightness by workers. Summary of the Invention

[0004] To address the technical problems mentioned above, this invention provides an integrated method for online monitoring and failure early warning of mechanical seal leakage. By acquiring video images and extracting the foreground image of air bubbles, a neural network is used to identify the extracted foreground image, which helps to avoid misjudgments and omissions caused by human error and improves the accuracy of sealing monitoring.

[0005] In a first aspect, the present invention provides an integrated method for online monitoring and failure early warning of mechanical seal leakage, comprising the following steps: Step 1: Close the valve to be tested and place it in water, and then ventilate the valve to be tested using an air source; Step 2: Acquire video images of the valve to be tested in the water; Step 3: Construct a dynamic background model based on the acquired video images; Step 4: Extract the foreground image based on the acquired dynamic background model; Step 5: Input the extracted foreground image into a deep neural network to determine the bubble and output the determination result; Step four includes: Acquire frames at time t, preliminarily determine the background point in the frames at time t based on the dynamic background model, and obtain the coordinates of the background point, as well as four consecutive frames at time t. , , and ; The inter-frame difference results are obtained by pairwise difference analysis of four consecutive frames. , and And the inter-frame difference results , and Perform grayscale transformation and binarization; The difference results of the morphological operation are ANDed using the morphological operation rules to determine the pixel region of the moving bubble; Determine whether the background points in the frame image at time t are in the moving region, and based on the determination result, redetermine the foreground and background points.

[0006] Further, the step of determining whether the background points in the frame image at time t are in the motion region, and re-determining the foreground and background points based on the determination result, includes: Based on the determined coordinates of the background points in the frame image at time t, determine whether the value of each background point in the image at time t is 1 in the first region. If so, the background point is still determined as a background point. If not, then determine whether the value of the background point in the second region is 1; if so, then determine whether the absolute difference between the grayscale value of the background point at time t and the grayscale value at time t+1 is greater than the threshold. If yes, then the background point is judged as the foreground point; otherwise, the background point is still judged as the background point.

[0007] Furthermore, step three includes: Take N consecutive frames from the acquired video images, with fixed points. The gray values ​​in the N-frame image sequence are respectively Therefore, the average gray value of each pixel in the N-frame sequence image is calculated respectively; The grayscale value of a new frame image is recorded as "1" or "0" depending on whether it is greater than the average grayscale value, and then combined into 16-bit hexadecimal data in a specific order. Starting from the second frame of the acquired N-frame image sequence, compare the hash value sequence of the previous frame frame by frame. If the Hamming distance between two frames is zero, they are marked as belonging to the same group; otherwise, they are considered as a new group. Finally, the N-frame image sequence is divided into n groups. , where n is a positive integer less than or equal to N; The starting frame image The hash values ​​are grouped as (in, [ ]); and the starting frame image Each group is associated with its respective hash group and The two most recent frames in and The difference images obtained by performing the difference operation are then subjected to a logical AND operation to obtain the intersection image of the two difference images. ; After thresholding using the Otsu algorithm, binarization is performed to obtain the difference image.

[0008] The starting frame image is acquired, and thresholding is performed using the Otsu algorithm to obtain salient images. ; Difference image and significant images Perform a logical OR operation to obtain the true target region image. ; Extract the moving target image from the starting frame ; The identified real target area Set as a valid region and initialize. The corresponding pixels are not the background. The corresponding pixel is used as the background, and then the remaining background pixels are used to fill the starting frame image.

[0009] Furthermore, step five includes: The extracted foreground image is input into the backbone convolutional neural network to obtain the feature vector of the bubble candidate region sub-image; A fully connected neural network is used to decode the feature vectors of the bubble candidate region sub-images, determine whether they are bubble images, and provide the bubble bounding box positions estimated by the network. Redundant detection boxes for the same bubble are merged using a non-maximum suppression algorithm to avoid multiple detection boxes appearing for the same bubble; Output the bubble recognition results.

[0010] Secondly, the present invention provides a computer-readable storage medium, which includes a stored program that, when the program is running, controls the power equipment where the computer-readable storage medium is located to execute the above-described integrated method for online monitoring and failure early warning of mechanical seal leakage.

[0011] The beneficial effects of this invention are as follows: This invention acquires video images, extracts the foreground image of the bubble, and uses a neural network to identify the extracted foreground image, which helps to avoid misjudgments and omissions caused by human error and improves the accuracy of sealing monitoring. Attached Figure Description

[0012] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0013] Figure 1 This is a flowchart of the integrated method for online monitoring and failure early warning of mechanical seal leakage according to the present invention; Figure 2 This is a schematic diagram of the mechanical seal leakage monitoring structure of the present invention.

[0014] Figure 3 This is a structural diagram of the video image acquisition of the present invention. Detailed Implementation

[0015] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0016] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, each technical and scientific term used in these embodiments has the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0017] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0018] In this invention, terms such as "upper," "lower," "left," "right," "front," "back," "vertical," "horizontal," "side," and "bottom" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used only to facilitate the description of the structural relationships of the various components or elements of this invention and do not specifically refer to any component or element in this invention. They should not be construed as limiting the invention.

