Collaborative Monitoring Method for Corrosion Prevention in Offshore Wind Farms Integrating Swarm Intelligence

Abstract

This invention belongs to the field of intelligent computing and offshore wind power monitoring technology, specifically involving a collaborative monitoring method for corrosion prevention in offshore wind farms that integrates swarm intelligence. The method includes: acquiring real-time salt spray deposition rate and real-time relative humidity; generating local initial screening; obtaining an environmental stress index; obtaining initial pheromones; calculating a comprehensive pheromone for assessing the swarm's collaborative filtration status; calculating the collaborative focusing degree for assessing high-risk targets across the network and guiding inspection resources; obtaining depth probing measurements characterizing substantial damage to the metal substrate; calculating a degradation risk score for extrapolating the probability of potential structural brittle fracture; and generating corrosion prevention scheduling instructions to guide monitoring status. This invention solves the technical problems in existing technologies, such as the lack of collaborative linkage among offshore wind farm nodes, the difficulty in extracting initial corrosion characteristics due to environmental noise, and the inability of inspection resources to dynamically focus on truly high-risk areas.

Description

Technical Field

[0001] This invention relates to the fields of intelligent computing and offshore wind power monitoring technology. More specifically, this invention relates to a collaborative monitoring method for corrosion prevention in offshore wind farms that integrates swarm intelligence. Background Technology

[0002] Offshore wind farms are exposed to extreme environments with high salt spray, high humidity, strong ultraviolet radiation, and the impact of ocean waves. Corrosion of wind turbine towers and underwater foundations is a fatal hidden danger that threatens the safe operation of the entire wind farm. Traditional corrosion monitoring mainly relies on manual periodic inspections or the installation of isolated corrosion sensors at key points.

[0003] However, existing technologies suffer from serious data silos and slow response issues in the vast offshore wind field applications. Single-point sensors can only reflect local physical changes and cannot detect the spatial spread trend of corrosion. Conventional drone or ship inspections usually use fixed-line scanning, which is not only inefficient but also easily wastes a lot of computing power and endurance in irrelevant areas.

[0004] Furthermore, salt spray deposition and temperature and humidity fluctuations in the marine environment create a strong spatiotemporal coupling effect. The initial microscopic corrosion characteristics caused by this environmental stress are easily masked by background noise such as wave stains and marine organism attachment. Existing automated systems lack a globally coordinated sensing mechanism, failing to aggregate and amplify weak anomalous signals scattered across different nodes through an attractor mechanism. This results in slow focusing speed and low positioning accuracy for high-risk corrosion areas. To address the technical problems of lack of coordinated linkage among offshore wind farm nodes, difficulty in extracting initial corrosion characteristics due to environmental noise, and the inability of inspection resources to dynamically focus on truly high-risk areas in existing technologies, this invention proposes a collaborative monitoring scheme based on swarm intelligence algorithms. Summary of the Invention

[0005] To address the technical problems in existing technologies, such as the lack of coordinated operation among offshore wind farm nodes, difficulties in extracting initial corrosion characteristics due to environmental noise, and the inability of inspection resources to dynamically focus on truly high-risk areas, this invention provides a collaborative monitoring method for offshore wind farm corrosion prevention that integrates swarm intelligence, including: Using temperature and humidity sensors and salt spray concentration detectors deployed on the surface of offshore wind turbines, real-time salt spray deposition rate and real-time relative humidity, characterizing external operating conditions, are obtained. Color space conversion and rough texture analysis are performed on the visual image stream to generate a local preliminary screening to characterize the suspected corrosion severity of the node's location. Based on the interaction between real-time salt spray deposition rate and real-time relative humidity, an environmental stress index is obtained. Based on the local preliminary screening and the environmental stress index, a positive amplification compensation operation is performed to obtain initial pheromones. According to the physical distance attenuation mechanism of chemical substances in nature, the initial pheromones are combined with the environmental stress index... The pressure index is used to calculate the comprehensive pheromone for assessing the collaborative aggregation filtering situation of the group; based on the nonlinear mutation amplification effect of the hyperbolic tangent function, combined with the comprehensive pheromone, the collaborative focusing degree is calculated to assess high-risk targets across the entire network and guide inspection resources; based on the magnitude of the collaborative focusing degree, the depth measurement is obtained to characterize substantial damage to the metal substrate; based on the nonlinear exponential penalty mechanism of physical wall thickness loss and environmental pressure, combined with the depth measurement and collaborative focusing degree, the degradation risk score is calculated to extrapolate the probability of potential structural brittle fracture; based on the magnitude of the degradation risk score, anti-corrosion scheduling instructions are generated to guide the monitoring status.

