Valve visual monitoring system and method

Abstract

The valve visual monitoring system comprises a valve, a driving device, an image collecting device and a data processing device, the valve is used for being installed on a pipeline, the valve is used for opening and closing the pipeline and adjusting the flow of a medium in the pipeline, and a first mark is arranged on the valve; the driving device is used for being connected to a pipeline and drives the valve to move between the opening position and the closing position. The image acquisition device acquires image information of the valve and / or the driving device, the image acquisition device comprises a color image acquisition module and a depth image acquisition module, the color image acquisition module acquires color image information of the valve and / or the driving device, and the depth image acquisition module acquires depth image information of the valve and / or the driving device; the data processing device receives and analyzes the image information and obtains the state of the valve according to the image information. The invention further discloses a valve visual monitoring method.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment status monitoring, and particularly to a valve visual monitoring system and method. Background Art

[0002] In modern industrial production and construction processes, valves are often used to control the direction, pressure, and flow rate of fluids in fluid systems, enabling the medium (including at least one of liquid, gas, and powder) in the piping and equipment to flow or stop and controlling its flow rate. Monitoring whether the valve is closed and the magnitude of the flow rate through the valve is of great significance for ensuring the safety of the medium.

[0003] However, since many valves are installed inside equipment or in harsh environments such as underground, high temperature, and high pressure, people cannot directly monitor the valve status; in the prior art, sensors are generally used to detect the valve status, but in environments such as high temperature, high pressure, strong corrosion, and high humidity, the performance of the sensors may be affected or even malfunction; moreover, some types of sensors are also affected by external magnetic fields, thus affecting the accuracy and reliability of detection.

[0004] Therefore, the present invention provides a valve visual monitoring system and method, which can effectively solve the above problems, has high environmental adaptability and strong anti-interference ability, and can provide images for easy fault diagnosis and analysis. By analyzing multi-modal visual information and cooperating with the markings set on the valve, the accuracy of valve status determination is improved. Summary of the Invention

[0005] In order to overcome the deficiencies of the prior art, the present invention provides a valve visual monitoring system and method, which has high environmental adaptability and strong anti-interference ability, and can provide images for easy fault diagnosis and analysis. By analyzing multi-modal visual information and cooperating with the markings set on the valve, the accuracy of valve status determination is improved.

[0006] The present invention discloses a valve visual monitoring system, including:

[0007] A valve, which is used to be installed on a pipeline, and is used to open and close the pipeline and adjust the flow rate of the medium in the pipeline. Among them, a first marking is provided on the valve;

[0008] A driving device, which is used to be connected to the pipeline, and the driving device drives the valve to move between an open position and a closed position;

[0009] An image acquisition device, which acquires image information of the valve and / or the driving device. Among them, the image acquisition device includes a color image acquisition module and a depth image acquisition module. The color image acquisition module acquires color image information of the valve and / or the driving device, and the depth image acquisition module acquires depth image information of the valve and / or the driving device;

[0010] A data processing device, which receives and analyzes the image information and obtains the state of the valve based on this.

[0011] As an improvement of the present invention, the data processing device includes a feature extraction module, a feature fusion module, and a simulation recognition module. The feature extraction module extracts various types of image features according to the image information. The feature fusion module fuses various types of image features at the same time to form a comprehensive feature vector. The simulation recognition module arranges the comprehensive feature vectors in chronological order to form a dynamic feature sequence, and uses a hidden Markov model to analyze the state of the valve according to the dynamic feature sequence.

[0012] As an improvement of the present invention, the feature extraction module includes a first feature extraction unit and a second feature extraction unit. The first feature extraction unit extracts the shape features and texture features of the valve and / or the driving device according to the color image information, and the second feature extraction unit extracts the spatial position features and depth distribution features of the valve and / or the driving device according to the depth image information.

[0013] As an improvement of the present invention, the feature fusion module includes a splicing fusion unit and a dimensionality reduction processing unit. The splicing fusion unit fuses various types of features at the same time to form a comprehensive feature vector, and the dimensionality reduction processing unit performs dimensionality reduction processing on the comprehensive feature vector.

