An event camera-based oil and water pipeline leakage monitoring method and system

Abstract

The application belongs to the field of power station oil-water pipeline monitoring, and particularly relates to an oil-water pipeline leakage monitoring method and system based on an event camera, which comprises the following steps: step S1. event data acquisition: at least one event camera is deployed near a potential leakage point of an oil-water pipeline to be monitored; step S2. dynamic droplet process capture triggering: dynamic brightness change information of a whole process of droplet formation, growth, separation and dropping caused by oil-water leakage is captured in real time, so that an event is formed; step S3. neural network feature extraction and analysis; and step S4. leakage warning and prompting. The application uses an event camera to monitor the leakage of an oil-water pipeline. Compared with visible light cooperating with a deep learning algorithm, the method can be applied to a high-reflectivity and low-contrast scene. Compared with existing devices relying on electrical monitoring, the method has the characteristics of wide monitoring range and high monitoring sensitivity. Compared with an infrared monitoring method, the method can be applied to the monitoring of normal-temperature oil-water leakage and the monitoring of temperature-changing oil-water leakage, and has the characteristic of wide application range.

Description

Technical Field

[0001] This application belongs to the field of oil and water pipeline monitoring in power plants, specifically involving a method and system for monitoring leaks in oil and water pipelines based on event cameras. Background Technology

[0002] In industrial production and infrastructure operation and maintenance, leak monitoring of oil and water pipelines is a crucial step in ensuring safety and reducing losses. Currently, the mainstream monitoring solutions in the industry can be divided into three main categories, each with significant differences in technical principles, applicable scenarios, and actual effectiveness, as detailed below: The first category is the monitoring solution using visible light cameras and deep learning algorithms, which is currently the most widely used and deployed mainstream technology. Its core logic is to acquire real-time images of the pipeline surface using a high-definition visible light camera, transmit the image data to the backend system, and use a pre-trained deep learning model (such as a CNN convolutional neural network) to identify abnormal features such as leaks and corrosion spots in the pipeline. The advantages of this solution are its mature technology, controllable cost, and ability to achieve large-scale, non-contact monitoring, making it particularly suitable for open environments such as open pipelines and large tank farms. However, its limitations are also significant: the monitoring effect is highly dependent on lighting conditions; at night, in rainy weather, or in dimly lit enclosed spaces, image clarity drops drastically, leading to a decrease in algorithm accuracy; even with additional lighting equipment, misjudgments may occur due to light reflection and shadow obstruction, making it difficult to meet the needs of 24-hour uninterrupted monitoring. The second type is the metal contact electrical monitoring solution, a typical representative of contact monitoring technology. Its design involves embedding pairs of metal contact sensors at critical leak-prone nodes such as pipeline flanges and valves. Under normal conditions, the contacts are open. When a leak occurs, the leaking oil or water acts as a conductive medium, creating an electrical circuit between the metal contacts and triggering an alarm signal. This solution is simple in structure and low in cost, suitable for fixed-point monitoring of specific high-risk nodes. However, its shortcomings are also obvious: firstly, the monitoring range is extremely small, only covering the local area where the contacts are located, rendering it ineffective for leaks in non-critical locations such as the middle section of the pipeline; secondly, the monitoring sensitivity is poor, requiring the leakage to reach a certain threshold sufficient to completely wet the contacts before triggering an alarm, often missing the optimal time to address initial leaks. Furthermore, in environments with viscous oil or poor water quality, problems such as contact failure and increased false alarm rates may occur. The third category is a monitoring solution that combines mobile inspection robots with multi-sensor fusion, primarily designed for specific scenarios such as cooling water pipelines. This solution uses an autonomously moving inspection robot as a carrier, equipped with a rotating camera and an infrared imager. The robot moves along a preset path to achieve dynamic inspection of the pipeline. Its core advantage lies in its ability to capture the temperature difference between the leaking liquid and the surrounding environment using infrared imaging technology. For example, a leaking low-temperature liquid in a cooling water pipe will form a clear low-temperature area in the infrared image, thus quickly locating the leak point. However, the applicability of this technology is strictly limited: it can only monitor liquids with significant temperature differences. For leaks of oil or water pipes at temperatures close to ambient, the infrared imager is difficult to identify, and in such cases, only visible light monitoring from the rotating camera can be relied upon, significantly reducing the effectiveness. Furthermore, the monitoring results are easily affected by the ambient temperature. In extreme temperature environments such as high-temperature workshops or winter outdoors, the temperature difference is compressed, significantly reducing the detection accuracy of the infrared sensor and even leading to missed detections. Summary of the Invention

[0003] This application aims to solve the above-mentioned problems in the prior art. It proposes a method and system for monitoring oil and water pipeline leaks based on an event camera. It can monitor the entire process of droplet formation and dripping in real time without the need for external light source supplementation. The image processing is realized through neural network, thereby triggering alarms for corresponding scenarios.