[0006] This invention combines visual image analysis with environmental condition data to perform multi-dimensional analysis and amplification of corrosion characteristics, effectively reducing the impact of environmental noise on the initial identification of corrosion features and improving the accuracy of corrosion anomaly identification. Simultaneously, based on the results of collaborative group analysis, high-risk areas are screened, and inspection routes are dynamically planned, allowing inspection resources to be directed towards truly high-risk corrosion areas, avoiding waste of resources in irrelevant areas and improving the efficiency of inspection and maintenance. Furthermore, by acquiring real structural damage data through depth sensing and combining it with multi-dimensional indicators to calculate risk scores, corrosion risk assessment becomes more realistic, and the generated scheduling instructions make anti-corrosion and maintenance work more targeted, improving the overall effectiveness of wind farm anti-corrosion monitoring.

[0007] Preferably, obtaining real-time salt spray deposition rate and real-time relative humidity characterizing external operating conditions includes: During the monitoring period, temperature and humidity sensors and salt spray concentration detectors deployed on the surface of offshore wind turbines are used to synchronously read the real-time salt spray deposition rate and real-time relative humidity of the corresponding nodes through the Internet of Things communication protocol.

[0008] Preferably, generating a local preliminary screening to characterize the suspected corrosion severity in the area where the node is located includes: The main control chip issues commands to control the optical modules of each node to acquire the visual image stream. The visual image stream is converted from the RGB color space to the HSV color space. Pixels that meet the preset rust hue range and high saturation range are selected as rust pixels using a threshold segmentation algorithm. The number of extracted rust pixels is divided by the total number of pixels in the visual image stream to obtain the color anomaly ratio. The image processing terminal uses the gray-level co-occurrence matrix to calculate the feature value reflecting the surface texture roughness of the visual image stream, which is denoted as texture roughness. The color anomaly ratio and texture roughness are linearly weighted and output as the local preliminary screening of the current coverage area of ​​the node, which is denoted as local preliminary screening.

[0009] Preferably, obtaining the environmental stress index includes: The critical humidity threshold characterizing the abrupt change in the electrochemical corrosion rate of metal surfaces is obtained from the database. Based on the critical humidity threshold, the proportion of temperature and humidity difference when the real-time relative humidity exceeds the critical humidity threshold is calculated. The proportion of temperature and humidity difference is multiplied by the logarithm of the real-time salt spray deposition rate to calculate the corrosion catalytic intensity characterizing the marine microclimate on the structural surface, which is denoted as the environmental stress index. The benchmark stress constant characterizing the background corrosion driving force under standard marine atmospheric conditions is obtained from the database.

[0010] Preferably, obtaining the initial pheromone includes: The local initial screening score and environmental stress index are obtained. Using the local initial screening score as the base, the environmental stress index is used to perform numerical amplification compensation on the local initial screening score through linear multiplication operation to generate a scalar value that characterizes the current physical status and future deterioration trend of the node, which is denoted as the initial pheromone.

[0011] Preferably, calculating the comprehensive pheromone used to assess the group's cooperative aggregation filtering posture includes: The initial pheromone arithmetic mean of all neighboring nodes within a preset communication radius of a unit node is calculated and denoted as the neighborhood average concentration; the average spatial Euclidean distance from the same unit node to the aforementioned neighboring nodes is calculated and denoted as the neighborhood average distance; the overall pheromone content satisfies the following expression: ; In the formula, Indicates comprehensive pheromones; The initial pheromone; It is an environmental stress index; Used as the baseline stress constant; The average concentration is the neighborhood concentration. The dimensionless average neighborhood distance; It is the first minimum positive number; It is the natural logarithm function.

[0012] This invention, by referencing the states of adjacent locations, can distinguish between accidental interference at a single location and large-scale real-world changes, making the assessment results more closely reflect the actual situation on-site. Simultaneously, it can make reasonable assessment corrections for locations surrounding the risk zone, identifying potential spread risks in advance, reducing the possibility of missed assessments, and improving the stability and reliability of the assessment results, providing a more accurate foundation for subsequent focused assessments.

[0013] Preferably, the co-focusing degree satisfies the following expression: ; In the formula, Indicates the degree of collaborative focus; For comprehensive pheromones; This represents the mean of the degree of aggregation. and These are the second and third smallest positive numbers; It is the hyperbolic tangent function.

[0014] This invention, through reasonable numerical differentiation, can suppress numerical interference in normal areas, make the characteristics of high-risk areas more obvious, reduce the impact of fluctuations in normal areas on risk assessment, and provide clear guidance for subsequent on-site operation planning.