[0014] As an improvement of the present invention, the simulation recognition module includes a dynamic feature construction unit and a model recognition unit. The dynamic feature construction unit arranges the comprehensive feature vectors in chronological order to form a dynamic feature sequence, and the model recognition unit trains a model according to past dynamic feature sequence samples, and uses the model to analyze the state of the valve according to the current dynamic feature sequence.

[0015] As an improvement of the present invention, the dimensionality reduction processing unit uses the principal component analysis algorithm to perform dimensionality reduction processing on the comprehensive feature vector.

[0016] As an improvement of the present invention, the first feature extraction unit extracts the contour edges of the valve and / or the driving device by using an edge detection algorithm, and then identifies the shape features of the valve and / or the driving device by using a Hough transform shape analysis method. The first feature extraction unit calculates the pixel gray level change rules of the color image in different directions and distances by using a gray level co-occurrence matrix algorithm, so as to obtain texture features.

[0017] As an improvement of the present invention, the second feature extraction unit calculates the three-dimensional coordinate information of the whole and each part of the valve and / or the driving device according to the depth image information, and analyzes the distribution of the whole and each part of the valve and / or the driving device in the depth direction.

[0018] ]>As an improvement of the present invention, the valve includes a valve body, a ball valve, a valve stem and a driving rod. The ball valve is arranged in the valve body, the valve stem passes through the valve body and is connected to the ball valve, one end of the driving rod is connected to the driving device, the other end of the driving rod is connected to the valve stem, and the driving device drives the driving rod to rotate, thereby driving the valve stem and the ball valve to rotate. The first identifier is arranged on the driving rod and / or at a position of the valve body close to the driving rod.

[0019] As an improvement of the present invention, the driving device includes an oil cylinder, a piston and a piston rod. The piston slides in the oil cylinder under the action of hydraulic oil, one end of the piston rod is connected to the piston, the other end of the piston rod is connected to the driving rod, and a second identifier is arranged on the surface of the piston rod.

[0020] The present invention also discloses a method for visually monitoring a valve, including:

[0021] Using an image acquisition device to acquire image information of the valve and / or the driving device. The image acquisition device includes a color image acquisition module and a depth image acquisition module. The color image acquisition module acquires color image information of the valve and / or the driving device, and the depth image acquisition module acquires depth image information of the valve and / or the driving device;

[0022] Using a feature extraction module to extract various image features according to the image information;

[0023] Using a feature fusion module to fuse various image features at the same time to form a comprehensive feature vector; <OPTIONAL>

[0024] Using an analog recognition module to arrange the comprehensive feature vectors in chronological order to form a dynamic feature sequence, and using a hidden Markov model to analyze the state of the valve according to the dynamic feature sequence.

[0025] The beneficial effects of the present invention are as follows: Through the above structure, during use, the color image acquisition module and the depth image acquisition module respectively acquire visual information of different modalities. The data processing device receives visual information of multiple modalities and comprehensively judges the state of the valve, improving the accuracy of judgment. Preferably, other modality image acquisition modules can also be added, such as an infrared camera, a light field camera, etc., to further increase visual information of other modalities and improve the accuracy of judging the state of the valve. For example, the infrared camera can detect whether there is heat generation due to excessive stress at the valve and whether there is medium leakage, etc. Preferably, the state of the valve can also be further assisted in judgment by means of the first identifier provided on the valve. For example, by setting the first identifier on the driving rod of the valve, it can make it easier for the data processing device to determine the position of the driving rod when analyzing the image information, and thus understand the opening state of the valve. For another example, multiple fan-shaped partitions can be set with the valve stem as the center, and the colors of each fan-shaped partition are different, so that it is easier for the data processing device to determine which partition the driving rod is in when analyzing the image information, and thus understand the opening state of the valve. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for use in the description of the embodiments. The following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0027] The present invention will be further described below in conjunction with the drawings and embodiments.

