Intelligent lubricating oil liquid level accurate monitoring system based on image recognition

Abstract

The invention belongs to the technical field of intelligent monitoring and maintenance of industrial equipment, and particularly relates to an intelligent lubricating oil liquid level accurate monitoring system based on image recognition. The lubricating oil peephole is used for displaying the storage amount of lubricating oil, and the light source and image real-time capturing unit is used for capturing and transmitting images of the liquid level of the lubricating oil and transmitting the images to the central processing unit in real time for analysis; the image recognition and processing unit comprises an image preprocessing module which is responsible for enhancing image quality, and the image recognition and processing unit integrates a template matching algorithm and a deep learning model based on Deeplabv3 +, can adapt to changing illumination and environment conditions, and effectively analyzes and recognizes a liquid level image; the display terminal has the functions of data storage, early warning signal emission and data display. The accuracy and reliability of lubricating oil liquid level monitoring are improved, accurate real-time detection on the lubricating oil liquid level under variable environmental conditions is achieved, and the equipment failure risk caused by the abnormal lubricating oil liquid level can be reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent monitoring and maintenance of industrial equipment, and in particular relates to an intelligent lubricating oil level precision monitoring system based on image recognition. Background Art

[0002] In today's context of increasing industrialization and automation, accurate monitoring and maintenance of the operating status of key equipment has become an important task to ensure production efficiency and equipment safety. Especially in heavy industrial fields such as nuclear power plants and chemical plants, lubricating oil is a key material to ensure the normal operation of mechanical equipment. Accurate monitoring of its liquid level is essential to avoid equipment failure and extend equipment life. However, traditional lubricating oil level monitoring methods mostly rely on manual inspection or basic mechanical sensors, which are not only inefficient, but also have potential safety risks in complex or dangerous environments.

[0003] With the advancement of technology, image recognition and deep learning technologies have shown their strong application potential in many fields, providing a new solution to improve traditional monitoring methods. By using advanced image processing technology and machine learning models, automatic and accurate monitoring of lubricating oil levels can be achieved, thereby improving monitoring efficiency, reducing labor costs, and improving the operating safety of equipment to a certain extent, which is of great significance for promoting industrial automation and improving equipment management efficiency. Summary of the invention

[0004] The purpose of the present invention is to provide an intelligent lubricating oil level precision monitoring system based on image recognition. The system integrates advanced image recognition technology and deep learning algorithm to improve the accuracy and reliability of lubricating oil level monitoring, realize accurate real-time detection of lubricating oil level under changing environmental conditions, and reduce the risk of equipment failure caused by abnormal lubricating oil level.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] An intelligent lubricating oil level precision monitoring system based on image recognition comprises a lubricating oil peephole, a light source and a real-time image capture unit, an image recognition and processing unit, and a display terminal; the lubricating oil peephole is used to display the storage amount of the lubricating oil, the light source and the real-time image capture unit are used to capture and transmit images of the lubricating oil surface and transmit these images to a central processing unit for analysis in real time; the image recognition and processing unit comprises an image preprocessing module responsible for enhancing the image quality, the image recognition and processing unit integrates a template matching algorithm and a deep learning model based on Deeplabv3+, can adapt to changing lighting and environmental conditions, and effectively analyze and recognize liquid level images; the display terminal functions include data storage, warning signal issuance, and data display.

[0007] The lubricating oil peephole is made of a transparent or translucent material.

[0008] The image preprocessing module is responsible for enhancing image quality and improving image recognizability through contrast adjustment and illumination correction optimization processing; the image preprocessing module is used for data screening, selecting high-quality and accurate-viewing-angle images from the collected pictures for data annotation, and the annotation content includes the height of the lubricating oil level and the container edge information.

[0009] The display terminal stores the liquid level data obtained through the image recognition and processing unit, automatically issues warning signals according to the set thresholds, notifies operators to take necessary measures, and displays liquid level information through an intuitive user interface, including real-time data and historical trend analysis, providing users with comprehensive liquid level monitoring and management functions.