[0015] Preferably, obtaining depth probing measurements characterizing substantial damage to the metallic substrate includes: Retrieve the initial design wall thickness corresponding to the node being tested from the database; obtain the benchmark threshold representing the healthy state of the metal structure surface without obvious corrosion from the database; obtain the cooperative focus of all nodes and sort them from high to low; extract nodes that are greater than the benchmark threshold and record them as nodes of interest; based on all extracted nodes of interest, dynamically reconstruct the inspection trajectory of the UAV and record it as the priority inspection trajectory of the UAV; control the inspection equipment to approach the node of interest, use the mounted ultrasonic thickness gauge to emit a probe wave that penetrates the coating, and analyze the echo time difference through the signal processing module of the ultrasonic thickness gauge, and obtain the current absolute thickness of the metal substrate by combining it with the system sound velocity; subtract the current absolute thickness from the initial design wall thickness and record the result as the depth probe measurement.

[0016] This invention can obtain more accurate data on the actual state of facility structures, avoiding the problem of inadequate verification of high-risk areas that may occur along fixed routes. Simultaneously, the acquired actual structural data can verify the preliminary assessment results, providing more realistic physical data support for subsequent risk assessments and improving the accuracy of the final judgment.

[0017] Preferably, the degradation risk score used to extrapolate the probability of potential structural brittle fracture includes: Retrieve the initial design wall thickness corresponding to the tested node from the database, and obtain the risk adjustment coefficient of the system's preset characteristic structural allowable loss limit; the degradation risk score satisfies the following expression: ; In the formula, Indicates the risk of degradation; For depth measurement; For the initial design wall thickness; This is the risk adjustment coefficient; To enhance collaborative focus; It is an exponential function.

[0018] This invention combines the on-site environment with the overall risk situation to more comprehensively assess the likelihood of future structural problems. Even if the current structural damage is not high, it can identify potential risks that may exist under specific conditions, reducing the chances of overlooking potential risks.

[0019] Preferably, generating anti-corrosion scheduling instructions to guide monitoring status includes: The system retrieves the safety threshold score representing the bottom line of operation and maintenance safety from the database, iterates through the degradation risk scores of the entire wind farm grid, records them as global degradation risk scores, and compares the global degradation risk scores with the safety threshold score. For areas where the degradation risk score is greater than the safety threshold score, the system automatically generates an electronic condition monitoring report containing three-dimensional coordinates and degradation risk scores, and simultaneously generates an anti-corrosion scheduling instruction containing the anti-corrosion paint spraying route and usage plan, which is then sent to the offshore automated maintenance platform.

[0020] The beneficial effects of this invention are as follows: It achieves full-process, multi-dimensional monitoring of corrosion status in offshore wind farms, upgrading corrosion monitoring from traditional single-point, static monitoring to global, dynamic, and collaborative monitoring, thus improving the technical level of corrosion monitoring in offshore wind farms. By improving the accuracy of corrosion anomaly identification and the efficiency of inspection and maintenance, potential corrosion hazards in wind farms can be detected and addressed in a timely manner, reducing corrosion damage to wind turbine structures, extending the service life of wind turbines, and lowering equipment maintenance costs. Simultaneously, effectively controlling the corrosion risks of wind turbines enhances the overall operational safety of offshore wind farms, reduces safety accidents caused by structural corrosion, ensures stable power generation in offshore wind farms, and provides strong technical support for the large-scale, high-quality development of the offshore wind power industry. Attached Figure Description

[0021] Figure 1 The flowchart of the collaborative monitoring method for corrosion prevention in offshore wind farms that integrates swarm intelligence, as illustrated in this invention, is shown in the schematic diagram. Figure 2 This diagram illustrates a comparison of the calculation results of the marine wind field node degradation risk score in this invention. Figure 3 The diagram illustrates a heatmap showing the distribution of marine wind field node degradation risk scores in this invention. Detailed Implementation

[0022] This invention discloses a collaborative monitoring method for corrosion prevention in offshore wind farms that integrates swarm intelligence, referring to... Figure 1 This includes steps S1-S4: S1: Using temperature and humidity sensors and salt spray concentration detectors deployed on the surface of offshore wind turbines, real-time salt spray deposition rate and real-time relative humidity, which characterize external operating conditions, are obtained.

[0023] It should be noted that offshore wind farms contain numerous distributed nodes, such as sensor terminals attached to the tower surface and inspection drones. Abstracting these heterogeneous devices into sensing worker bees in a swarm intelligence network is the first step in constructing a collaborative monitoring topology. Simultaneously, collecting basic environmental parameters is a prerequisite for subsequent analysis of external corrosion acceleration pressures. This invention obtains initial boundary conditions for the impact of marine microclimate on structural corrosion using initial node coordinates, real-time salt spray deposition rate, and real-time relative humidity, providing fundamental data for subsequent calculations of environmental stress indices.