[0028] Figure 1 It is a schematic diagram of the module connection of the present invention;

[0029] Figure 2 It is a schematic diagram of the use state of the present invention;

[0030] Figure 3 It is a schematic diagram of the feature extraction module of the present invention;

[0031] Figure 4 It is a schematic diagram of the feature fusion module of the present invention; [[ID=##]]

[0032] Figure 5 It is a schematic diagram of the module recognition module of the present invention;

[0033] Figure 6 It is a schematic diagram of the valve and the driving device of the present invention in the valve open state;

[0034] Figure 7 It is a schematic diagram of the valve and the driving device of the present invention in the valve closed state;

[0035] Figure 8It is a cross-sectional view of the valve and drive device of the present invention in the valve open state;

[0036] Figure 9 It is a cross-sectional view of the valve and drive device of the present invention in the valve closed state;

[0037] Figure 10 It is a flowchart of the valve visual monitoring method of the present invention. Detailed implementation manners

[0038] To make the above objects, features, and advantages of the present application more obvious and understandable, the following will describe the detailed implementation manners of the present application in conjunction with the accompanying drawings. Many specific details are set forth in the following description to fully understand the present application. However, the present application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without departing from the connotation of the present application. Therefore, the present application is not limited by the specific embodiments disclosed below.

[0039] In the description of the present application, it should be understood that if terms such as "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc. appear, the orientation or positional relationship indicated by these terms is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation of the present application.

[0040] In addition, if terms such as "first" and "second" appear, these terms are only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include at least one of such features. In the description of the present application, if the term "plurality" appears, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise specifically defined.

[0041] In this application, unless otherwise clearly defined and limited, if terms such as "installed", "connected", "linked", "fixed", etc. appear, these terms should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or integrated; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two components or the interaction relationship between two components, unless otherwise clearly defined. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0042] In this application, unless otherwise clearly defined and limited, if there is a description such as a first feature being "on" or "under" a second feature, its meaning can be that the first and second features are in direct contact, or the first and second features are indirectly in contact through an intermediate medium. Moreover, the first feature being "above", "over" and "on top of" the second feature can mean that the first feature is directly above or obliquely above the second feature, or simply means that the first feature is at a higher horizontal level than the second feature. The first feature being "under", "beneath" and "underneath" the second feature can be that the first feature is directly below or obliquely below the second feature, or simply means that the first feature is at a lower horizontal level than the second feature.

[0043] It should be noted that if an element is referred to as "fixed to" or "disposed on" another element, it can be directly on the other element or there can also be an intermediate element. If an element is considered to be "connected" to another element, it can be directly connected to the other element or there may be an intermediate element at the same time. If so, the terms "vertical", "horizontal", "up", "down", "left", "right" and similar expressions used in this application are only for the purpose of illustration and do not represent the only implementation.

[0044] Refer to Figures 1 to 9 , a valve visual monitoring system, comprising:

[0045] A valve 100, the valve 100 is used to be installed on a pipeline 200, the valve 100 is used to open and close the pipeline 200 and adjust the flow rate of the medium in the pipeline 200, wherein, a first identifier 101 is provided on the valve 100;

[0046] A driving device 300, the driving device 300 is used to be connected to the pipeline 200, and the driving device 300 drives the valve 100 to move between an open position and a closed position;

[0047] An image acquisition device 400, the image acquisition device 400 acquires image information of the valve 100 and / or the driving device 300, wherein the image acquisition device 400 includes a color image acquisition module 410 and a depth image acquisition module 420, the color image acquisition module 410 acquires color image information of the valve 100 and / or the driving device 300, and the depth image acquisition module 420 acquires depth image information of the valve 100 and / or the driving device 300;

[0048] A data processing device 500, the data processing device 500 receives and analyzes the image information and thereby obtains the state of the valve 100.

[0049] With the above structure, in use, the color image acquisition module 410 and the depth image acquisition module 420 respectively acquire visual information of different modalities. The data processing device 500 receives visual information of multiple modalities and comprehensively judges the state of the valve 100 to improve the accuracy of judgment; preferably, other modality image acquisition modules can be added, such as an infrared camera, a light field camera, etc., to further increase visual information of other modalities and improve the accuracy of judging the state of the valve 100. For example, the infrared camera can detect whether there is heat generation due to excessive stress at the valve or whether there is medium leakage, etc.; preferably, the state of the valve 100 can be further assisted in judgment by means of the first identifier 101 provided on the valve 100. For example, by setting the first identifier 101 on the driving rod 113 of the valve 100, the data processing device 500 can more easily determine the position of the driving rod 113 when analyzing the image information, and thus know the opening state of the valve 100; for another example, multiple fan-shaped partitions can be set with the valve stem 112 as the center, and the colors of each fan-shaped partition are different, so that the data processing device 500 can more easily determine which partition the driving rod 113 is in when analyzing the image information, and thus know the opening state of the valve 100.