[0010] The light source and real-time image capture unit is used to capture the lubricating oil surface image illuminated by the light source and collect the camera screenshots of the lubricating oil peephole to form a data set.

[0011] The template matching algorithm is used to preliminarily locate the specific area in the lubricating oil container. The template matching algorithm is based on correlation analysis and locates specific features or objects by comparing the similarity between the input image and the pre-defined template. It is used to quickly locate the oil level gauge in the lubricating oil container. The template matching algorithm first selects a group of representative oil level gauge images to build a template library for subsequent image matching analysis. For each image to be analyzed, the correlation coefficient between it and each template in the template library is calculated to determine the matching result with the highest similarity, accurately locate the position of the oil level gauge, and provide a basis for liquid level recognition.

[0012] The template matching algorithm formula is:

[0013]

[0014] R(x,y): correlation coefficient at position (x,y), used to measure the similarity between the template and the image;

[0015] R(x,y): the pixel value of the input image at position (x,y);

[0016] The average pixel value of the input image;

[0017] T adj (x, y): The pixel value of the template image at position (x, y) adjusted to fit the specific conditions;

[0018] The average pixel value of the adjusted template image;

[0019] x',y': The offset relative to the (x,y) position during the summation.

[0020] The Deeplabv3+ model is refined to adapt to the specific characteristics of the lubricating oil container, including the gloss, color, and texture of the oil surface; the model structure formula is:

[0021]

[0022] S: Image segmentation result after processing by Deeplabv3+ model;

[0023] :Adjusted Deeplabv3+ model functions to process specific types of image data;

[0024] l loc : Image data extracted from the input image and located in a specific area.

[0025] The lubricant level is accurately calculated based on the output of the Deeplabv3+ model, which combines the convolutional neural network (CNN) and dilated convolution technology to understand complex scenes in the image while maintaining high resolution.

[0026] The calculation of the lubricating oil level first calculates the absolute height L of the liquid level, and then calculates the percentage P of the liquid level to the total height of the container. The formula for calculating the absolute height L of the liquid level and the percentage P of the liquid level is as follows:

[0027] The calculation formula of the absolute liquid level L is:

[0028] L=HD

[0029] The calculation formula of liquid level percentage P is:

[0030]

[0031] H: indicates the total height of the container.

[0032] D: indicates the distance from the liquid surface to the top of the container.

[0033] L: indicates the absolute height of the liquid level.

[0034] P: Indicates the percentage of liquid level to the total height of the container.

[0035] The beneficial effects achieved by the present invention are:

[0036] The implementation of the intelligent lubricating oil level recognition system of the present invention significantly improves the accuracy and efficiency of lubricating oil level monitoring in industrial equipment. By integrating advanced image recognition technology and deep learning algorithms, the system can accurately identify the liquid level under various lighting and environmental conditions, greatly reducing the risk of equipment failure caused by abnormal liquid levels. In addition, the system's automatic early warning function can promptly remind operators to take necessary measures, effectively avoiding potential operational interruptions and maintenance costs. The image preprocessing module further enhances the recognizability of the image, making liquid level detection more accurate and reliable.

[0037] The intuitive display and data management functions of the display terminal of the present invention enable operators to easily grasp the liquid level status and realize data-driven decision support. In addition, the flexibility and scalability of the system mean that it can be easily applied to different types of industrial environments, further promoting the development of intelligent manufacturing and industrial automation. In general, the present invention effectively addresses the limitations of traditional liquid level monitoring methods through technological innovation and provides an efficient and reliable solution for industrial equipment management. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a schematic diagram of an intelligent lubricating oil level accurate monitoring system based on image recognition;

[0039] Figure 2 Flow chart of the lubricating oil level monitoring method. DETAILED DESCRIPTION

[0040] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0041] An intelligent lubricating oil level recognition system combining image recognition technology and deep learning includes: a lubricating oil peephole, a light source and real-time image capture unit, an image recognition and processing unit, and a display terminal; the lubricating oil peephole is used to display the storage amount of lubricating oil, providing operators with an intuitive and convenient means of lubricating oil level monitoring. The peephole is located in a convenient position of the container and is made of transparent or translucent material to ensure that the lubricating oil level can be clearly observed in various working environments.