[0024] Specifically, by utilizing temperature and humidity sensors and salt spray concentration detectors deployed on the surface of offshore wind turbines, real-time salt spray deposition rate and real-time relative humidity, characterizing external operating conditions, are obtained, including: During the monitoring period, temperature and humidity sensors and salt spray concentration detectors deployed on the surface of offshore wind turbines are used to synchronously read the real-time salt spray deposition rate and real-time relative humidity of the corresponding nodes through the Internet of Things communication protocol.

[0025] Thus, real-time salt spray deposition rate and real-time relative humidity, which characterize external working conditions, were obtained.

[0026] S2: By performing color space conversion and rough texture analysis on the visual image stream, a local preliminary screening is generated to characterize the suspected corrosion severity of the area where the node is located; based on the interaction between real-time salt spray deposition rate and real-time relative humidity, an environmental stress index is obtained; based on the local preliminary screening and the environmental stress index, a positive amplification compensation operation is performed to obtain the initial pheromone.

[0027] It should be noted that directly calculating the corrosion area with high precision in massive image data is computationally expensive. This invention utilizes an edge computing terminal to perform lightweight color space conversion and texture analysis on the acquired visual image stream, quickly providing a preliminary qualitative score characterizing the suspected corrosion level, thus providing an initial anchor point for subsequent group pheromone allocation.

[0028] Specifically, by performing color space transformation and rough texture analysis on the visual image stream, a preliminary local screening is generated to characterize the suspected severity of corrosion in the region where the node is located, including: The main control chip issues commands to control the optical modules of each node to acquire the visual image stream. The visual image stream is converted from the RGB color space to the HSV color space. Pixels that meet the preset rust hue range and high saturation range are selected as rust pixels using a threshold segmentation algorithm. The number of extracted rust pixels is divided by the total number of pixels in the visual image stream to obtain the color anomaly ratio. The image processing terminal uses the gray-level co-occurrence matrix to calculate the feature value reflecting the surface texture roughness of the visual image stream, which is denoted as texture roughness. The color anomaly ratio and texture roughness are linearly weighted and output as the local preliminary screening of the current coverage area of ​​the node, which is denoted as local preliminary screening.

[0029] Thus, a preliminary local screening was obtained, characterizing the suspected severity of corrosion in the areas where each node is located.

[0030] It should be noted that the same initial rust deteriorates at a significantly different rate in dry and high-salt-spray, high-humidity environments. This invention integrates salt spray and humidity data to calculate a dimensionless stress index, aiming to provide a kinetic dimension reference for distinguishing between static defects and high-potential corrosion points.

[0031] Preferably, the environmental stress index is obtained based on the interaction between real-time salt spray deposition rate and real-time relative humidity, including: The critical humidity threshold characterizing the abrupt change in the electrochemical corrosion rate of metal surfaces is obtained from the database. Based on the critical humidity threshold, the proportion of temperature and humidity difference when the real-time relative humidity exceeds the critical humidity threshold is calculated. The proportion of temperature and humidity difference is multiplied by the logarithm of the real-time salt spray deposition rate to calculate the corrosion catalytic intensity characterizing the marine microclimate on the structural surface, which is denoted as the environmental stress index. The benchmark stress constant characterizing the background corrosion driving force under standard marine atmospheric conditions is obtained from the database.

[0032] Thus, an environmental stress index characterizing the intensity of marine climate corrosion catalysis was obtained.

[0033] It should be noted that in swarm intelligence algorithms, pheromones are the core medium guiding group behavior. This invention combines the appearance score of image recognition with the coercive pressure imposed by the environment to generate an initial pheromone concentration, aiming to give each node its initial importance in the swarm cooperative network.

[0034] Preferably, based on local initial screening and environmental stress index, a positive amplification compensation operation is performed to obtain initial pheromones, including: The local initial screening score and environmental stress index are obtained. Using the local initial screening score as the base, the environmental stress index is used to perform numerical amplification compensation on the local initial screening score through linear multiplication operation to generate a scalar value that characterizes the current physical status and future deterioration trend of the node, which is denoted as the initial pheromone.

[0035] Thus, the initial pheromones of the node's physical status and deterioration trend were obtained.