[0050] In this embodiment, the data processing device 500 includes a feature extraction module 510, a feature fusion module 520, and a simulation recognition module 530. The feature extraction module 510 extracts various types of image features according to the image information. The feature fusion module 520 fuses various types of image features at the same time to form a comprehensive feature vector. The simulation recognition module 530 arranges the comprehensive feature vector in chronological order to form a dynamic feature sequence, and uses a hidden Markov model to analyze the state of the valve 100 according to the dynamic feature sequence. Through the setting of the above structure, the feature extraction module 510 can specifically extract various types of image features according to the image information, such as the relative position and angle between the valve body 110 and the drive rod 113 of the valve 100, scratches and corrosion on the valve surface, etc., excluding useless information and improving the analysis accuracy and speed. The feature fusion module 520 fuses various types of image features at the same time to form a comprehensive feature vector, ensuring the unity in time, so as to understand various types of image features of the valve 100 at the same time, facilitating subsequent classification or recognition tasks and improving the analysis accuracy. Based on a large number of dynamic feature sequence samples of different states of the valve, such as fully closed, partially opened, fully opened, and faulty, etc., train the HMM (hidden Markov model). The HMM can model the feature sequence with time series, and realize the classification and recognition of the valve state by analyzing the transition probability between hidden states in the sequence and the emission probability of observed features under each hidden state. For example, when the valve 100 starts to open gradually from the fully closed state, the HMM model can accurately judge that the valve 100 is in an intermediate state during the opening process and predict its subsequent possible state change trend according to the gradual change of the position feature of the drive rod 113 in the feature sequence and the corresponding changes in texture features and depth distribution features. In practical applications, input the dynamic feature sequence of the valve to be recognized into the trained HMM model, and the HMM model will output the most likely valve state category and give the corresponding state confidence score, so that the operator can evaluate and make decisions on the recognition result.

[0051] In this embodiment, the feature extraction module 510 includes a first feature extraction unit 511 and a second feature extraction unit 512. The first feature extraction unit 511 extracts the shape features and texture features of the valve 100 and / or the driving device 300 according to the color image information. The second feature extraction unit 512 extracts the spatial position features and depth distribution features of the valve 100 and / or the driving device 300 according to the depth image information. Through the above structure setting, the first feature extraction unit 511 can extract the shape features and texture features of the valve 100 and / or the driving device 300 according to the color image information. Among them, the texture features can describe information such as the texture roughness and directionality of the valve surface, which is convenient for subsequent judgment of the wear condition or surface material change of the valve. For example, if there are scratches or corrosion on the valve surface, it can be accurately identified through the change of texture features. The shape features describe the basic shape elements of the valve 100 and the geometric relationship between them. According to the change of the relative position or ratio of these shape elements, the state of the valve 100 can be judged. For example, when the valve 100 is opened, the valve stem 112 extends out, which will change the relative position relationship between the valve stem 112 and the valve body 110 in the image, thus reflecting the change in the shape feature parameters. The second feature extraction unit 512 extracts the spatial position features and depth distribution features of the valve 100 and / or the driving device 300 according to the depth image information. Among them, the spatial position features can include the three-dimensional coordinate information of each component of the valve 100. By comparing with the standard coordinate model in the open or closed state of the valve 100 under the standard state, the opening degree of the valve 100 can be directly judged. The depth distribution feature can express the distribution of the whole valve 100 and each part in the depth direction, such as calculating the depth histogram and counting the proportion of the number of pixels in different depth intervals. For some valves with special structures, such as multi-stage pressure reducing valves, different valve states will cause changes in the depth distribution of the internal flow path. Through the depth distribution feature, this state change can be effectively identified, and then the opening degree state of the valve can be judged.