[0004] To achieve the above objectives, the technical solution of the present invention is as follows: A method for monitoring leaks in oil and water pipelines based on event cameras includes the following steps: Step S1. Event Data Acquisition: Deploy at least one event camera near the potential leak point of the oil and water pipeline to be monitored. No supplementary lighting is required. The event camera generates asynchronous event stream data in response to changes in brightness in the scene. Step S2. Dynamic droplet process capture trigger: Utilizing the inherent sensitivity of the event camera to brightness changes, the dynamic brightness change information of the entire process of droplet formation, growth, detachment and droplet falling caused by oil and water leakage is captured in real time, thereby forming an event; Step S3. Neural Network Feature Extraction and Analysis: Input the event stream data containing dynamic information of the entire process from droplet formation to droplet falling into the pre-trained neural network model; Step S4. Leakage warning and prompt: When the neural network model identifies a droplet behavior pattern that matches the preset leakage characteristics, it generates a leak warning signal and outputs a prompt message.

[0005] Furthermore, in step S1, the event camera deployment locations include flange connections of oil and water pipelines, valves, pump body seals, weld seams, or areas prone to pipe corrosion; the event stream data includes pixel coordinates. timestamp and polarity information , represented as .

[0006] Furthermore, in step S2, the dynamic brightness change information is reflected as the aggregation pattern of the event stream data in the corresponding spatiotemporal domain; the event camera spontaneously and with low latency captures the dynamic process of the droplets through its inherent sensitivity to brightness changes. Furthermore, the event camera is a LUCID triton2-evs.

[0007] Furthermore, an event accumulation camera output step is provided between steps S2 and S3. The event accumulation camera output step is to superimpose the formed events according to a specified number of events to form an image.

[0008] Furthermore, in step S3, the neural network model is configured as follows: Step S31. Extract spatiotemporal features related to droplet dynamic behavior from event stream data. The spatiotemporal features related to droplet dynamic behavior include one or more combinations of multiple droplet event spatial correlation features formed by droplet contour change features, droplet size growth rate features, droplet suspension time features, droplet detachment trajectory features, and droplet fall frequency features. Step S32. Based on the spatiotemporal features extracted in step S31, identify whether there is a droplet behavior pattern that conforms to the preset leakage characteristics.

[0009] Furthermore, before step S3, event data preprocessing is also included: preprocessing the acquired event stream data, the preprocessing including at least one of event accumulation, noise filtering, and background activity suppression.

[0010] Furthermore, median filtering is used for noise removal, and its formula is:

[0011] in, This represents the sorted sequence of pixel values. Indicates the window size.

[0012] Furthermore, in step S3, the neural network model is one or more of the following combined architectures: convolutional neural network, recurrent neural network, graph neural network, or Transformer network.

[0013] Furthermore, in step S31, The characteristic equation for droplet profile change is:

[0014] where, R is the radius of curvature at the point (x, z), and b is the radius of curvature at the origin. is the tilt angle at the point (x, z), and the coefficient is the surface tension. is the acceleration due to gravity; x and z are the coordinates relative to the origin o; after determining the origin information, substitute the point coordinates (x, z), the radius of curvature R, and the tilt angle , and calculate the radius of curvature b of the origin and the coefficient and compare with the pre-set b and in the pattern to determine whether dripping occurs.

[0015] The flying and scattering state of the droplet in the air is described by the following formula for the resistance D:

[0016] where, C is the resistance coefficient, is the air density A is the cross-sectional area of the water droplet, and v is the velocity of the water droplet relative to the wind. Analyze the current motion state of the droplet by combining the calculation of the resistance D with the gravity G and the buoyancy F. When G > D + F, the droplet is in the state of accelerating downward or decelerating upward; when G = D + F, the droplet is in the uniform state; when G < D + F, the droplet is in the state of decelerating downward or accelerating upward.