[0036] S3: Based on the physical distance attenuation mechanism of chemical substances in nature, combined with the initial pheromone and the environmental stress index, calculate the comprehensive pheromone used to assess the group's collaborative aggregation filtering status; based on the nonlinear mutation amplification effect of the hyperbolic tangent function, combined with the comprehensive pheromone, calculate the collaborative focusing degree used to assess high-risk targets across the entire network and guide inspection resources.

[0037] It should be noted that isolated pheromones are prone to artifacts and false alarms due to momentary sensor failures or reflections from local water stains. This invention simulates the physical diffusion mechanism of chemical substances in nature and aggregates neighborhood states through the distance decay law. It aims to use group consensus to filter isolated noise and at the same time generate a strong pheromone aggregation effect in truly large-scale corrosion areas.

[0038] Specifically, based on the physical distance attenuation mechanism of chemical substances in nature, and combining the initial pheromone with the environmental stress index, a comprehensive pheromone for assessing the group's cooperative aggregation filtering behavior is calculated, including: The arithmetic mean of the initial pheromone of all neighboring nodes within the preset communication radius of the statistical unit node is denoted as the neighborhood average concentration; the average spatial Euclidean distance from the same unit node to the above-mentioned neighboring nodes is calculated and denoted as the neighborhood average distance.

[0039] The overall pheromone content satisfies the following expression: ; In the formula, Indicates comprehensive pheromones; The initial pheromone; It is an environmental stress index; Used as the baseline stress constant; The average concentration is the neighborhood concentration. The dimensionless average neighborhood distance; It is the first minimum positive number; It is the natural logarithm function.

[0040] In the formula, By using a logarithmic function to smooth the gain of environmental stress, the pheromones of nodes in extreme high salt spray areas are robustly amplified. It follows the inverse square law of decay and absorbs pheromones emitted by neighboring nodes. This means that even if a node has only average initial screening, if it is surrounded by severe corrosion, its overall concentration will be rapidly increased, thus identifying the true high-risk spread zone.

[0041] For example, if a node itself , , After neighborhood statistics, it was found that , , Calculation Calculation results Retain to three decimal places.

[0042] Thus, a comprehensive pheromone for assessing the group's collaborative aggregation filtering situation was obtained.

[0043] It should be noted that the value of the aforementioned preset communication radius depends on the node deployment density on the surface of the offshore wind turbine and the wireless communication protocol used. If the communication radius is set too small, the target node will not be able to obtain enough data from neighboring nodes, making it difficult to leverage the advantages of group collaboration; if it is set too large, it will introduce data noise from unrelated structural surfaces at excessive distances, resulting in blurred spatial characteristics. Preferably, in the corrosion monitoring scenario of offshore wind turbine towers or pile foundations, the preset communication radius is set to a range of 5 to 20 meters; in a preferred embodiment of the present invention, the preset communication radius is set to 10 meters to ensure that each target node can stably and effectively interact with 3 to 8 neighboring nodes in its physical space.

[0044] It should be noted that in a network composed of massive nodes, a mechanism is needed to define which areas are worth deploying drones for close-range in-depth exploration. This invention introduces a hyperbolic tangent nonlinear mapping, placing node pheromones within the network's background mean for game-theoretic application. This suppresses values ​​in ordinary areas, amplifies anomalous mutation points, and generates focusing weights to guide inspection resources.

[0045] Preferably, based on the nonlinear abrupt amplification effect of the hyperbolic tangent function, and combined with comprehensive pheromones, the collaborative focusing degree used to evaluate high-risk targets across the entire network and guide inspection resources is calculated, including: The comprehensive pheromone content of all nodes in the entire network is statistically analyzed and its arithmetic mean is calculated, which is denoted as the mean of clustering degree.

[0046] Co-focusing satisfies the following expression: ; In the formula, Indicates the degree of collaborative focus; For comprehensive pheromones; This represents the mean of the degree of aggregation. and These are the second and third smallest positive numbers; It is the hyperbolic tangent function.

[0047] In the formula, Using the global mean as the base, the local pheromone levels are amplified quadratically to highlight the abnormally high value areas that deviate from the normal population. By utilizing the nonlinear activation properties of the hyperbolic tangent function, when the node concentration is higher than the mean of aggregation, a positive gain is applied and the concentration tends to saturate; when it is lower than the mean of aggregation, it is quickly suppressed to zero or negative, thus completing the screening of focused targets with high discriminative power.

[0048] For example, If at this time the entire network , Calculation This indicates that the score for ordinary regions tends to be 0, while the high-risk node receives a higher focus index. Calculation results Round to two decimal places.

[0049] This provides a collaborative focus for assessing high-risk targets across the entire network and guiding inspection resources.