[0019] In this invention, terms such as "fixed connection," "connected," and "linked" should be interpreted broadly, indicating a fixed connection, an integral connection, or a detachable connection; a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can determine the specific meaning of these terms in this invention based on the specific circumstances, and they should not be construed as limitations on the invention.

[0020] Example 1: like Figure 1 As shown in the figure, this embodiment provides a method for online monitoring and early warning of mechanical seal leakage, including the following steps: S1: Close the valve to be tested and place it in water, and then ventilate the valve to be tested through an air source; Specifically, such as Figure 2 As shown, close the valve to be tested and connect the inlet to the air source to supply air to the valve to be tested, while immersing the valve to be tested completely in water; S2: Acquire video images of the valve to be inspected in the water; Specifically, such as Figure 3 As shown, video images of the valve to be tested in water are obtained by shooting from the front with backlight illumination; S3: Construct a dynamic background model based on the acquired video images; Includes the following steps: A1: Extract a sequence of N consecutive frames from the acquired video images, with fixed points. The gray values ​​in the N-frame image sequence are respectively Therefore, the average gray value of each pixel in the N-frame sequence image is calculated respectively; in, This is the starting frame of an N-frame image sequence.

[0021] A2: Record the grayscale value of the new frame image as "1" or "0" based on whether it is greater than the average grayscale value, and combine them into 16-bit hexadecimal data in a specific order; A3: Starting from the second frame of the acquired N-frame image sequence, compare the hash value sequence of the previous frame frame by frame. If the Hamming distance between two frames is zero, they are marked as belonging to the same group; otherwise, they are considered as a new group. Finally, the N-frame image sequence is divided into n groups. , where n is a positive integer less than or equal to N. A4: Set the starting frame image The hash values ​​are grouped as (in, [ ]); and the starting frame image Each group is associated with its respective hash group and The two most recent frames in and The difference images obtained by performing the difference operation are then subjected to a logical AND operation to obtain the intersection image of the two difference images. The formula is as follows: , ; ; Where p is a pre-defined threshold value range for hash values ​​related to the target's speed. Generally, the faster the target moves, the larger this threshold p becomes. An upper limit threshold is also set. and lower threshold To ensure the algorithm has good robustness, i.e. .

[0022] A5: After thresholding using the Otsu algorithm, binarization is performed to obtain the difference image. The formula is as follows: ; in, This represents the operation of binarization using a threshold.

[0023] A6: Obtain the starting frame image and perform thresholding using the Otsu algorithm to obtain salient images. ; A7: Difference image and significant images Perform a logical OR operation to obtain the true target region image. ,Right now .

[0024] A8: Extract the moving target image from the starting frame. ; A9: The identified real target area Set as a valid region and initialize. The corresponding pixels are not the background. The corresponding pixel is set as the background, and then the remaining background pixels are filled into the starting frame image; S4: Extract the foreground image based on the acquired dynamic background model; B1: Acquire frame images at time t, preliminarily determine the background point in the frame images at time t based on the dynamic background model, and obtain the coordinates of the background point, as well as four consecutive frames at time t. , , and .

[0025] B2: Perform pairwise differences on four consecutive frames to obtain three inter-frame difference results. , and And the inter-frame difference results , and Perform grayscale transformation and binarization; The specific calculation formula is as follows: ; ; ; in, It is the threshold used for differential binarization.

[0026] B3: Perform an AND operation on the difference results using morphological algorithms to determine the pixel region of the moving bubble; the specific formula is shown below: ; ; in, and The region equal to 1 is the region where the moving bubble pixels are located.

[0027] For ease of explanation, Designated as the first area, This is designated as the second region.

[0028] B4: Determine whether the background points in the frame image at time t are in the motion region, and redetermine the foreground and background points based on the determination result.

[0029] Specifically, it includes the following steps: B4-1: Based on the determined background point coordinates of the frame image at time t, determine the first region for each background point in the image at time t. If the value in is 1, then the background point will still be judged as a background point. B4-2: If not, then determine that the background point is in the second region. If the value in the value field is 1, then determine whether the absolute difference between the grayscale value of the background point at time t and the grayscale value at time t+1 is greater than the threshold. If yes, then the background point is judged as the foreground point; otherwise, the background point is still judged as the background point.

[0030] B5: Extract the foreground image based on the redefined foreground and background points.

[0031] S5: Input the extracted foreground image into a deep neural network to identify bubbles and output the recognition results.

[0032] Specifically, it includes the following steps: S5-1: Input the extracted foreground image into the trained MobileNetV3 backbone convolutional neural network to obtain the feature vector of the bubble candidate region sub-image; S5-2: A fully connected neural network is used to decode the feature vectors of the bubble candidate region sub-image, determine whether it is a bubble image, and give the location of the bubble bounding box estimated by the network. S5-3: Redundant detection boxes for the same bubble are merged using a non-maximum suppression algorithm to avoid multiple detection boxes appearing for the same bubble; S5-4: Output bubble recognition results.