[0017] The terminal velocity of the droplet is:

[0018] where, is the diameter of the droplet, is the density of the droplet, is the density of the surrounding medium, is the acceleration due to gravity, is the viscosity of the surrounding medium.

[0019] Furthermore, the identification of the droplet behavior pattern that conforms to the preset leakage characteristics in the step S32 includes: distinguishing the droplet behavior patterns formed by normal condensate, splash, and continuous leakage; when the leakage droplet behavior pattern is identified, based on the spatial distribution of the event data, determine the specific location or area where the leakage occurs.

[0020] Even further, each pattern is specifically: Normal condensate pattern: Radius of curvature b of the origin: In the state of normal condensate, the droplet presents a stable spherical shape, and the radius of curvature b of the origin is on the order of millimeters to centimeters. Terminal velocity V rd : In the state of normal condensate, the droplet is in a relatively static state, and the terminal velocity V of the droplet rd≈0; Splash Mode: Origin curvature radius b: In the splash state, the origin curvature radius b is on the order of micrometers to millimeters; Terminal speed V rd During the splashing process, the terminal velocity V of the droplet rd It will increase.

[0021] Persistent leakage mode: Origin curvature radius b: Under continuous leakage conditions, the origin curvature radius b will vary within a stable range as the leakage conditions change, ranging from millimeters to centimeters.

[0022] Terminal speed V rd Under continuous leakage conditions, the terminal velocity Vrd of the droplet reaches a stable value. Furthermore, the droplet determination process is as follows: First, determine the terminal speed V. rd If it is 0, then identify the radius of curvature b. If it is in the range of millimeters to centimeters, it is classified as normal condensate. Secondly, if the terminal velocity Vrd is not 0, the radius of curvature b is considered to be splashing water if it is on the order of micrometers to millimeters, and to be continuous leakage if it is on the order of millimeters to centimeters.

[0023] Furthermore, the droplet determination process is as follows: For different droplets with the same size characteristics, the coefficient β is strongly correlated with the droplet density, and different droplet types can be distinguished by different coefficients β.

[0024] Furthermore, the prompt information output in step S4 includes one or more of the following: visual alarm indication, audible alarm, leak location information, leak severity level, and maintenance work order generation information; the leak severity level is classified according to one or more of the following factors: the identified droplet formation rate, droplet size, and droplet falling frequency.

[0025] A camera-based system for monitoring leaks in oil and water pipelines includes: Event camera module: Deployed near potential leak points in the oil-water pipeline to be monitored, it is used to capture dynamic brightness changes throughout the entire process from droplet formation to dripping caused by oil-water leakage, and output asynchronous event stream data; Data processing and analysis module: Communicates with the event camera module to receive event stream data; the module contains a pre-trained neural network model to extract spatiotemporal features related to droplet dynamic behavior from the event stream data, and based on the extracted spatiotemporal features, identifies whether there are droplet behavior patterns that conform to preset leakage characteristics; Early warning output module: Connected to the data processing and analysis module, when a droplet behavior pattern that matches the preset leakage characteristics is identified, it generates and outputs a leak warning signal and prompt information.

[0026] Furthermore, it also includes an event data preprocessing unit: used to accumulate, filter noise or suppress background activity in the raw event stream data output by the event camera module, and then input the processed data into the data processing and analysis module.

[0027] Furthermore, the data processing and analysis module is deployed on edge computing devices, local servers, or cloud servers.

[0028] The advantages of this application are: 1. This application employs an event camera to monitor leaks in oil and water pipelines. Compared to visible light combined with deep learning algorithms, this method is applicable to high-reflectivity, low-contrast scenarios. Compared to existing devices relying on electrical monitoring, it features a wider monitoring range and higher sensitivity. Compared to infrared monitoring methods, it is suitable for monitoring oil and water leaks at both normal temperatures and those occurring under temperature variations, demonstrating broad applicability.

