Video monitoring method and system based on wind power plant area

Abstract

The invention relates to the technical field of video monitoring, in particular to a video monitoring method and system based on a wind power plant area, and the method comprises the following steps: employing an intelligent panoramic array camera, setting the size of an aperture, adjusting the exposure time, checking the image definition in a night vision mode, recording the frequency and power of a fog penetration function, and covering the monitoring range of a wind power plant. And obtaining a panoramic real-time image. According to the method, the moving pixel points in the panoramic image are analyzed and compared with the static background, the moving target is identified, potential safety hazards or faults are found, the position of the ship is compared with the preset navigation path, the potential route deviation behavior can be effectively predicted and warned, the wind power plant is regarded as a network, and the wind power generation efficiency is improved. A fault propagation path is analyzed through a graph theory model, the fault processing flow is optimized, mastering of the operation state of the wind power plant is enhanced, making and execution of a maintenance plan are effectively guided, and stable operation and efficient maintenance of the wind power plant are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of video surveillance, and particularly to a video surveillance method and system based on a wind farm site. Background Art

[0002] The technical field of video surveillance mainly involves using video cameras or camera systems to capture, record, and analyze video images for monitoring and protecting the safety of personnel, property, and information. It is widely applied in multiple industries, including retail, education, banking, transportation, and critical infrastructure. With the progress of technology, video surveillance systems can not only monitor in real time but also identify specific patterns, behaviors, and abnormal events through advanced algorithms and machine learning, improving the safety level and response speed. Modern video surveillance also combines cloud technology, big data, and Internet of Things technology, making the video surveillance system more intelligent and networked.

[0003] Among them, the video surveillance method for a wind farm site refers to using video surveillance technology to monitor and manage the internal safety and operation status of a wind farm. The main uses of this method include monitoring the working status of wind turbines, detecting and warning of potential safety hazards, recording maintenance work, and ensuring the safety of staff. Through real-time video feedback, operators can quickly understand the operation of the wind farm, respond promptly to faults or safety issues, and effectively improve the operation efficiency and safety standards of the wind farm.

[0004] Existing video surveillance technologies are limited by the performance of equipment in wind farm applications. Especially in adverse weather conditions such as fog or at night, the image quality and clarity of traditional cameras drop significantly, affecting the monitoring effect and accuracy. Traditional video surveillance systems mainly rely on manual review and inspection of video content, which is not only inefficient but also prone to missed detections due to operator fatigue. In the face of large-scale wind farm operation and maintenance management, this inefficient monitoring method is difficult to meet the high requirements for real-time and comprehensiveness. In terms of safety management, existing technologies lack an active warning mechanism and can only analyze problems after the fact, which to a certain extent delays the implementation of countermeasures and leads to the occurrence or deterioration of safety accidents. For example, in a wind farm, if abnormal behaviors such as a ship deviating from its course cannot be detected and warned in time, it will cause equipment damage or more serious safety accidents, affecting the safe operation and power production of the entire wind farm. Summary of the Invention

[0005] The purpose of the present invention is to solve the drawbacks existing in the prior art, and to propose a video surveillance method and system based on a wind farm site.

[0006] To achieve the above purpose, the present invention adopts the following technical solution. A video surveillance method based on a wind farm site includes the following steps:

[0007] S1: Use an intelligent panoramic array camera, set the aperture size, adjust the exposure time, check the image clarity in the night vision mode, record the frequency and power of the fog penetration function, cover the monitoring range of the wind farm, and obtain panoramic real-time images;

[0008] S2: Analyze the moving pixel points in the panoramic real-time images, set the color and range of the moving pixel points, perform pixel point contour extraction, compare the extracted data with the static background, and generate moving target data;

[0009] S3: Perform pattern matching on the moving target data, mark the timestamps and position coordinates of illegal behaviors, classify and record abnormal events, and generate filtered target records;

[0010] S4: Compare the vessel positions in the filtered target records with the predetermined navigation routes, calculate the offset and speed changes of the navigation routes, perform deviation evaluation, and give early warnings to generate intelligent warning signals;

[0011] S5: According to the intelligent warning signals, set the nodes and edges of the graph theory model, regard the wind farm as a network, analyze the fault propagation paths, adjust the weights of the nodes and edges to perform path simulation, and construct a fault propagation prediction model;

[0012] S6: Based on the fault propagation prediction model, formulate emergency response measures, adjust the frequency and scope of the maintenance plan, set the key nodes for regular inspections, and allocate maintenance tasks to generate maintenance and emergency response plans.

[0013] As a further solution of the present invention, the panoramic real-time images include images in the high-definition night vision mode, views in the multi-frequency fog penetration mode, and the covered monitoring area. The moving target data includes the color range of the detected moving pixel points, the contour extraction data, and the background comparison results. The filtered target records include the timestamps, position coordinates, and classifications of the analyzed illegal behaviors. The intelligent warning signals include warning signals based on path offset and speed changes, and warning classifications of abnormal events. The fault propagation prediction model includes the nodes and edges of the wind farm network analysis, the fault propagation paths, and the node weight simulation. The maintenance and emergency response plans include the strategies of emergency measures, the regular inspections of key nodes, and the allocation of maintenance tasks.

[0014] As a further solution of the present invention, the steps of using an intelligent panoramic array camera, setting the aperture size, adjusting the exposure time, checking the image clarity in the night vision mode, recording the frequency and power of the fog penetration function, covering the monitoring range of the wind farm, and obtaining panoramic real-time images are specifically as follows:

[0015] S101: Use an intelligent panoramic array camera. First, set the aperture to the maximum value to optimize the light intake, adjust the exposure time to match the low-light environment at night, and optimize the image clarity through parameter adjustment to obtain an optimized night vision image.