[0033] Example 2: This embodiment provides a computer-readable storage medium including a stored program, wherein, when the program is executed, the computer-readable storage medium controls the execution of the operation method of the compressed air energy storage system in a medium-deep saline aquifer described in Embodiment 2.

[0034] The same or similar parts between the various embodiments in this specification can be referred to mutually. In particular, the terminal embodiments are basically similar to the method embodiments, so the description is relatively simple, and the relevant parts can be referred to the description in the method embodiments.

[0035] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or units may be electrical, mechanical, or other forms.

[0036] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0037] Additionally, it should be noted that the flowcharts in the accompanying drawings illustrate methods according to embodiments of this disclosure. In the descriptions corresponding to the flowcharts or block diagrams in the drawings, the operations or steps corresponding to different blocks may occur in a different order than disclosed in the description; sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps may actually be executed substantially in parallel, or sometimes in reverse order, depending on the function involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions.

[0038] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An integrated method for online monitoring and failure early warning of mechanical seal leakage, characterized in that, Includes the following steps: Step 1: Close the valve to be tested and place it in water, and then ventilate the valve to be tested using an air source; Step 2: Acquire video images of the valve to be tested in the water; Step 3: Construct a dynamic background model based on the acquired video images; Step 4: Extract the foreground image based on the acquired dynamic background model; Step 5: Input the extracted foreground image into a deep neural network to determine the bubble and output the determination result; Step four includes: Acquire frames at time t, preliminarily determine the background point in the frames at time t based on the dynamic background model, and obtain the coordinates of the background point, as well as four consecutive frames at time t. , , and ; The inter-frame difference results are obtained by pairwise difference analysis of four consecutive frames. , and And the inter-frame difference results , and Perform grayscale transformation and binarization; The difference results of the morphological operation are ANDed using the morphological operation rules to determine the pixel region of the moving bubble; Determine whether the background points in the frame image at time t are in the motion region. Based on the determination result, redetermine the foreground and background points, and obtain the foreground image based on the determined foreground and background points.

2. The integrated method for online monitoring and failure early warning of mechanical seal leakage according to claim 1, characterized in that, The step of determining whether the background points in the frame image at time t are in the motion region, and re-determining the foreground and background points based on the determination result, includes: Based on the determined coordinates of the background points in the frame image at time t, determine whether the value of each background point in the image at time t is 1 in the first region. If so, the background point is still determined as a background point. If not, then determine whether the value of the background point in the second region is 1; if so, then determine whether the absolute difference between the grayscale value of the background point at time t and the grayscale value at time t+1 is greater than the threshold. If yes, then the background point is judged as the foreground point; otherwise, the background point is still judged as the background point.

3. The integrated method for online monitoring and failure early warning of mechanical seal leakage according to claim 1, characterized in that, Step three includes: Take N consecutive frames from the acquired video images, with fixed points. The gray values ​​in the N-frame image sequence are respectively Therefore, the average gray value of each pixel in the N-frame sequence image is calculated respectively; The grayscale value of a new frame image is recorded as "1" or "0" depending on whether it is greater than the average grayscale value, and then combined into 16-bit hexadecimal data in a specific order. Starting from the second frame of the acquired N-frame image sequence, compare the hash value sequence of the previous frame frame by frame. If the Hamming distance between two frames is zero, they are marked as belonging to the same group; otherwise, they are considered as a new group. Finally, the N-frame image sequence is divided into n groups. , where n is a positive integer less than or equal to N; The starting frame image The hash values ​​are grouped as (in, [ ]); and the starting frame image Each group is associated with its respective hash group and The two most recent frames in and The difference images obtained by performing the difference operation are then subjected to a logical AND operation to obtain the intersection image of the two difference images. ; After thresholding using the Otsu algorithm, binarization is performed to obtain the difference image. The starting frame image is acquired, and thresholding is performed using the Otsu algorithm to obtain salient images. ; Difference image and significant images Perform a logical OR operation to obtain the true target region image. ; Extract the moving target image from the starting frame ; The identified real target area Set as a valid region and initialize. The corresponding pixels are not the background. The corresponding pixel is used as the background, and then the remaining background pixels are used to fill the starting frame image.

4. The integrated method for online monitoring and failure early warning of mechanical seal leakage according to claim 1, characterized in that, Step five includes: The extracted foreground image is input into the backbone convolutional neural network to obtain the feature vector of the bubble candidate region sub-image; A fully connected neural network is used to decode the feature vectors of the bubble candidate region sub-images, determine whether they are bubble images, and provide the bubble bounding box positions estimated by the network. Redundant detection boxes for the same bubble are merged using a non-maximum suppression algorithm to avoid multiple detection boxes appearing for the same bubble; Output the bubble recognition results.

5. A computer-readable storage medium comprising a stored program, characterized in that, During program execution, the power equipment containing the computer-readable storage medium is controlled to perform the integrated method for online monitoring and failure early warning of mechanical seal leakage as described in any one of claims 1 to 4.

Images