[0029] 2. This application can monitor oil and water pipeline status in real time without the need for external light source supplementation, adapt to monitoring scenarios in low light environment, and provide alarm reminders for multiple scenarios. Attached Figure Description

[0030] Figure 1 This is a flowchart of the method used in this application. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of the embodiments of the invention clearer, the technical solutions of the embodiments of the invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the invention, not all embodiments. The components of the embodiments of the invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0032] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0033] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0034] In the description of this invention, it should be noted that the terms "upper," "vertical," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use, or the orientation or positional relationship commonly understood by those skilled in the art. They are only used for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention. In addition, the terms "first," "second," etc., are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0035] This invention presents an aircraft surface feature segmentation method based on contour constraint optimization, which uses a deep learning network to segment targets in an image. The method learns feature information from the image through a feature extraction backbone network, then fits the outer contour constraint of the target based on this feature information, initially segmenting the target in the image. Finally, the target contour constraint is used to optimize the segmentation result, achieving high-precision segmentation of each target instance in the image.

[0036] Example 1 like Figure 1 As shown, a method for monitoring leaks in oil and water pipelines based on event cameras includes the following steps: Step S1. Event Data Acquisition: Deploy at least one event camera near the potential leak point of the oil and water pipeline to be monitored. No supplementary lighting is required. The event camera generates asynchronous event stream data in response to changes in brightness in the scene. Step S2. Dynamic droplet process capture trigger: Utilizing the inherent sensitivity of the event camera to brightness changes, the dynamic brightness change information of the entire process of droplet formation, growth, detachment and droplet falling caused by oil and water leakage is captured in real time, thereby forming an event; Step S3. Neural Network Feature Extraction and Analysis: Input the event stream data containing dynamic information of the entire process from droplet formation to droplet falling into the pre-trained neural network model; Step S4. Leakage Warning and Alert: When the neural network model identifies a droplet behavior pattern that matches the preset leakage characteristics, it generates a leak warning signal and outputs an alert message. The method workflow is as follows: Figure 1 As shown.

[0037] Example 2 like Figure 1 As shown, a method for monitoring leaks in oil and water pipelines based on event cameras includes the following steps: Step S1. Event Data Acquisition: Deploy at least one event camera near the potential leak point of the oil and water pipeline to be monitored. No supplementary lighting is required. The event camera generates asynchronous event stream data in response to changes in brightness in the scene. Step S2. Dynamic droplet process capture trigger: Utilizing the inherent sensitivity of the event camera to brightness changes, the dynamic brightness change information of the entire process of droplet formation, growth, detachment and droplet falling caused by oil and water leakage is captured in real time, thereby forming an event; Step S3. Neural Network Feature Extraction and Analysis: Input the event stream data containing dynamic information of the entire process from droplet formation to droplet falling into the pre-trained neural network model; Step S4. Leakage Warning and Alert: When the neural network model identifies a droplet behavior pattern that matches the preset leakage characteristics, it generates a leak warning signal and outputs an alert message. The method workflow is as follows: Figure 1 As shown.

[0038] In step S1, the event camera deployment locations include flange connections of oil and water pipelines, valves, pump body seals, weld seams, or areas prone to pipe corrosion; the event stream data includes pixel coordinates. timestamp and polarity information , represented as .

[0039] In step S2, the dynamic brightness change information is reflected as the aggregation pattern of the event stream data in the corresponding spatiotemporal domain; through the inherent sensitivity of the event camera to brightness changes, it can spontaneously and with low latency capture the dynamic process of the droplets without external triggering or fixed frame rate imaging. The event camera is a LUCID triton2-evs.

[0040] Between steps S2 and S3, there is an event accumulation camera output step, which involves superimposing the formed events into a specified number of images.

[0041] In step S3, the neural network model is configured as follows: Step S31. Extract spatiotemporal features related to droplet dynamic behavior from event stream data. The spatiotemporal features related to droplet dynamic behavior include one or more combinations of multiple droplet event spatial correlation features formed by droplet contour change features, droplet size growth rate features, droplet suspension time features, droplet detachment trajectory features, and droplet fall frequency features. Step S32. Based on the spatiotemporal features extracted in step S31, identify whether there is a droplet behavior pattern that conforms to the preset leakage characteristics.

[0042] Before step S3, it also includes event data preprocessing: preprocessing the acquired event stream data, and the preprocessing includes at least one of event accumulation, noise filtering, and background activity suppression to enhance the effective event signal related to the droplet.

[0043] Noise filtering uses median filtering (Median Filter), and its formula is:

[0044] where represents the sorted pixel value sequence, represents the window size.

[0045] In step S3, the neural network model is one or a combination of architectures such as a convolutional neural network, a recurrent neural network, a graph neural network, or a Transformer network.