[0016] S102: Based on the optimized night vision image, set the frequency and power of the fog penetration function, record the effects of the fog penetration function in different environments, and evaluate the fog penetration performance under different conditions to generate fog penetration effect evaluation data.

[0017] S103: Adopt the fog penetration effect evaluation data to adjust the monitoring range of the intelligent panoramic array camera, check and cover the wind farm, and generate panoramic real-time images.

[0018] As a further solution of the present invention, the steps of analyzing the moving pixel points in the panoramic real-time image, setting the color and range of the moving pixel points, performing pixel point contour extraction, and comparing the extracted data with the static background to generate moving target data are specifically as follows:

[0019] S201: Analyze the moving pixel points in the panoramic real-time image, set the color threshold and pixel range for identifying the moving pixel points, extract the moving pixel points through color filtering and size screening, and generate moving pixel initialization data.

[0020] S202: Based on the moving pixel initialization data, perform pixel point contour extraction operations, draw the outer contour of the moving pixel points through edge detection technology, and iteratively optimize the structure of the pixel points to obtain contour optimization data.

[0021] S203: Compare the contour optimization data with the static background in the panoramic real-time image, identify the differences between the background and the moving target, and generate moving target data.

[0022] As a further solution of the present invention, the steps of performing pattern matching on the moving target data, marking the timestamps and position coordinates of illegal behaviors, classifying and recording abnormal events, and generating screened target records are specifically as follows:

[0023] S301: Perform pattern matching on the moving target data, use the defined behavior patterns for comparison and identification, mark the data points that match the known illegal behavior patterns, record the timestamps and position coordinates of the data points, and generate behavior matching records.

[0024] S302: Based on the behavior matching records, classify the identified illegal behaviors, sort and record the events according to the behavior characteristics and occurrence frequencies, and make an event classification table.

[0025] S303: Use the event classification table to evaluate and record the classified abnormal events, evaluate the criticality and urgency, formulate response measures, and generate a screened target record.

[0026] As a further solution of the present invention, the steps of comparing the ship position in the screened target record with a predetermined sailing path, calculating the offset and speed change of the sailing path, performing deviation evaluation, and giving an early warning to generate an intelligent early warning signal are specifically as follows:

[0027] S401: Use the screened target record, according to the real-time position data of the ship, adopt the Kalman filtering algorithm to continuously estimate the ship position and analyze the path offset, and generate path offset data;

[0028] S402: Based on the path offset data, record the speed of the ship at different time points, analyze the relationship between speed and time, and calculate the rate of change of speed to obtain a speed change analysis result;

[0029] S403: Use the speed change analysis result to perform deviation evaluation, combine the data of path offset and speed change, and through a set risk threshold, if exceeded, trigger an alarm to generate an intelligent early warning signal.

[0030] As a further solution of the present invention, the formula of the Kalman filtering algorithm is as follows:

[0031] x a|k = x k|k-1 + A k (z k - H k x k|k-1 )

[0032] Among them, x a|k represents the position estimate at the k-th moment, x k|k-1 represents the predicted position before the k-th moment, A k is the Kalman gain at the k-th moment, z k is the actual observed position at the k-th moment, and H k is the observation matrix.

[0033] As a further solution of the present invention, according to the intelligent early warning signal, set the nodes and edges of the graph theory model, regard the wind farm as a network, analyze the fault propagation path, adjust the weights of the nodes and edges to perform path simulation, and the steps of constructing a fault propagation prediction model are as follows:

[0034] S501: Based on the intelligent early warning signal, set the nodes of the graph theory model, representing multiple key devices and connection paths in the wind farm, and by analyzing the device functions and positions associated with the early warning signal, define the basic structure of the nodes and edges to generate initial structure data;

[0035] S502: Based on the initial structure data, adjust the node and edge weights in the graph. The weight adjustment is based on the risk level and the criticality of the connection, perform a simulation operation of the fault propagation path, predict potential fault diffusion effects through the simulation, and obtain the fault propagation simulation result;

[0036] S503: Use the fault propagation simulation result, combine real-time data and historical fault data, perform dynamic adjustment and optimization, optimize the consistency and response speed of the prediction, and build a fault propagation prediction model.

[0037] As a further solution of the present invention, based on the fault propagation prediction model, formulate emergency response measures, adjust the frequency and scope of the maintenance plan, set key nodes for regular inspections, and allocate maintenance tasks. The steps to generate the maintenance and emergency response plan are specifically as follows:

[0038] S601: Based on the fault propagation prediction model, formulate emergency response measures, prioritize determining response strategies and resource allocation according to the fault risk areas predicted by the model, and generate a resource allocation plan;

[0039] S602: According to the resource allocation plan, adjust the frequency and scope of the maintenance plan, perform inspections on risk nodes and key equipment, expand the maintenance coverage area, and match the potential fault points of the prediction model to obtain the adjusted maintenance plan;

[0040] S603: Adopt the adjusted maintenance plan, set key nodes for regular inspections, and allocate maintenance tasks and responsible persons for the key nodes. Through task allocation and progress monitoring, generate an emergency response plan.