[0042] The light source and real-time image capture unit are used to capture and transmit images of the lubricating oil level and transmit these images to the central processor in real time for analysis. The light source part adopts a technology that can automatically adjust the brightness according to the ambient light to ensure that a uniform and clear image can be obtained under different lighting conditions.

[0043] The image recognition and processing unit first includes an image preprocessing module, which is responsible for enhancing image quality and improving image recognizability through optimization processing such as contrast adjustment and lighting correction; then, the unit integrates a template matching algorithm and a deep learning model based on Deeplabv3+. Through this integrated approach, the system can not only adapt to changing lighting and environmental conditions, but also effectively analyze and identify liquid level images with high accuracy.

[0044] The display terminal functions include data storage, warning signal issuance and data display. This terminal can not only safely store the liquid level data obtained through the image recognition and processing unit, but also automatically issue warning signals according to the set threshold value, so as to promptly notify the operator to take necessary measures. In addition, the terminal displays liquid level information through an intuitive user interface, including real-time data and historical trend analysis, thus providing users with comprehensive liquid level monitoring and management functions.

[0045] A liquid level detection method based on the device comprises the following steps:

[0046] The light source and real-time image capture unit is used to capture the lubricating oil surface image illuminated by the light source, mainly to collect camera screenshots of the lubricating oil peephole to form a data set.

[0047] The image preprocessing module of the image recognition and processing unit is used to enhance image quality and data screening, and select high-quality and accurate-viewing-angle images from the collected pictures for data annotation, and the annotation content includes key information such as the height of the lubricating oil level and the edge of the container.

[0048] The image recognition and processing unit includes a template matching algorithm, wherein the template matching algorithm is used to preliminarily locate a specific area in a lubricating oil container. The template matching algorithm used takes into account various possible placement angles and lighting conditions of the lubricating oil container to improve the adaptability and accuracy of the algorithm.

[0049] The template matching algorithm is mainly based on correlation analysis, and locates specific features or objects by comparing the similarity between the input image and the predefined template. In the present invention, this algorithm is used to quickly locate the oil level gauge in the lubricating oil container.

[0050] The template matching algorithm first selects a set of representative oil level gauge images to build a template library for subsequent image matching analysis. For each image to be analyzed, the system will calculate its correlation coefficient with each template in the template library to determine the matching result with the highest similarity. In this way, the system can accurately locate the approximate position of the oil level gauge and provide a basis for liquid level identification.

[0051] In order to further improve the robustness and applicability of the template matching algorithm, the present invention has comprehensively optimized the algorithm. This includes multi-angle and multi-scale analysis of the template to adapt to the different viewing angles and size changes that may occur in the oil level gauge, and adaptive improvements for different lighting conditions. These optimization measures significantly enhance the accuracy and stability of the algorithm, ensuring efficient liquid level detection in various environments.

[0052] The algorithm implementation formula is adjusted to:

[0053]

[0054] Where:

[0055] R(x,y): The correlation coefficient at position (x,y), used to measure the similarity between the template and the image.

[0056] R(x,y): The pixel value of the input image at position (x,y).

[0057] The average pixel value of the input image.

[0058] T adj (x,y): The pixel value at position (x,y) of the template image adjusted to fit the specific conditions.

[0059] The average pixel value of the adjusted template image.

[0060] x',y': The offset relative to the (x,y) position during the summation.

[0061] The image recognition and processing unit includes deep learning analysis based on the Deeplabv3+ model, which is refined to adapt to the specific characteristics of the lubricating oil container, including the gloss, color and texture of the oil surface.

[0062] The model structure adjustment formula is:

[0063]

[0064] Where:

[0065] S: Image segmentation result after processing by Deeplabv3+ model.

[0066] : Deeplabv3+ model functions adapted to process specific types of image data.

[0067] l loc : Image data extracted from the input image and located in a specific area.