[0052] In this embodiment, the feature fusion module 520 includes a splicing fusion unit 521 and a dimensionality reduction processing unit 522. The splicing fusion unit 521 fuses various types of features at the same time to form a comprehensive feature vector. The dimensionality reduction processing unit 522 performs dimensionality reduction processing on the comprehensive feature vector. Through the above structure setting, the texture feature vector, shape feature vector of the color image, spatial position feature vector and depth distribution feature vector of the depth image are spliced in sequence to form a comprehensive feature vector containing multiple modal information, making full use of the complementarity between different modal features, improving the description ability of the valve state, and further improving the accuracy of the state judgment of the valve 100.

[0053] In this embodiment, the simulation recognition module 530 includes a dynamic feature construction unit 531 and a model recognition unit 532. The dynamic feature construction unit 531 arranges the comprehensive feature vectors in chronological order to form a dynamic feature sequence. The model recognition unit 532 trains a model based on past dynamic feature sequence samples, and uses the model to analyze the state of the valve 100 according to the current dynamic feature sequence. Through the above structural arrangement, the feature vectors extracted from multiple consecutive frames of images and fused and dimension-reduced are arranged in chronological order to form a dynamic feature sequence. This sequence not only contains the static feature information of the valve at a certain moment, but also reflects the trend of feature changes over a period of time, and can better reflect the dynamic change process of the valve state. For example, for a valve that is slowly opening or closing, the spatial position feature of its valve stem will show a gradually rising or falling trend in the sequence.

[0054] In this embodiment, the dimension reduction processing unit 522 uses the principal component analysis algorithm to perform dimension reduction processing on the comprehensive feature vectors. Through the above structural arrangement, since the dimension of the fused feature vectors is relatively high, there may be redundant information, increasing the computational complexity. The PCA (Principal Component Analysis) algorithm is used to perform dimension reduction processing on it. PCA projects the original high-dimensional feature space onto a low-dimensional subspace through linear transformation, so that the projected features can retain the main information while minimizing the correlation between features and data redundancy as much as possible. For example, reducing the fused feature vectors originally in the hundreds of dimensions to dozens of dimensions not only reduces the burden of subsequent calculations, but also ensures that key information is not lost, improves the processing efficiency and classification accuracy of feature data, and can significantly improve the data processing ability and speed.

[0055] In this embodiment, the first feature extraction unit 511 uses an edge detection algorithm to extract the contour edges of the valve 100 and / or the driving device 300, and then uses the Hough transform shape analysis method to identify the shape features of the valve 100 and / or the driving device 300. The first feature extraction unit 511 uses the gray level co-occurrence matrix algorithm to calculate the pixel gray level change rules of the color image in different directions and distances, so as to obtain texture features.

[0056] In this embodiment, the second feature extraction unit 512 calculates the three-dimensional coordinate information of the whole and each part of the valve 100 and / or the driving device 300 according to the depth image information, and analyzes the distribution of the whole and each part of the valve 100 and / or the driving device 300 in the depth direction.

[0057] In this embodiment, the valve 100 includes a valve body 110, a ball valve 111, a valve stem 112, and a drive rod 113. The ball valve 111 is disposed inside the valve body 110, the valve stem 112 passes through the valve body 110 and is connected to the ball valve 111. One end of the drive rod 113 is connected to the drive device 300, and the other end of the drive rod 113 is connected to the valve stem 112. The drive device 300 drives the drive rod 113 to rotate, thereby driving the valve stem 112 and the ball valve 111 to rotate. The first identifier 101 is disposed on the drive rod 113 and / or at a position on the valve body 110 close to the drive rod 113. With the above structure, during use, the drive device 300 drives the drive rod 113 to rotate, thereby driving the valve stem 112 and the ball valve 111 to rotate, so as to change the state of the valve 100, such as an open state, a closed state, etc. By disposing the first identifier 101 on the drive rod 113 and / or at a position on the valve body 110 close to the drive rod 113, the relative position relationship between the valve body 110 and the drive rod 113 can be more clearly shown in the image, thereby facilitating subsequent judgment of the state of the valve 100 and making the judgment of the state of the valve 100 more accurate.