[0041] A video monitoring system based on a wind farm area. The video monitoring system based on the wind farm area is used to execute the above-mentioned video monitoring method for the wind farm area. The system includes:

[0042] The image collection module sets the aperture size of the intelligent panoramic array camera, adjusts the exposure time, enables the night vision function, checks the image clarity, records the fog penetration frequency and power, covers the entire wind farm, and generates panoramic real-time image data;

[0043] The image analysis module analyzes the moving pixel points in the panoramic real-time image data, adjusts the color range of the moving pixel points, extracts the pixel point contours, and compares them with the static background to generate moving target data;

[0044] The event monitoring module filters the moving target data, marks the time and location of abnormal activities, classifies and records different abnormal events, performs pattern matching, and generates abnormal event records;

[0045] The path deviation analysis module compares the vessel positions in the abnormal event records with the predetermined route, calculates the deviation and speed changes, performs deviation assessment, and generates route deviation alerts.

[0046] The maintenance scheduling module sets the nodes and edges of the graph theory model according to the route deviation alerts, simulates the fault propagation path, constructs a fault propagation prediction model, formulates emergency response measures, assigns maintenance tasks, and generates maintenance and emergency response plans.

[0047] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0048] In the present invention, the high-definition panoramic real-time images collected by the intelligent panoramic array camera significantly improve the monitoring efficiency of the wind farm and the accuracy of image analysis. By using advanced image processing techniques, especially in night vision and fog penetration modes, it is ensured that there is no limitation by environmental conditions and clear vision can be maintained in various climates. By analyzing the moving pixel points in the panoramic image and comparing them with the static background, moving targets can be quickly and accurately identified, and potential safety hazards or faults can be detected in a timely manner. This highly automated monitoring and analysis process reduces the dependence on manual intervention, improves the response speed and comprehensiveness of monitoring. By comparing the vessel position with the predetermined navigation path, potential deviation behaviors can be effectively predicted and alerted, further enhancing the initiative and predictability of the field safety management. Regarding the wind farm as a network and analyzing the fault propagation path through the graph theory model not only optimizes the fault handling process but also enhances the understanding of the operating state of the wind farm, effectively guiding the formulation and implementation of the maintenance plan, and ensuring the stable operation and efficient maintenance of the wind farm. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a schematic diagram of the working process of the present invention;

[0050] Figure 2 It is a detailed flowchart of S1 of the present invention;

[0051] Figure 3 It is a detailed flowchart of S2 of the present invention;

[0052] Figure 4 It is a detailed flowchart of S3 of the present invention;

[0053] Figure 5 It is a detailed flowchart of S4 of the present invention;

[0054] Figure 6 It is a detailed flowchart of S5 of the present invention;

[0055] Figure 7 It is a detailed flowchart of S6 of the present invention;

[0056] Figure 8This is the system flowchart of the present invention. Detailed implementation manners

[0057] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0058] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the accompanying drawings. It is only for the convenience of describing the present invention 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 to the present invention. In addition, in the description of the present invention, "a plurality of" means two or more unless otherwise specifically defined.

[0059] Please refer to Figure 1 , the present invention provides a technical solution, a video monitoring method based on a wind farm area, including the following steps:

[0060] S1: Use an intelligent panoramic array camera, set the aperture size, adjust the exposure time, check the image clarity in the night vision mode, record the frequency and power of the fog penetration function by adjusting the laser irradiation angle, cover the monitoring range of the entire wind farm, and obtain panoramic real-time images;

[0061] S2: Analyze the moving pixel points in the panoramic real-time images, set the color and size range of the moving pixel points, perform pixel point contour extraction, compare the extracted data with the static background, and generate moving target data;

[0062] S3: Perform pattern matching on the moving target data, use the stored illegal behavior features, mark the time stamps and position coordinates of the illegal behaviors, classify and record the abnormal events, and generate a screened target record;

[0063] S4: Compare the ship positions in the screened target record with the predetermined navigation paths, calculate the offset and speed changes of the navigation paths, perform deviation evaluation, and give an early warning to generate an intelligent early warning signal;

[0064] S5: According to the intelligent early warning signal, set the nodes and edges of the graph theory model, regard the wind farm as a network, analyze and predict the propagation path of the fault, perform path simulation by adjusting the weights of the nodes and edges, and construct a fault propagation prediction model;

[0065] S6: Based on the fault propagation prediction model, formulate emergency response measures, adjust the frequency and scope of the maintenance plan, set key nodes for regular inspections, perform automated maintenance scheduling, and allocate maintenance tasks to generate maintenance and emergency response plans.

[0066] The panoramic real-time images include images in high-definition night vision mode, views in multi-frequency fog-penetrating mode, and the monitored areas covered. The moving target data includes the color range of detected moving pixel points, contour extraction data, and background contrast results. The screened target records include the timestamps, location coordinates, and classifications of analyzed illegal behaviors. The intelligent early warning signals include early warning signals based on path deviation and speed changes, and early warning classifications of abnormal events. The fault propagation prediction model includes nodes and edges of wind farm network analysis, fault propagation paths, and node weight simulations. The maintenance and emergency response plans include the strategies of emergency measures, regular inspections of key nodes, and the allocation of maintenance tasks.