[0068] The image recognition and processing unit includes the calculation of the lubricant level, which can be accurately calculated based on the output of the Deeplabv3+ model. The Deeplabv3+ model combines convolutional neural networks (CNN) and dilated convolution techniques to understand complex scenes in images while maintaining high resolution.

[0069] The calculation of the lubricating oil level first calculates the absolute height L of the liquid level, and then calculates the percentage P of the liquid level to the total height of the container, providing the user with intuitive liquid level information. Through this calculation method, the system can provide users with accurate lubricating oil level information under various environmental conditions, supporting effective equipment maintenance and operation decisions.

[0070] The formula for calculating the absolute height L of the liquid level and the liquid level percentage P is as follows:

[0071] The calculation formula of the absolute liquid level L is:

[0072] L=HD

[0073] The calculation formula of liquid level percentage P is:

[0074]

[0075] Where:

[0076] H: indicates the total height of the container.

[0077] D: indicates the distance from the liquid surface to the top of the container.

[0078] L: indicates the absolute height of the liquid level.

[0079] P: Indicates the percentage of liquid level to the total height of the container.

[0080] The display terminal integrates data storage, warning issuance and display, realizes safe storage of liquid level data, threshold-triggered warning and intuitive display, and ensures efficient liquid level monitoring and management.

[0081] The present invention provides an intelligent lubricating oil level precision monitoring system based on image recognition, which aims to achieve real-time and precise monitoring of the lubricating oil level in industrial equipment through advanced image processing technology and deep learning algorithm.

[0082] like Figure 1 As shown, the system mainly includes four parts: a lubricating oil peephole, a light source and real-time image capture unit, an image recognition and processing unit, and a display terminal.

[0083] The lubricant oil peephole is conveniently located on the container and is made of transparent or translucent material to ensure that the lubricant oil level can be clearly observed in various working environments.

[0084] The light source and real-time image capture unit is responsible for capturing images of the lubricating oil level and transmitting these images to the central processor for analysis in real time. The light source part uses technology that can automatically adjust the brightness according to the ambient light to ensure uniform and clear images under different lighting conditions.

[0085] The core function of the image recognition and processing unit lies in its advanced image preprocessing module, which is specially designed to enhance the quality of captured images, thereby providing higher-precision input data for subsequent image recognition and analysis. The main steps of image preprocessing include contrast adjustment, illumination correction, noise filtering, and edge enhancement, as follows:

[0086] The contrast adjustment makes the distinction between the lubricating oil level and the background more obvious. This process involves adjusting the dynamic range to ensure that important features in the image can be accurately identified.

[0087] The illumination correction step automatically adjusts the brightness and contrast of the image through an algorithm, compensating for image areas that are too dark or too bright, and ensuring the uniformity and visibility of the image.

[0088] The noise filtering can remove random noise in the image through various filtering techniques to improve the quality of the image.

[0089] The edge enhancement emphasizes the edges in the image, making the liquid level line and the container edge clearer.

[0090] After completing the image quality optimization, the image preprocessing module further performs data annotation operations on the image, accurately marking the height of the lubricating oil level, the edge of the container and other key features, and provides important reference data for subsequent image analysis and liquid level monitoring.

[0091] The image recognition and processing unit integrates a template matching algorithm to preliminarily locate a specific area within the lubricant container. The algorithm specifically considers the different possible placement angles and lighting conditions of the lubricant container, thereby improving the adaptability and accuracy of the algorithm.

[0092] The template matching algorithm is based on correlation analysis and achieves the location of a specific feature or object by comparing the similarity between the input image and the predefined template. In the present invention, the algorithm is used to quickly identify the position of the oil level gauge in the lubricating oil container.

[0093] The template matching algorithm selects a series of representative oil level gauge images to build a template library for subsequent image matching analysis. For each image to be analyzed, the system calculates its correlation coefficient with each template in the template library to identify the most similar match. This method enables the system to accurately estimate the position of the oil level gauge, laying the foundation for accurate identification of the liquid level.