[0058] In this embodiment, the drive device 300 includes an oil cylinder 310, a piston 320, and a piston rod 330. The piston 320 slides inside the oil cylinder 310 under the action of hydraulic oil. One end of the piston rod 330 is connected to the piston 320, and the other end of the piston rod 330 is connected to the drive rod 113. A second identifier 301 is provided on the surface of the piston rod 330. With the above structure, during use, the pressure exerted by the hydraulic oil drives the piston 320 to slide inside the oil cylinder 310, thereby driving the piston rod 330. The piston rod 330 drives the drive rod 113 to drive the valve 100 to open or close. A second identifier 301 is provided on the surface of the piston rod 330. For example, a plurality of partitions of different colors are provided on the surface of the piston rod 330 along its extending direction. Then, the length of the piston rod 330 extending out of the oil cylinder 310 can be judged according to the color on the piston rod 330 exposed outside the oil cylinder 310, thereby judging the state of the valve 100 and improving the accuracy of the judgment. Moreover, the first identifier 101 on the valve 100 can be combined, and the two are comprehensively judged to further improve the accuracy of the judgment.

[0059] Refer to Figures 1 to 10 , a valve vision monitoring method, including:

[0060] Step S1, use the image acquisition device 400 to acquire the image information of the valve 100 and / or the driving device 300. Among them, the image acquisition device 400 includes a color image acquisition module 410 and a depth image acquisition module 420. The color image acquisition module 410 acquires the color image information of the valve 100 and / or the driving device 300, and the depth image acquisition module 420 acquires the depth image information of the valve 100 and / or the driving device 300;

[0061] Step S2, use the feature extraction module 510 to extract various types of image features according to the image information;

[0062] Step S3, use the feature fusion module 520 to fuse various types of image features at the same time to form a comprehensive feature vector;

[0063] Step S4, use the analog recognition module 530 to arrange the comprehensive feature vectors in chronological order to form a dynamic feature sequence, and use the hidden Markov model to analyze the state of the valve 100 according to the dynamic feature sequence.

[0064] Through the setting of the above structure, the color image acquisition module 410 and the depth image acquisition module 420 are used to acquire visual information of different modalities respectively. The feature extraction module 510 can extract various types of image features targeted according to the image information, such as the relative position and angle between the valve body 110 and the driving rod 113 of the valve 100, scratches and corrosion on the valve surface, etc., excluding useless information, and improving the analysis accuracy and speed; the feature fusion module 520 fuses various types of image features at the same time to form a comprehensive feature vector, ensuring the unity in time, so as to understand various types of image features of the valve 100 at the same time, facilitating subsequent classification or recognition tasks, and improving the analysis accuracy; based on a large number of dynamic feature sequence samples of different states of the valve, such as fully closed, partially opened, fully opened, faulty, etc., train the HMM (hidden Markov model). The HMM can model the feature sequence with time series. By analyzing the transition probability between hidden states in the sequence and the emission probability of observed features under each hidden state, the classification and recognition of the valve state are realized. For example, when the valve 100 starts to open gradually from the fully closed state, the HMM model can accurately judge that the valve 100 is in an intermediate state during the opening process and predict its subsequent possible state change trend according to the gradual change of the position feature of the driving rod 113 in the feature sequence and the corresponding changes in texture features and depth distribution features. In practical applications, the dynamic feature sequence of the valve to be recognized is input into the trained HMM model, and the HMM model will output the most likely valve state category and give the corresponding state confidence score, so that the operator can evaluate and make decisions on the recognition results.

Claims

1. A valve visual monitoring system, characterized in that, Including: A valve (100) for installation on a pipeline (200), the valve (100) being used to open and close the pipeline (200) and regulate the flow rate of the medium in the pipeline (200), wherein a first identifier (101) is provided on the valve (100); A driving device (300) for connection to the pipeline (200), the driving device (300) driving the valve (100) to move between an open position and a closed position; An image acquisition device (400) for acquiring image information of the valve (100) and / or the driving device (300), wherein the image acquisition device (400) includes a color image acquisition module (410) and a depth image acquisition module (420), the color image acquisition module (410) acquiring color image information of the valve (100) and / or the driving device (300), and the depth image acquisition module (420) acquiring depth image information of the valve (100) and / or the driving device (300); A data processing device (500) for receiving and analyzing the image information and obtaining the state of the valve (100) based thereon.