[0067] Please refer to Figure 2 , for the steps of using an intelligent panoramic array camera, setting the aperture size, adjusting the exposure time, checking the image clarity in night vision mode, recording the frequency and power of the fog-penetrating function, and covering the monitoring area of the wind farm to obtain panoramic real-time images:

[0068] S101: When using an intelligent panoramic array camera, first set the aperture to the maximum value to optimize the light input, adjust the exposure time to match the low-light environment at night, and through parameter adjustment, optimize the image clarity. The execution process of obtaining the optimized night vision image is as follows;

[0069] When using an intelligent panoramic array camera for night photography, maximizing the aperture setting is beneficial to improving the light input. Further, finely adjust the exposure time to adapt to the low-light environment. These settings of the camera must be precisely controlled during operation to obtain clear night vision images. By fine-tuning the camera parameters, the image quality can be significantly improved. The maximization of the aperture and the adjustment of the exposure time need to be dynamically configured based on the real-time feedback of the ambient light. The optimization of the light input and the matching of the exposure time are obtained by testing the image effects under different settings and conducting comparative analysis. The operator can adjust the settings according to the actual situation to ensure that each captured image is in the best state and obtain the optimized night vision image.

[0070] S102: Based on the optimized night vision image, set the frequency and power of the fog-penetrating function, record the effects of the fog-penetrating function in different environments, and evaluate the fog-penetrating performance under different conditions to generate the fog-penetrating effect evaluation data. The execution process is as follows;

[0071] Set the frequency and power of the fog-penetrating function and calculate the fog-penetrating effect evaluation data according to the formula where F tRepresents the defogging frequency, L represents the inductance of the defogging coil, and C represents the capacitance. Detailed explanation of the formula and the derivation process of the formula calculation: In defogging technology, an LC circuit is used to set the required defogging frequency, where the inductance L and the capacitance C are specific values ​​obtained through experiments. Taking the inductance of 0.1 Henry and the capacitance of 0.01 Farad as an example, the defogging frequency is calculated using the formula: This indicates that the fog penetration frequency is 1.59Hz, which is suitable for specific low-light environments and foggy conditions and helps optimize image clarity.

[0072] S103: Using the fog penetration effect evaluation data, adjusting the monitoring range of the intelligent panoramic array camera, checking the coverage of the wind farm, and generating a panoramic real-time image. The execution process is as follows;

[0073] Based on the evaluation data of the fog penetration effect, the monitoring range of the intelligent panoramic array camera is adjusted to cover the wind farm, which is an expansion of the application of the technology. Through the panoramic function of the intelligent camera, the image not only includes the overall picture of the wind farm, but also includes environmental details and changes. This adjustment is implemented by analyzing the image quality after fog penetration and evaluating the visibility of each area of ​​the wind farm. The expansion of the monitoring range allows operators to fully monitor the operating status of the wind farm from the central control room, ensure the safety and efficiency of the wind farm, and generate panoramic real-time images.

[0074] See also Figure 3 , analyze the moving pixels in the panoramic real-time image, set the color and range of the moving pixels, perform pixel contour extraction, compare the extracted data with the static background, and generate the moving target data in the following steps:

[0075] S201: Analyze the moving pixels in the panoramic real-time image, set the color threshold and pixel range for identifying the moving pixels, extract the moving pixels through color filtering and size screening, and generate the moving pixel initialization data. The execution process is as follows;

[0076] In panoramic real-time image analysis, the identification of moving pixels depends on setting an appropriate color threshold, which needs to be adjusted according to changes in ambient light and the color characteristics of the target itself. The determination of the pixel range is based on the evaluation of the target size and movement speed. Color filtering and size screening work together in the extraction process of moving pixels. The process involves batch image processing technologies, such as dynamic threshold adjustment and real-time color tracking technology. The capture of each moving pixel is achieved through advanced image recognition algorithms to ensure that only pixels that meet the set threshold are considered to be moving, generating motion pixel initialization data.

[0077] S202: Based on the motion pixel initialization data, a pixel point contour extraction operation is performed, the outline of the motion pixel point is drawn by edge detection technology, and the structure of the pixel point is iteratively optimized to obtain the contour optimization data. The execution process is as follows;

[0078] Perform the pixel contour extraction operation and calculate the contour optimization data according to the formula where E(x,y) represents the edge intensity at the pixel point (x,y), and G x and G y represent the gradients in the x and y directions. Explanation of the formula and the derivation process of the formula calculation: Edge detection technology uses the gradient method to identify the edge regions in the image, and the positions with larger gradient values correspond to the edges. The pixel gradients in the x and y directions can be calculated by the Sobel operator. Given that the gradient of the pixel point (x,y) in the x direction is 30 and the gradient in the y direction is 40, then the calculation of the edge intensity is: E(x,y) = 30 2 + 40 2 = 2500. This indicates that the edge intensity at the pixel point (x,y) is relatively high, indicating that this is the edge part of the target. By iterating this calculation, the contour structure of the entire target can be optimized.

[0079] S203: Compare the contour optimization data with the static background in the panoramic real-time image to identify the differences between the background and the moving target. The execution process of generating the moving target data is as follows;

[0080] According to the contour optimization data, comparing with the static background in the panoramic real-time image is a key step in identifying the moving target, which requires high precision and response speed. Through comparative analysis, the differences between the background and the moving target can be accurately distinguished. The process involves complex image segmentation techniques and target recognition algorithms. The identification of the differences between the background and the moving target depends on the precise application of image processing techniques to ensure the accuracy and reliability of the data, which is crucial for subsequent monitoring and response operations, and generate the moving target data.