[0050] S4: Based on the magnitude of the co-focusing degree, obtain the depth measurement that characterizes the substantial damage to the metal substrate; according to the nonlinear exponential penalty mechanism of physical wall thickness loss and environmental pressure, combine the depth measurement and co-focusing degree to calculate the deterioration risk score for extrapolating the probability of potential structural brittle fracture; based on the magnitude of the deterioration risk score, generate anti-corrosion scheduling instructions to guide the monitoring status.

[0051] It should be noted that although basic imagery and swarm intelligence computing can pinpoint hotspots, the ultimate basis for corrosion prevention and maintenance is the degree of wear and tear on the metal itself. Based on the aforementioned high-focus area, this invention dynamically generates the approach flight path of unmanned equipment and uses penetrating ultrasonic measurement to obtain the real physical data of metal substrate peeling.

[0052] Specifically, based on the magnitude of the co-focusing intensity, depth measurements characterizing substantial damage to the metallic substrate are obtained, including: Retrieve the initial design wall thickness corresponding to the node being tested from the database; obtain the benchmark threshold representing the healthy state of the metal structure surface without obvious corrosion from the database; obtain the cooperative focus of all nodes and sort them from high to low; extract nodes that are greater than the benchmark threshold and record them as nodes of interest; based on all extracted nodes of interest, dynamically reconstruct the inspection trajectory of the UAV and record it as the priority inspection trajectory of the UAV; control the inspection equipment to approach the node of interest, use the mounted ultrasonic thickness gauge to emit a probe wave that penetrates the coating, and analyze the echo time difference through the signal processing module of the ultrasonic thickness gauge, and obtain the current absolute thickness of the metal substrate by combining it with the system sound velocity; subtract the current absolute thickness from the initial design wall thickness and record the result as the depth probe measurement.

[0053] Thus, a depth measurement was obtained to characterize substantial physical damage to the metal structure.

[0054] It should be noted that simple depth measurement is a linear concept, but in complex marine environments, damage accompanied by high environmental stress can easily lead to fatigue fracture. This invention couples the absolute loss of physical detection with the co-focusing degree, which represents the combined environmental pressure, in order to deduce the probability of the final catastrophic degradation risk.

[0055] Preferably, based on the nonlinear exponential penalty mechanism of physical wall thickness loss and environmental pressure, combined with depth measurement and synergistic focusing, a degradation risk score is calculated to extrapolate the probability of potential structural brittle fracture, including: Retrieve the initial design wall thickness corresponding to the node under test from the database, and obtain the risk adjustment coefficient of the system's preset characteristic structure allowable loss limit.

[0056] The degradation risk score satisfies the following expression: ; In the formula, Indicates the risk of degradation; For depth measurement; For the initial design wall thickness; This is the risk adjustment coefficient; To enhance collaborative focus; It is an exponential function.

[0057] In the formula, It characterizes the proportion of wall thickness loss in the most basic physical structure; while The result calculated by the swarm intelligence network is then transformed into a nonlinear exponential penalty term. Even if the physical wall thickness loss is not significant at the moment, if the point is in a harsh microenvironment with high swarm concentration, the potential risk score for structural brittle fracture will increase exponentially. This represents the exponential risk increment of the probability of structural brittle fracture caused by a harsh microenvironment.

[0058] For example, if a high-risk node is measured millimeters millimeters, setting , ; calculation yielded The risk index reached approximately 30.19%, far exceeding the 10% calculated solely based on thickness. (Calculation results) Retain to four decimal places.

[0059] It should be noted that, Figure 2A comparison chart of the calculated degradation risk scores for offshore wind farm nodes shows the numerical difference between the simple thickness loss ratio and the comprehensive degradation risk score. The score corresponding to the simple thickness loss ratio is 10.0%, while the score corresponding to the comprehensive degradation risk score is 30.19%, with the comprehensive degradation risk score being significantly higher than that of the simple thickness loss ratio. This difference indicates that this comprehensive degradation risk scoring method can combine the physical losses of nodes with multi-dimensional environmental influencing factors, more comprehensively reflecting the degradation risk status of nodes and helping to identify the degree of risk that cannot be fully reflected by a single thickness loss indicator.

[0060] It should be noted that, Figure 3 A heatmap showing the distribution of degradation risk scores for nodes in an offshore wind farm illustrates the spatial distribution of these scores within the wind farm area. In the region with an X-coordinate of approximately 4–8 km and a Y-coordinate of approximately 2–6 km, node degradation risk scores generally fall within the 20%–40% range, exhibiting a clear clustering of high values. In other areas of the wind farm, node degradation risk scores are mostly concentrated in the 0%–20% range, with overall lower values. This difference in distribution indicates that this scoring method can combine the spatial location of nodes within the wind farm with differences in environmental pressure, allowing areas with higher degradation risk to receive higher scores, thereby distinguishing the degradation risk status of nodes at different locations within the wind farm.