2. The valve vision monitoring system according to claim 1, wherein The data processing device (500) includes a feature extraction module (510), a feature fusion module (520) and a simulation recognition module (530). The feature extraction module (510) extracts various image features according to the image information. The feature fusion module (520) fuses various image features at the same time to form a comprehensive feature vector. The simulation recognition module (530) arranges the comprehensive feature vectors in chronological order to form a dynamic feature sequence, and analyzes the state of the valve (100) according to the dynamic feature sequence using a hidden Markov model.

3. The valve vision monitoring system according to claim 2, characterized in that, The feature extraction module (510) includes a first feature extraction unit (511) and a second feature extraction unit (512). The first feature extraction unit (511) extracts the shape features and texture features of the valve (100) and / or the driving device (300) according to the color image information. The second feature extraction unit (512) extracts the spatial position features and depth distribution features of the valve (100) and / or the driving device (300) according to the depth image information.

4. The valve vision monitoring system according to claim 2, wherein, The feature fusion module (520) includes a splicing fusion unit (521) and a dimensionality reduction processing unit (522). The splicing fusion unit (521) fuses various features at the same time to form a comprehensive feature vector. The dimensionality reduction processing unit (522) performs dimensionality reduction processing on the comprehensive feature vector.

5. The valve vision monitoring system according to claim 2, wherein The simulation recognition module (530) includes a dynamic feature construction unit (531) and a model recognition unit (532). The dynamic feature construction unit (531) arranges the comprehensive feature vectors in chronological order to form a dynamic feature sequence. The model recognition unit (532) trains a model based on past dynamic feature sequence samples and uses the model to analyze the state of the valve (100) according to the current dynamic feature sequence.

6. The valve vision monitoring system according to claim 3, characterized in that, The first feature extraction unit (511) uses an edge detection algorithm to extract the contour edges of the valve (100) and / or the driving device (300), and then uses the Hough transform shape analysis method to identify the shape features of the valve (100) and / or the driving device (300). The first feature extraction unit (511) uses the gray-level co-occurrence matrix algorithm to calculate the pixel gray-level change rules of the color image in different directions and distances, so as to obtain texture features.

7. The valve vision monitoring system according to claim 3, wherein The second feature extraction unit (512) calculates the three-dimensional coordinate information of the whole and each part of the valve (100) and / or the driving device (300) according to the depth image information, and analyzes the distribution of the whole and each part of the valve (100) and / or the driving device (300) in the depth direction.

8. The valve vision monitoring system according to claim 1, characterized in that The valve (100) includes a valve body (110), a ball valve (111), a valve stem (112) and a driving rod (113). The ball valve (111) is arranged in the valve body (110), the valve stem (112) passes through the valve body (110) and is connected to the ball valve (111). One end of the driving rod (113) is connected to the driving device (300), and the other end of the driving rod (113) is connected to the valve stem (112). The driving device (300) drives the driving rod (113) to rotate, thereby driving the valve stem (112) and the ball valve (111) to rotate. The first identifier (101) is arranged on the driving rod (113) and / or at the position of the valve body (110) close to the driving rod (113).

9. The valve vision monitoring system according to claim 8, characterized in that, The driving device (300) includes an oil cylinder (310), a piston (320) and a piston rod (330). The piston (320) slides in the oil cylinder (310) under the action of hydraulic oil. One end of the piston rod (330) is connected to the piston (320), and the other end of the piston rod (330) is connected to the driving rod (113). A second identifier (301) is arranged on the surface of the piston rod (330).

10. A valve visual monitoring method, characterized in that, Comprising: Use an image acquisition device (400) to acquire image information of the valve (100) and / or the drive device (300). Among them, the image acquisition device (400) includes a color image acquisition module (410) and a depth image acquisition module (420). The color image acquisition module (410) acquires color image information of the valve (100) and / or the drive device (300), and the depth image acquisition module (420) acquires depth image information of the valve (100) and / or the drive device (300); Use a feature extraction module (510) to extract various image features according to the image information; Use a feature fusion module (520) to fuse various image features at the same time to form a comprehensive feature vector; Use an analog recognition module (530) to arrange the comprehensive feature vectors in chronological order to form a dynamic feature sequence, and use a hidden Markov model to analyze the state of the valve (100) according to the dynamic feature sequence.

Images