[0081] Please refer to Figure 4 to perform pattern matching on the moving target data, mark the timestamps and position coordinates of illegal behaviors, and classify and record the abnormal events. The specific steps for generating the filtered target records are as follows:

[0082] S301: Perform pattern matching on the moving target data, use the defined behavior patterns for comparative identification, mark the data points that match the known illegal behavior patterns, and record the timestamps and position coordinates of the data points. The execution process of generating the behavior matching records is as follows;

[0083] Use the defined behavior patterns for comparative identification and calculate the behavior matching records according to the formula where M i represents the behavior matching degree of the i-th data point, w j represents the weight of the j-th behavior pattern, and x ijRepresents the eigenvalue of the i-th data point under the j-th behavior pattern. Detailed formula explanation and formula calculation derivation process: Behavior pattern matching uses weighted sum to determine the consistency between the data point and the defined pattern. Consider a specific example where a data point has the following eigenvalues: for three behavior patterns (Pattern A, B, C), the eigenvalues x are 0.2, 0.8, and 0.1 respectively. If the weights w of the patterns are set to 0.5, 0.3, and 0.2, then the matching degree is calculated as follows: M i = 0.5·0.2 + 0.3·0.8 + 0.2·0.1 = 0.34. This indicates that the matching degree of this data point with the defined behavior pattern is 0.34. A lower matching degree means that the data point does not match well with any illegal behavior pattern, which helps to determine whether the data point belongs to illegal behavior.

[0084] S302: Based on the behavior matching records, classify the identified illegal behaviors, sort and record the events according to the behavior characteristics and occurrence frequencies. The execution process of creating an event classification table is as follows;

[0085] Classifying and sorting these records according to the behavior matching records is a key link in the entire monitoring system. The identified illegal behaviors need to be classified according to the behavior characteristics and occurrence frequencies, including detailed information and statistical data of various behaviors. Sorting the events not only reflects the severity of the events but also depends on the occurrence frequencies, enabling the monitoring personnel to quickly identify high-risk behavior patterns and prioritize their handling. Based on in-depth analysis of the behavior matching records, it ensures the effective identification and timely response to illegal behaviors and creates an event classification table.

[0086] S303: Use the event classification table to evaluate and record the classified abnormal events, evaluate the criticality and urgency, formulate response measures, and generate the execution process of screening target records as follows;

[0087] The evaluation and recording of abnormal events using the event classification table is an optimization of the monitoring system. The process not only evaluates the criticality and urgency of each type of event but also formulates corresponding response measures, providing a clear basis for subsequent monitoring and intervention, ensuring the rapid response ability of the system, and reflecting in detail the handling priorities of various abnormal events, providing key support for achieving more effective resource allocation and crisis management and generating screening target records.

[0088] Please refer to Figure 5 , compare the vessel positions in the screening target records with the predetermined sailing routes, calculate the offset and speed changes of the sailing routes, perform deviation evaluation, and issue early warnings. The specific steps for generating intelligent early warning signals are as follows:

[0089] S401: The execution process of screening target records, using the Kalman filtering algorithm based on the real-time position data of the vessel to continuously estimate the vessel position and analyze path deviation, and generating path deviation data is as follows;

[0090] The formula of the Kalman filtering algorithm is as follows:

[0091] x a|k = x k|k-1 + A k (z k - H k x k|k-1 )

[0092] Among them, x a|k represents the position estimate at the k-th moment, x k|k-1 represents the predicted position before the k-th moment, A k is the Kalman gain at the k-th moment, z k is the actual observed position at the k-th moment, H k is the observation matrix.

[0093] Detailed explanation of the formula and the derivation process of formula calculation: Set the specific values of each item in the formula. Let x k|k-1 be the position prediction without considering the observation at the k-th moment, set as 5000 meters; let z k be the actual observed position at the k-th moment, obtained through GPS data, set as 5050 meters; let H k be the observation matrix, simplified to 1 here, directly mapping the estimation of the state space to the observation space; let A k be the Kalman gain, calculated according to the error covariance and the observation noise covariance, and set the calculated gain as 0.1.

[0094] Substitute these values into the formula for calculation:

[0095] x a|k = 5000 + 0.1×(5050 - 1×5000)

[0096] = 5000 + 0.1×50

[0097] = 5000 + 5

[0098] = 5005

[0099] The calculated value of x a|k is 5005 meters, indicating the optimal position estimate after considering the observation at the k-th moment. This shows that the position estimate updated by the Kalman filter has been improved compared to before, being closer to the position measured by the actual GPS, indicating that the prediction error has been effectively reduced. The result further strengthens the accuracy of route tracking, providing an accurate data basis for further analysis and adjustment of route deviation.

[0100] S402: Based on the path offset data, record the speed of the vessel at different time points, analyze the relationship between speed and time, calculate the rate of change of speed, and the execution process for obtaining the speed change analysis result is as follows;

[0101] Analyze the relationship between speed and time, and calculate the rate of change of speed. According to the formula Calculate the speed change analysis result. In the formula, r v represents the rate of change of speed, Δv represents the change in speed, and Δt represents the change in time. Detailed explanation of the formula and the derivation process of formula calculation: When analyzing the vessel speed data, the calculation of the rate of change is crucial. Consider a specific example. If within two different time points (for example, a 5-minute interval), the speed of the vessel increases from 20 knots to 25 knots, then the change in speed Δv is 5 knots, and the change in time Δt is 5 minutes. Therefore, the calculation of the rate of change of speed is: indicating that the speed of the vessel increases by 1 knot per minute. This rate of change of speed helps to evaluate the acceleration of the vessel and is an important measure of the vessel's motion state.