[0061] Thus, a degradation risk score was obtained for extrapolating the probability of potential structural brittle fracture. It should be noted that the ultimate goal of collaborative monitoring of corrosion prevention in offshore wind farms is to prevent problems before they occur and guide operation and maintenance. This invention is the execution terminal of the system. By converting complex algorithm scoring into engineering-executable operation instructions, it realizes a closed loop from intelligent perception to automated maintenance.

[0062] Preferably, based on the magnitude of the deterioration risk score, anti-corrosion scheduling instructions are generated to guide the monitoring status, including: The system retrieves the safety threshold score representing the bottom line of operation and maintenance safety from the database, iterates through the degradation risk scores of the entire wind farm grid, records them as global degradation risk scores, and compares the global degradation risk scores with the safety threshold score. For areas where the degradation risk score is greater than the safety threshold score, the system automatically generates an electronic condition monitoring report containing three-dimensional coordinates and degradation risk scores, and simultaneously generates an anti-corrosion scheduling instruction containing the anti-corrosion paint spraying route and usage plan, which is then sent to the offshore automated maintenance platform.

[0063] Thus, the collaborative monitoring of corrosion prevention in offshore wind farms based on fusion swarm intelligence has been completed.

[0064] While various embodiments of the invention have been shown and described in this specification, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention.

Claims

1. A collaborative monitoring method for corrosion prevention in offshore wind farms integrating swarm intelligence, characterized in that, include: By using temperature and humidity sensors and salt spray concentration detectors deployed on the surface of offshore wind turbines, real-time salt spray deposition rate and real-time relative humidity, which characterize external operating conditions, can be obtained. By performing color space conversion and rough texture analysis on the visual image stream, a local preliminary screening is generated to characterize the suspected corrosion severity of the area where the node is located; based on the interaction between real-time salt spray deposition rate and real-time relative humidity, an environmental stress index is obtained; based on the local preliminary screening and the environmental stress index, a positive amplification compensation operation is performed to obtain the initial pheromone. Based on the physical distance attenuation mechanism of chemical substances in nature, and combined with the initial pheromone and environmental stress index, a comprehensive pheromone is calculated to assess the collaborative aggregation filtering situation of the group; based on the nonlinear mutation amplification effect of the hyperbolic tangent function, and combined with the comprehensive pheromone, a collaborative focusing degree is calculated to assess high-risk targets across the entire network and guide inspection resources. Based on the magnitude of the co-focusing degree, a depth measurement is obtained to characterize the substantial damage to the metal substrate; according to the nonlinear exponential penalty mechanism of physical wall thickness loss and environmental pressure, combined with the depth measurement and co-focusing degree, a deterioration risk score is calculated to extrapolate the probability of brittle fracture of potential structures. Based on the magnitude of the degradation risk score, anti-corrosion scheduling instructions are generated to guide the monitoring status.

2. The method for collaborative monitoring of corrosion prevention in offshore wind farms integrating swarm intelligence as described in claim 1, characterized in that, The acquisition of real-time salt spray deposition rate and real-time relative humidity, characterizing external operating conditions, includes: During the monitoring period, temperature and humidity sensors and salt spray concentration detectors deployed on the surface of offshore wind turbines are used to synchronously read the real-time salt spray deposition rate and real-time relative humidity of the corresponding nodes through the Internet of Things communication protocol.

3. The method for collaborative monitoring of corrosion prevention in offshore wind farms integrating swarm intelligence as described in claim 1, characterized in that, The preliminary local screening for the suspected corrosion severity in the region where the node is located includes: The main control chip issues commands to control the optical modules of each node to acquire the visual image stream. The visual image stream is converted from the RGB color space to the HSV color space. Pixels that meet the preset rust hue range and high saturation range are selected as rust pixels using a threshold segmentation algorithm. The number of extracted rust pixels is divided by the total number of pixels in the visual image stream to obtain the color anomaly ratio. The image processing terminal uses the gray-level co-occurrence matrix to calculate the feature value reflecting the surface texture roughness of the visual image stream, which is denoted as texture roughness. The color anomaly ratio and texture roughness are linearly weighted and output as the local preliminary screening of the current coverage area of ​​the node, which is denoted as local preliminary screening.