[0046] Furthermore, in the said step S31, The droplet profile change characteristic equation is

[0047] where R is the radius of curvature at the point (x, z), b is the radius of curvature at the origin, is the tilt angle at the point (x, z), and the coefficient is the surface tension, is the acceleration due to gravity; x, z are the coordinates relative to the origin o; after determining the origin information, substitute the point coordinates (x, z), the radius of curvature R, and the tilt angle , and calculate the radius of curvature b at the origin and the coefficient and compare with the preset b and in the pattern to determine whether a droplet occurs.

[0048] The flying state of the droplet in the air is described by the following formula for the drag D:

[0049] where C is the drag coefficient of 0.60, is the air density (about A is the cross-sectional area of the water droplet, and v is the velocity of the water droplet relative to the wind; By combining the calculation of the drag D with the gravity G and the buoyancy F, analyze the current motion state of the droplet. When G > D + F, the droplet is in the state of accelerating downward or decelerating upward. When G = D + F, the droplet is in the uniform state. When G < D + F, the droplet is in the state of decelerating downward or accelerating upward. For example, when spraying upward, it first decelerates upward and then accelerates downward.

[0050] The terminal velocity of the droplet is:

[0051] in, It is the droplet diameter. It is the droplet density. It is the density of the surrounding medium. It is gravitational acceleration. It is the viscosity of the surrounding medium. This is calculated... The speed is analyzed to determine the leakage status of the droplets, thereby providing different levels of alarms.

[0052] The step S32 of identifying droplet behavior patterns that conform to preset leakage characteristics includes: distinguishing droplet behavior patterns formed by normal condensation, splashing and continuous leakage; when a leakage droplet behavior pattern is identified, determining the specific location or area where the leakage occurs based on the spatial distribution of event data.

[0053] The specific modes are as follows: Normal condensate mode: The radius of curvature at the origin, b: Under normal condensation conditions, droplets usually exhibit a stable spherical or near-spherical shape, so the radius of curvature at the origin, b, is relatively large, ranging from millimeters to centimeters. Terminal speed V rd Under normal condensation conditions, the droplets are in a relatively stationary state, and the terminal velocity of the droplets is V. rd ≈0; Splash Mode: Origin radius of curvature b: In the splashing state, due to the splitting of the droplet by external force, the origin radius of curvature b may decrease, ranging from micrometers to millimeters; Terminal speed V rd During the splashing process, the terminal velocity V of the droplet rd It will increase significantly, depending on the mass of the droplet and the initial velocity at the time of splashing.

[0054] Persistent leakage mode: Origin curvature radius b: Under continuous leakage conditions, the origin curvature radius b may vary within a stable range as leakage conditions change, ranging from millimeters to centimeters.

[0055] Terminal speed V rd Under continuous leakage conditions, the terminal velocity Vrd of the droplet reaches a stable value. The droplet determination process is as follows: First, determine the terminal speed V. rd If it is 0, then identify the radius of curvature b. If it is in the range of millimeters to centimeters, it is classified as normal condensate. Secondly, if the terminal velocity Vrd is not 0, the radius of curvature b is considered to be splashing water if it is on the order of micrometers to millimeters, and to be continuous leakage if it is on the order of millimeters to centimeters.

[0056] The droplet determination process is as follows: The coefficient β can distinguish different types of droplets. For different droplets of the same size, the coefficient β is strongly correlated with droplet density, which is 850-920 kg / m³ for crude oil. The water content is 998.2 kg / Different droplet types can be distinguished by using different coefficients β.

[0057] The prompt information output in step S4 includes one or more of the following: visual alarm indication, audible alarm, leak location information, leak severity level, and maintenance work order generation information; the leak severity level is classified according to one or more of the following factors: the identified droplet formation rate, droplet size, and droplet falling frequency.

[0058] Example 3 A camera-based system for monitoring leaks in oil and water pipelines includes: Event camera module: Deployed near potential leak points in the oil-water pipeline to be monitored, it is used to capture dynamic brightness changes throughout the entire process from droplet formation to dripping caused by oil-water leakage, and output asynchronous event stream data; Data processing and analysis module: Communicates with the event camera module to receive event stream data; the module contains a pre-trained neural network model to extract spatiotemporal features related to droplet dynamic behavior from the event stream data, and based on the extracted spatiotemporal features, identifies whether there are droplet behavior patterns that conform to preset leakage characteristics; Early warning output module: Connected to the data processing and analysis module, when a droplet behavior pattern that matches the preset leakage characteristics is identified, it generates and outputs a leak warning signal and prompt information.