[0102] S403: Use the speed change analysis result to perform deviation assessment. Combine the data of path offset and speed change. Through the set risk threshold, if it is exceeded, an alarm will be triggered, and the execution process for generating an intelligent early warning signal is as follows;

[0103] When using the speed change analysis result to perform deviation assessment, the key lies in how to use the data of path offset and speed change to determine whether to trigger an alarm. The process is completed by setting a risk threshold. If the speed change analysis result shows that the rate of change of speed exceeds the preset risk threshold, the system will automatically trigger an alarm. This deviation assessment mechanism ensures that the vessel can respond in a timely manner when encountering potential risks. Based on a comprehensive risk assessment, including rapid speed changes and path offsets, it aims to warn the crew and monitoring system in advance to prevent accidents and generate intelligent early warning signals.

[0104] Please refer to Figure 6 According to the intelligent early warning signal, set the nodes and edges of the graph theory model. Regard the wind farm as a network, analyze the fault propagation path, and adjust the weights of the nodes and edges to perform path simulation. The specific steps for constructing a fault propagation prediction model are as follows:

[0105] S501: Based on the intelligent early warning signal, set the nodes of the graph theory model, representing multiple key devices and connection paths in the wind farm. By analyzing the device functions and locations associated with the early warning signal, define the basic structure of the nodes and edges, and the execution process for generating the initial structure data is as follows;

[0106] By analyzing the device functions and locations associated with the warning signals, initial structure data is generated according to the formula G=(V,E). In the formula, G represents the graph theory model, V represents the node set of devices, and E represents the edge set of connection paths. Explanation of the formula and the derivation process of formula calculation: In a wind farm monitoring system, constructing a graph theory model is an abstract representation of the devices and connection paths within the system. For example, assume a wind farm includes three key devices as nodes, and these devices are connected through certain paths to form the basic structure of the graph. The definitions of nodes and edges are based on the functions and locations of the devices:

[0107] V = {v1, v2, v3}

[0108] and

[0109] E = {(v1, v2), (v2, v3), (v3, v1)}

[0110] Such a graph model helps to understand and analyze the interconnection relationships and fault propagation paths between devices, providing initial structure data for subsequent risk assessment and fault response.

[0111] S502: Based on the initial structure data, adjust the weights of the nodes and edges in the graph. The weight adjustment is based on the risk level and connection criticality, and perform the simulation operation of the fault propagation path. By simulating, predict the potential fault diffusion effect. The execution process of obtaining the fault propagation simulation result is as follows;

[0112] Adjusting the weights of the nodes and edges in the graph based on the initial structure data is a complex process. This process needs to consider the risk level and criticality of each node and edge. The weight adjustment is based on these factors to perform the simulation operation of the fault propagation path. Through simulation, the potential fault diffusion effect can be predicted. The results of this simulation help the technical team better understand how faults spread in the device network of the wind farm when a specific fault occurs, and take corresponding preventive measures and response strategies. The implementation of the simulation operation is based on accurate mathematical models and calculation methods to ensure the accuracy and practicality of the simulation results, and obtain the fault propagation simulation result.

[0113] S503: Use the fault propagation simulation results, combined with real-time data and historical fault data, to perform dynamic adjustment and optimization, optimize the consistency and response speed of the prediction, and the execution process of constructing the fault propagation prediction model is as follows;

[0114] Using the fault propagation simulation results in combination with real-time data and historical fault data for dynamic adjustment and optimization is the core link in constructing a fault propagation prediction model. This process not only improves the consistency of prediction but also enhances the system's response speed. The optimization process includes algorithm adjustment and parameter fine-tuning to ensure that the prediction model can effectively map various faults and anomalies that occur in actual operations. Through continuous data analysis and model updates, it can reflect the equipment status and fault trends in real time, providing scientific decision-making support for the maintenance and operation of wind farms and constructing a fault propagation prediction model.

[0115] Please refer to Figure 7 , based on the fault propagation prediction model, formulate emergency response measures, adjust the frequency and scope of the maintenance plan, set key nodes for regular inspections, and allocate maintenance tasks. The steps to generate the maintenance and emergency response plan are specifically as follows:

[0116] S601: Based on the fault propagation prediction model, formulate emergency response measures. According to the fault risk areas predicted by the model, prioritize the determination of response strategies and resource allocation. The execution process of generating the resource allocation plan is as follows;

[0117] According to the fault risk areas predicted by the model, formulate emergency response measures and generate a resource allocation plan according to the formula where R represents the overall effectiveness of the response measures, p z represents the probability of the z-th fault risk area, and a z represents the effectiveness of the corresponding response strategy. Explanation of the formula and the derivation process of formula calculation: When a fault occurs, the response measures for each risk area need to be considered comprehensively based on the probability of the fault occurrence and the effectiveness of the strategy. For example, if there are three main risk areas with fault occurrence probabilities of 0.2, 0.5, and 0.3 respectively, and the corresponding strategy effectiveness scores are 80, 60, and 90. Then the overall effectiveness of the response measures is calculated as: R = 0.2·80 + 0.5·60 + 0.3·90 = 16 + 30 + 27 = 73. This score reflects the expected effectiveness of the overall emergency response measures, helping decision-makers understand the effect of the current emergency plan and optimize resource allocation accordingly.

[0118] S602: According to the resource allocation plan, adjust the frequency and scope of the maintenance plan, conduct inspections on risk nodes and key equipment, expand the maintenance coverage area, and match the potential fault points of the prediction model. The execution process of obtaining the adjusted maintenance plan is as follows;

[0119] Adjust the frequency and scope of the maintenance plan according to the resource allocation plan to more effectively conduct preventive inspections on risk nodes and key equipment. The expanded maintenance coverage area should match the potential failure points of the prediction model. In this way, problems can be detected and repaired in a timely manner before a failure occurs. The adjustment of the maintenance plan is based on the analysis of the model output and historical maintenance data to ensure that each risk node receives sufficient attention, reduce the overall failure rate of the system, and provide a solid guarantee for the operation safety of the wind farm, resulting in an adjusted maintenance plan.