4. The method for collaborative monitoring of corrosion prevention in offshore wind farms integrating swarm intelligence as described in claim 1, characterized in that, The obtained environmental stress index includes: The critical humidity threshold characterizing the abrupt change in the electrochemical corrosion rate of metal surfaces is obtained from the database. Based on the critical humidity threshold, the proportion of temperature and humidity difference when the real-time relative humidity exceeds the critical humidity threshold is calculated. The proportion of temperature and humidity difference is multiplied by the logarithm of the real-time salt spray deposition rate to calculate the corrosion catalytic intensity characterizing the marine microclimate on the structural surface, which is denoted as the environmental stress index. The benchmark stress constant characterizing the background corrosion driving force under standard marine atmospheric conditions is obtained from the database.

5. The method for collaborative monitoring of corrosion prevention in offshore wind farms integrating swarm intelligence as described in claim 1, characterized in that, The process of obtaining the initial pheromone includes: The local initial screening score and environmental stress index are obtained. Using the local initial screening score as the base, the environmental stress index is used to perform numerical amplification compensation on the local initial screening score through linear multiplication operation to generate a scalar value that characterizes the current physical status and future deterioration trend of the node, which is denoted as the initial pheromone.

6. The method for collaborative monitoring of corrosion prevention in offshore wind farms integrating swarm intelligence as described in claim 1, characterized in that, The calculation is used to evaluate the comprehensive pheromone used to assess the group's collaborative aggregation filtering posture, including: The initial pheromone arithmetic mean of all neighboring nodes within a preset communication radius of a unit node is calculated and denoted as the neighborhood average concentration; the average spatial Euclidean distance from the same unit node to the aforementioned neighboring nodes is calculated and denoted as the neighborhood average distance; the overall pheromone content satisfies the following expression: ； In the formula, Indicates comprehensive pheromones; The initial pheromone; It is an environmental stress index; Used as the baseline stress constant; The average concentration is the neighborhood concentration. The dimensionless average neighborhood distance; It is the first minimum positive number; It is the natural logarithm function.

7. The method for collaborative monitoring of corrosion prevention in offshore wind farms integrating swarm intelligence as described in claim 1, characterized in that, The collaborative focus degree satisfies the following expression: ； In the formula, Indicates the degree of collaborative focus; For comprehensive pheromones; This represents the mean of the degree of aggregation. and These are the second and third smallest positive numbers; It is the hyperbolic tangent function.

8. The method for collaborative monitoring of corrosion prevention in offshore wind farms integrating swarm intelligence as described in claim 1, characterized in that, The depth probing measurement used to characterize substantial damage to the metallic substrate includes: Retrieve the initial design wall thickness corresponding to the node being tested from the database; obtain the benchmark threshold representing the healthy state of the metal structure surface without obvious corrosion from the database; obtain the cooperative focus of all nodes and sort them from high to low; extract nodes that are greater than the benchmark threshold and record them as nodes of interest; based on all extracted nodes of interest, dynamically reconstruct the inspection trajectory of the UAV and record it as the priority inspection trajectory of the UAV; control the inspection equipment to approach the node of interest, use the mounted ultrasonic thickness gauge to emit a probe wave that penetrates the coating, and analyze the echo time difference through the signal processing module of the ultrasonic thickness gauge, and obtain the current absolute thickness of the metal substrate by combining it with the system sound velocity; subtract the current absolute thickness from the initial design wall thickness and record the result as the depth probe measurement.

9. The method for collaborative monitoring of corrosion prevention in offshore wind farms integrating swarm intelligence as described in claim 1, characterized in that, The calculation used to extrapolate the degradation risk score for the probability of potential structural brittle fracture includes: Retrieve the initial design wall thickness corresponding to the tested node from the database, and obtain the risk adjustment coefficient of the system's preset characteristic structural allowable loss limit; the degradation risk score satisfies the following expression: ； In the formula, Indicates the risk of degradation; For depth measurement; For the initial design wall thickness; This is the risk adjustment coefficient; To enhance collaborative focus; It is an exponential function.

10. The method for collaborative monitoring of corrosion prevention in offshore wind farms integrating swarm intelligence as described in claim 1, characterized in that, The generation of anti-corrosion scheduling instructions to guide monitoring status includes: The system retrieves the safety threshold score representing the bottom line of operation and maintenance safety from the database, iterates through the degradation risk scores of the entire wind farm grid, records them as global degradation risk scores, and compares the global degradation risk scores with the safety threshold score. For areas where the degradation risk score is greater than the safety threshold score, the system automatically generates an electronic condition monitoring report containing three-dimensional coordinates and degradation risk scores, and simultaneously generates an anti-corrosion scheduling instruction containing the anti-corrosion paint spraying route and usage plan, which is then sent to the offshore automated maintenance platform.

Images