[0094] The algorithm implementation formula is adjusted to:

[0095]

[0096] Where:

[0097] R(x,y): The correlation coefficient at position (x,y), used to measure the similarity between the template and the image.

[0098] R(x,y): The pixel value of the input image at position (x,y).

[0099] : The average pixel value of the input image.

[0100] T adj (x,y): The pixel value at position (x,y) of the template image adjusted to fit the specific conditions.

[0101] : The average pixel value of the adjusted template image.

[0102] x',y': The offset relative to the (x,y) position during the summation.

[0103] The image recognition and processing unit uses deep learning analysis based on the Deeplabv3+ model to refine specific features of the lubricant container, including the gloss, color, and texture of the oil surface. By adjusting the Deeplabv3+ model to adapt to these specific image features, the model's accuracy in monitoring the lubricant level is enhanced.

[0104] The model structure adjustment formula is defined as follows:

[0105]

[0106] Where:

[0107] S: Image segmentation result after processing by Deeplabv3+ model.

[0108] : Deeplabv3+ model functions adapted to process specific types of image data.

[0109] l loc: Image data extracted from the input image and located in a specific area.

[0110] The image recognition and processing unit is further responsible for calculating the lubricant level, using the data output by the Deeplabv3+ model to accurately determine the lubricant level. The Deeplabv3+ model combines convolutional neural networks (CNN) and dilated convolution techniques, allowing the model to accurately understand complex scenes in images while maintaining high resolution.

[0111] The calculation of the liquid level includes two steps: first, the absolute height L of the liquid level is calculated, and then the percentage P of the liquid level in the total height of the container is calculated to provide users with intuitive liquid level information. This calculation method enables the system to provide users with accurate lubricating oil level information under different environmental conditions, supporting effective equipment maintenance and operation decisions.

[0112] The calculation formulas for the absolute liquid level L and the liquid level percentage P are as follows:

[0113] L=HD

[0114]

[0115] Where:

[0116] H: indicates the total height of the container.

[0117] D: indicates the distance from the liquid surface to the top of the container.

[0118] L: indicates the absolute height of the liquid level.

[0119] P: Indicates the percentage of liquid level to the total height of the container.

[0120] The display terminal integrates data storage, warning issuance and data display functions, realizing safe storage of liquid level data, threshold trigger warning and intuitive display. This integrated design ensures the efficiency of liquid level monitoring and management, allowing operators to obtain liquid level information in a timely manner and take corresponding response measures.

[0121] The present invention discloses an intelligent lubricating oil level precision monitoring system based on image recognition. The system realizes real-time and precise monitoring of lubricating oil level in industrial equipment by integrating advanced image processing technology and deep learning algorithm. In complex working environments, such as changes in lighting and different container placement angles, the system can still accurately identify the lubricating oil level, significantly improving the accuracy and efficiency of monitoring.

[0122] In actual applications, when the lubricating oil level changes and may affect the normal operation of the equipment, the system can issue an early warning in time, provide accurate level information to operators, and support effective equipment maintenance and operation decisions. This not only avoids equipment failures caused by abnormal lubricating oil levels, but also significantly reduces maintenance costs and downtime, and improves production efficiency.

[0123] The integrated design of the display terminal of the present invention, including data storage, early warning issuance and intuitive display functions, ensures the high efficiency of liquid level monitoring management, enables operators to easily grasp the liquid level status, and realizes data-driven decision support. In short, the present invention effectively addresses the limitations of traditional liquid level monitoring methods through technological innovation and provides an efficient and reliable solution for industrial equipment management.

Claims

1. An intelligent lubricating oil level precision monitoring system based on image recognition, characterized in that: It includes a lubricating oil peephole, a light source and real-time image capture unit, an image recognition and processing unit, and a display terminal; the lubricating oil peephole is used to display the storage amount of lubricating oil, the light source and real-time image capture unit are used to capture and transmit images of the lubricating oil surface and transmit these images to the central processor for analysis in real time; the image recognition and processing unit includes an image preprocessing module, which is responsible for enhancing image quality; the image recognition and processing unit integrates a template matching algorithm and a deep learning model based on Deeplabv3+, which can adapt to changing lighting and environmental conditions and effectively analyze and identify liquid level images; the display terminal functions include data storage, warning signal issuance, and data display.