[0059] It also includes an event data preprocessing unit: used to accumulate, filter noise or suppress background activity in the raw event stream data output by the event camera module, and then input the processed data into the data processing and analysis module.

[0060] The data processing and analysis module is deployed on edge computing devices, local servers, or cloud servers.

Claims

1. A method for monitoring leaks in oil and water pipelines based on event cameras, characterized in that, It includes the following steps: Step S1. Event data acquisition: Deploy at least one event camera near the potential leakage points of the oil and water pipelines to be monitored. Without light source supplementation, the event camera generates asynchronous event stream data in response to the brightness change in the scene; Step S2. Trigger for capturing the dynamic droplet process: Utilize the inherent sensitivity of the event camera to brightness change to capture in real time the dynamic brightness change information of the whole process from droplet formation, growth, detachment to dripping caused by oil and water leakage, thereby forming events; Step S3. Neural network feature extraction and analysis: Input the event stream data containing the dynamic information of the whole process from droplet formation to dripping into a pre-trained neural network model; Step S4. Leakage warning and prompt: When the neural network model identifies a droplet behavior pattern that conforms to the preset leakage characteristics, generate a warning signal for running, overflowing, dripping, and leaking, and output a prompt message.

2. The method for monitoring leaks in oil and water pipelines based on an event camera according to claim 1, characterized in that, In step S1, the event camera deployment locations include flange connections of oil and water pipelines, valves, pump seals, weld seams, and areas prone to pipe corrosion; the event stream data includes pixel coordinates. timestamp and polarity information , represented as .

3. The method for monitoring leaks in oil and water pipelines based on event cameras according to claim 1, characterized in that, In step S2, the dynamic brightness change information is embodied as the aggregation pattern of the event stream data in the corresponding spatio-temporal domain; the dynamic process of the droplet is captured by the event camera for the brightness change.

4. The method for monitoring leaks in oil and water pipelines based on an event camera according to claim 1, characterized in that, An event accumulation camera image generation step is provided between step S2 and step S3. The event accumulation camera image generation step is to superimpose the formed events for a specified event to form an image.

5. The method for monitoring leaks in oil and water pipelines based on an event camera according to claim 1, characterized in that, In step S3, the neural network model is configured as follows: Step S31. Extract spatio-temporal features related to the dynamic behavior of the droplet from the event stream data. The spatio-temporal features related to the dynamic behavior of the droplet include one or a combination of multiple droplet event spatial correlation features such as droplet contour change features, droplet size growth rate features, droplet suspension time features, droplet detachment trajectory features, and droplet dripping frequency features; Step S32. Based on the spatio-temporal features extracted in step S31, identify whether there is a droplet behavior pattern that conforms to the preset leakage characteristics.

6. The method for monitoring leaks in oil and water pipelines based on an event camera according to claim 1, characterized in that, Before step S3, event data preprocessing is also included: Preprocess the acquired event stream data, and the preprocessing includes at least one of event accumulation, noise filtering, and background activity suppression.

7. The method for monitoring leaks in oil and water pipelines based on an event camera according to claim 5, characterized in that, Median filtering is used for noise filtering, and its formula is: in, This represents the sorted sequence of pixel values. Indicates the window size.

8. A method for monitoring leaks in oil and water pipelines based on an event camera, as described in claim 1, characterized in that, In step S3, the neural network model is one or a combination architecture of a convolutional neural network, a recurrent neural network, a graph neural network, or a Transformer network.

9. A method for monitoring leaks in oil and water pipelines based on an event camera, as described in claim 5, characterized in that, In step S31, The equation for the droplet contour change feature is Where R is the radius of curvature at point (x, z), and b is the radius of curvature at the origin. Let be the angle of inclination at point (x, z), and the coefficient be . For surface tension, It is the acceleration due to gravity; x and z are the coordinates relative to the origin o; after determining the origin information, substitute the point coordinates (x, z), radius of curvature R, and tilt angle. Find the radius of curvature b at the origin and the coefficient. With the pre-set pattern b, Compare and determine whether dripping has occurred; The splashing state of the droplet in the air is described by the following formula for the resistance D: Where C is the drag coefficient, It is air density A is the cross-sectional area of ​​the water droplet, and v is the speed of the water droplet relative to the wind. Analyze the current motion state of the droplet by combining the calculation of the resistance D with the gravity G and the buoyancy F. When G > D + F, the droplet is in the state of accelerating falling or decelerating rising; when G = D + F, the droplet is in the uniform state; when G < D + F, the droplet is in the state of decelerating falling or accelerating rising; terminal velocity of the droplet for: in, It is the droplet diameter. It is the droplet density. It is the density of the surrounding medium. It is gravitational acceleration. It is the viscosity of the surrounding medium.