[0120] S603: Adopt the adjusted maintenance plan, set the key nodes for regular inspections, and assign maintenance tasks and responsible persons for the key nodes. The execution process of generating the emergency response plan through task assignment and progress monitoring is as follows;

[0121] Adopt the adjusted maintenance plan to set the key nodes for regular inspections, and assign maintenance tasks and responsible persons for the nodes. This is a crucial step to ensure the effective implementation of maintenance work. Through clear task assignment and progress monitoring, the completion status of each work can be effectively tracked, which not only improves the efficiency of maintenance work but also ensures the operating stability of key equipment. Based on a comprehensive assessment of the failure risk and resource availability, ensure that a rapid and effective response can be made in case of a failure to protect the continuous operation of the wind farm and generate an emergency response plan.

[0122] Please refer to Figure 8 , a video monitoring system based on the wind farm area. The video monitoring system based on the wind farm area is used to execute the above-mentioned video monitoring method for the wind farm area. The system includes:

[0123] The image collection module sets the aperture size of the intelligent panoramic array camera, adjusts the exposure time, enables the night vision function, checks the image clarity, records the fog penetration frequency and power, covers the entire wind farm, and generates panoramic real-time image data;

[0124] The image analysis module analyzes the moving pixel points in the panoramic real-time image data, adjusts the color range of the moving pixel points, extracts the pixel point contours, and compares them with the static background to generate moving target data;

[0125] The event monitoring module filters the moving target data, marks the time and location of abnormal activities, classifies and records different abnormal events, performs pattern matching, and generates abnormal event records;

[0126] The path deviation analysis module compares the ship position in the abnormal event record with the predetermined route, calculates the offset and speed change, performs deviation evaluation, and generates a route deviation alarm;

[0127] The maintenance scheduling module sets the nodes and edges of the graph theory model according to the route deviation alarm, simulates the fault propagation path, constructs a fault propagation prediction model, formulates emergency response measures, allocates maintenance tasks, and generates maintenance and emergency response plans.

[0128] The above are only the preferred embodiments of the present invention, and do not limit the present invention in other forms. Any person skilled in the art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical content of the technical solution of the present invention, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. A video monitoring method based on a wind farm area, characterized in that: The following steps are involved: Use an intelligent panoramic array camera to set the aperture size, adjust the exposure time, check the image clarity in night vision mode, record the frequency and power of the fog penetration function, cover the monitoring range of the wind farm, and obtain panoramic real-time images; Analyze the moving pixels in the panoramic real-time image, set the color and range of the moving pixels, perform pixel contour extraction, compare the extracted data with the static background, and generate moving target data; Performing pattern matching on the mobile target data, marking the timestamp and location coordinates of illegal behaviors, classifying and recording abnormal events, and generating screening target records; Comparing the vessel position in the screening target record with the predetermined navigation path, calculating the deviation and speed change of the navigation path, performing deviation evaluation, and issuing an early warning to generate an intelligent early warning signal; According to the intelligent early warning signal, the nodes and edges of the graph theory model are set, the wind farm is regarded as a network, the fault propagation path is analyzed, the weights of the nodes and edges are adjusted to perform path simulation, and a fault propagation prediction model is constructed; Based on the fault propagation prediction model, emergency response measures are formulated, the frequency and scope of the maintenance plan are adjusted, key nodes for regular inspections are set, and maintenance tasks are assigned to generate maintenance and emergency response plans.

2. The video monitoring method based on a wind farm area according to claim 1 is characterized in that: The panoramic real-time image includes images in high-definition night vision mode, views in multi-frequency fog-penetrating mode and covered monitoring areas; the moving target data includes the color range of detected moving pixels, contour extraction data and background comparison results; the screening target record includes the timestamp, location coordinates and classification of illegal behaviors after analysis; the intelligent early warning signal includes early warning signals based on path deviation and speed change, and early warning classification of abnormal events; the fault propagation prediction model includes nodes and edges of wind farm network analysis, fault propagation path and node weight simulation; the maintenance and emergency response plan includes strategies for emergency measures, regular inspection of key nodes, and allocation of maintenance tasks.

3. The video monitoring method based on a wind farm area according to claim 1 is characterized in that: Using an intelligent panoramic array camera, setting the aperture size, adjusting the exposure time, checking the image clarity in night vision mode, recording the frequency and power of the fog penetration function, and covering the monitoring range of the wind farm, the specific steps for obtaining panoramic real-time images are as follows: Use the smart panoramic array camera to first set the aperture to the maximum value to optimize the amount of light entering, adjust the exposure time to match the low-light environment at night, and optimize the image clarity by adjusting the parameters to obtain the optimized night vision image; Based on the optimized night vision image, the frequency and power of the fog penetration function are set, the effect of the fog penetration function under differentiated environments is recorded, and the fog penetration performance under differentiated conditions is evaluated to generate fog penetration effect evaluation data; The fog penetration effect evaluation data is used to adjust the monitoring range of the intelligent panoramic array camera, check the coverage of the wind farm, and generate a panoramic real-time image.