2. The intelligent lubricating oil level precise monitoring system based on image recognition according to claim 1 is characterized in that: The lubricating oil peephole is made of a transparent or translucent material.

3. The intelligent lubricating oil level precise monitoring system based on image recognition according to claim 1 is characterized in that: The image preprocessing module is responsible for enhancing image quality and improving image recognizability through contrast adjustment and illumination correction optimization processing; the image preprocessing module is used for data screening, selecting high-quality and accurate-viewing-angle images from the collected pictures for data annotation, and the annotation content includes the height of the lubricating oil level and the container edge information.

4. The intelligent lubricating oil level precise monitoring system based on image recognition according to claim 1 is characterized in that: The display terminal stores the liquid level data obtained through the image recognition and processing unit, automatically issues warning signals according to the set thresholds, notifies operators to take necessary measures, and displays liquid level information through an intuitive user interface, including real-time data and historical trend analysis, providing users with comprehensive liquid level monitoring and management functions.

5. The intelligent lubricating oil level precise monitoring system based on image recognition according to claim 1 is characterized in that: The light source and real-time image capture unit is used to capture the lubricating oil surface image illuminated by the light source and collect the camera screenshots of the lubricating oil peephole to form a data set.

6. The intelligent lubricating oil level precise monitoring system based on image recognition according to claim 1 is characterized in that: The template matching algorithm is used to preliminarily locate the specific area in the lubricating oil container. The template matching algorithm is based on correlation analysis and locates specific features or objects by comparing the similarity between the input image and the pre-defined template. It is used to quickly locate the oil level gauge in the lubricating oil container. The template matching algorithm first selects a group of representative oil level gauge images to build a template library for subsequent image matching analysis. For each image to be analyzed, the correlation coefficient between it and each template in the template library is calculated to determine the matching result with the highest similarity, accurately locate the position of the oil level gauge, and provide a basis for liquid level recognition.

7. The intelligent lubricating oil level precise monitoring system based on image recognition according to claim 6 is characterized by: The template matching algorithm formula is: R(x,y): correlation coefficient at position (x,y), used to measure the similarity between the template and the image; R(x,y): the pixel value of the input image at position (x,y); The average pixel value of the input image; T adj (x, y): The pixel value of the template image at position (x, y) adjusted to fit the specific conditions; The average pixel value of the adjusted template image; x',y': The offset relative to the (x,y) position during the summation.

8. The intelligent lubricating oil level precise monitoring system based on image recognition according to claim 1 is characterized in that: The Deeplabv3+ model is refined to adapt to the specific characteristics of the lubricating oil container, including the gloss, color, and texture of the oil surface; the model structure formula is: S: Image segmentation result after processing by Deeplabv3+ model; Deeplabv3+ model functions adapted to process specific types of image data; l loc : Image data extracted from the input image and located in a specific area.

9. The intelligent lubricating oil level precise monitoring system based on image recognition according to claim 8 is characterized in that: The lubricant level is accurately calculated based on the output of the Deeplabv3+ model, which combines the convolutional neural network (CNN) and dilated convolution technology to understand complex scenes in the image while maintaining high resolution.

10. The intelligent lubricating oil level precise monitoring system based on image recognition according to claim 9 is characterized in that: The calculation of the lubricating oil level first calculates the absolute height L of the liquid level, and then calculates the percentage P of the liquid level to the total height of the container. The formula for calculating the absolute height L of the liquid level and the percentage P of the liquid level is as follows: The calculation formula of the absolute liquid level L is: L=HD The calculation formula of liquid level percentage P is: H: indicates the total height of the container. D: indicates the distance from the liquid surface to the top of the container. L: indicates the absolute height of the liquid level. P: Indicates the percentage of liquid level to the total height of the container.

Images