10. A method for monitoring leaks in oil and water pipelines based on an event camera, as described in claim 5, characterized in that, In step S32, identifying the droplet behavior pattern that conforms to the preset leakage characteristics includes: distinguishing the droplet behavior patterns formed by normal condensate water, splashing, and continuous leakage; when identifying the leakage droplet behavior pattern, determine the specific location or area where the leakage occurs based on the spatial distribution of the event data.

11. A method for monitoring leaks in oil and water pipelines based on an event camera, as described in claim 10, characterized in that, Each pattern is specifically as follows: Normal condensate water pattern: Origin curvature radius b: In the state of normal condensate water, the droplet presents a stable spherical shape, and the origin curvature radius b is in the order of millimeters to centimeters; Terminal speed V rd Under normal condensation conditions, the droplets are in a relatively stationary state, and the terminal velocity of the droplets is V. rd ≈0; Splash Mode: Origin curvature radius b: In the splash state, the origin curvature radius b is on the order of micrometers to millimeters; Terminal speed V rd During the splashing process, the terminal velocity V of the droplet rd It will increase; Persistent leakage mode: Origin curvature radius b: Under continuous leakage conditions, the origin curvature radius b will vary within a stable range as the leakage conditions change, ranging from millimeters to centimeters; Terminal speed V rd Under continuous leakage conditions, the terminal velocity Vrd of the droplet reaches a stable value.

12. The method for monitoring oil and water pipeline leaks based on an event camera according to claim 10, characterized in that, The droplet determination process is as follows: First, determine the terminal speed V. rd If it is 0, then identify the radius of curvature b. If it is in the range of millimeters to centimeters, it is classified as normal condensate. Secondly, if the terminal velocity Vrd is not 0, the radius of curvature b is considered to be splashing water if it is on the order of micrometers to millimeters, and to be continuous leakage if it is on the order of millimeters to centimeters.

13. A method for monitoring leaks in oil and water pipelines based on an event camera, as described in claim 10, characterized in that, Furthermore, the droplet identification process is as follows: under the same size characteristics, different droplets have a strong correlation between the coefficient β and the droplet density, and different droplet types can be distinguished by different coefficients β.

14. The method for monitoring oil and water pipeline leaks based on an event camera according to claim 1, characterized in that, The prompt information output in step S4 includes one or more of the following: visual alarm indication, audible alarm, leak location information, leak severity level, and maintenance work order generation information; the leak severity level is classified according to one or more of the following factors: the identified droplet formation rate, droplet size, and droplet falling frequency.

15. A system for monitoring leaks in oil and water pipelines based on event cameras, characterized in that, include: Event camera module: Deployed near potential leak points in the oil-water pipeline to be monitored, it is used to capture dynamic brightness changes throughout the entire process from droplet formation to dripping caused by oil-water leakage, and output asynchronous event stream data; Data processing and analysis module: Communicates with the event camera module to receive event stream data; the module contains a pre-trained neural network model to extract spatiotemporal features related to droplet dynamic behavior from the event stream data, and based on the extracted spatiotemporal features, identifies whether there are droplet behavior patterns that conform to preset leakage characteristics; Early warning output module: Connected to the data processing and analysis module, when a droplet behavior pattern that matches the preset leakage characteristics is identified, it generates and outputs a leak warning signal and prompt information.

16. A system for monitoring leaks in oil and water pipelines based on an event camera, as described in claim 15, is characterized in that... It also includes an event data preprocessing unit: used to accumulate, filter noise or suppress background activity in the raw event stream data output by the event camera module, and then input the processed data into the data processing and analysis module.

17. A system for monitoring leaks in oil and water pipelines based on an event camera, as described in claim 15, characterized in that, The data processing and analysis module is deployed on edge computing devices, local servers, or cloud servers.

Images