4. The video monitoring method based on a wind farm area according to claim 1 is characterized in that: The steps of analyzing the moving pixels in the panoramic real-time image, setting the color and range of the moving pixels, performing pixel contour extraction, and comparing the extracted data with the static background to generate moving target data are as follows: Analyze the moving pixels in the panoramic real-time image, set the color threshold and pixel range for identifying the moving pixels, extract the moving pixels through color filtering and size screening, and generate moving pixel initialization data; Based on the motion pixel initialization data, a pixel point contour extraction operation is performed, the outer contour of the motion pixel point is drawn by edge detection technology, and the structure of the pixel point is iteratively optimized to obtain contour optimization data; The contour optimization data is compared with a static background in a panoramic real-time image, the difference between the background and the moving target is identified, and moving target data is generated.

5. The video monitoring method based on a wind farm area according to claim 1 is characterized in that: The steps of pattern matching the mobile target data, marking the timestamp and location coordinates of illegal behaviors, classifying and recording abnormal events, and generating screening target records are as follows: Performing pattern matching on the mobile target data, performing comparative identification using defined behavior patterns, marking data points that match known illegal behavior patterns, recording timestamps and location coordinates of the data points, and generating behavior matching records; Based on the behavior matching records, the identified illegal behaviors are classified, and the events are sorted and recorded according to the behavior characteristics and occurrence frequency, and an event classification table is prepared; The event classification table is used to evaluate and record the classified abnormal events, evaluate the criticality and urgency, formulate response measures, and generate screening target records.

6. The video monitoring method based on a wind farm area according to claim 1 is characterized in that: The steps of comparing the ship position in the screening target record with the predetermined navigation path, calculating the deviation and speed change of the navigation path, performing deviation evaluation, and issuing an early warning to generate an intelligent early warning signal are specifically as follows: Using the screening target record, according to the real-time position data of the ship, using the Kalman filter algorithm, continuous estimation of the ship's position and path deviation analysis are performed to generate path deviation data; Based on the path deviation data, the speed of the vessel at the differentiated time points is recorded, the relationship between the speed and time is analyzed, and the rate of change of the speed is calculated to obtain a speed change analysis result; The speed change analysis results are used to perform deviation assessment, and the path deviation and speed change data are combined to generate an intelligent early warning signal by triggering an alarm if a set risk threshold is exceeded.

7. The video monitoring method based on a wind farm area according to claim 6 is characterized in that: The formula of the Kalman filter algorithm is as follows: x a|k =x k|k-1 +A k (z k -H k x k|k-1 ) Among them, x a|k represents the position estimate at the kth moment, x k|k-1 represents the predicted position before the kth moment, A k is the Kalman gain at the kth moment, z k is the actual observed position at the kth moment, H k is the observation matrix.

8. The video monitoring method based on a wind farm area according to claim 1 is characterized in that: According to the intelligent early warning signal, the nodes and edges of the graph theory model are set, the wind farm is regarded as a network, the fault propagation path is analyzed, the weights of the nodes and edges are adjusted to perform path simulation, and the steps of building a fault propagation prediction model are as follows: Based on the intelligent early warning signal, nodes of a graph theory model are set to represent multiple key devices and connection paths in the wind farm, and the basic structure of nodes and edges is defined by analyzing the functions and locations of the devices associated with the early warning signal to generate initial structure data; Based on the initial structure data, adjusting the weights of nodes and edges in the graph, the weight adjustment is based on the risk level and the criticality of the connection, performing a simulation operation of the fault propagation path, and predicting the potential fault diffusion effect through simulation to obtain a fault propagation simulation result; The fault propagation simulation results are used in combination with real-time data and historical fault data to perform dynamic adjustment and optimization, optimize the consistency and response speed of prediction, and build a fault propagation prediction model.

9. The video monitoring method based on a wind farm area according to claim 1 is characterized in that: Based on the fault propagation prediction model, emergency response measures are formulated, the frequency and scope of the maintenance plan are adjusted, key nodes for regular inspections are set, and maintenance tasks are assigned. The specific steps for generating maintenance and emergency response plans are as follows: Based on the fault propagation prediction model, emergency response measures are formulated, response strategies and resource allocation are prioritized according to the fault risk areas predicted by the model, and resource allocation plans are generated; According to the resource allocation plan, the frequency and scope of the maintenance plan are adjusted, risk nodes and key equipment are inspected, the maintenance coverage area is expanded, potential failure points of the prediction model are matched, and an adjusted maintenance plan is obtained; The adjusted maintenance plan is adopted to set key nodes for regular inspection, and maintenance tasks and responsible persons are assigned to the key nodes. Through task allocation and progress monitoring, an emergency response plan is generated.

10. A video monitoring system based on a wind farm area, characterized in that: According to the video monitoring method based on a wind farm area according to any one of claims 1 to 9, the system comprises: The image collection module sets the aperture size of the intelligent panoramic array camera, adjusts the exposure time, enables the night vision function, checks the image clarity, records the frequency and power of fog penetration, covers the entire wind farm, and generates panoramic real-time image data; The image analysis module analyzes the moving pixels in the panoramic real-time image data, adjusts the color range of the moving pixels, extracts the outline of the pixels, and compares them with the static background to generate moving target data; The event monitoring module screens the mobile target data, marks the time and location of abnormal activities, classifies and records differentiated abnormal events, performs pattern matching, and generates abnormal event records; The path deviation analysis module compares the vessel position in the abnormal event record with the predetermined route, calculates the deviation and speed change, performs deviation assessment, and generates a route deviation alarm; The maintenance scheduling module sets the nodes and edges of the graph theory model according to the route deviation alarm, simulates the fault propagation path, builds a fault propagation prediction model, formulates emergency response measures, allocates maintenance tasks, and generates maintenance and emergency response plans.

Images