A wind turbine corrosion state monitoring system and protection method

Abstract

This application discloses a corrosion status monitoring system and protection method for wind turbine generators, relating to the field of corrosion monitoring technology for wind power equipment. The system includes: a sensor module for real-time acquisition and output of multi-source corrosion data; the sensor module includes a salt spray sensor, a reference electrode, and a temperature and humidity sensor; a data processing module, communicatively connected to the sensor module, and containing a corrosion analysis algorithm, for receiving and storing multi-source corrosion data, performing fusion analysis on the multi-source corrosion data based on the corrosion analysis algorithm, calculating the current corrosion rate, and predicting the remaining protection life of key components; and a decision output module, connected to the data processing module, for generating a corresponding corrosion risk assessment report or outputting corresponding maintenance instructions based on the current corrosion rate and the predicted remaining protection life of key components. This invention has the advantages of achieving multi-parameter collaborative monitoring and real-time data analysis, and improving the accuracy of corrosion protection.

Description

Technical Field

[0001] This invention relates to the field of corrosion monitoring technology for wind power generation equipment, specifically to a corrosion status monitoring system and protection method for wind turbine generator sets. Background Technology

[0002] As a crucial piece of clean energy equipment, wind turbine generators are typically installed in harsh coastal or offshore environments, where they are exposed to corrosive conditions such as high salt spray, high humidity, and fluctuating temperatures for extended periods. These environmental factors lead to severe electrochemical and stress corrosion of the generator's metal structures (such as towers, foundation rings, and nacelles), significantly shortening equipment lifespan and potentially causing structural safety hazards.

[0003] Currently, the corrosion monitoring methods commonly used in the industry have significant limitations: on the one hand, they mainly rely on a single type of corrosion sensor or periodic manual inspections, making it impossible to achieve multi-parameter collaborative monitoring; on the other hand, they lack effective real-time data analysis capabilities, making it difficult to detect potential corrosion risks in a timely manner. Specifically, this manifests as follows: salt spray concentration monitoring and metal potential detection are independent of each other, making it impossible to comprehensively assess corrosion risks; manual inspections are conducted at long intervals, making it difficult to capture dynamic changes in corrosion in a timely manner; corrosion rate calculations mostly use static empirical formulas, which cannot adapt to complex environmental changes; and maintenance decisions rely on manual experience and judgment, resulting in low response efficiency.

[0004] Especially in marine environments, wind turbines face even more severe corrosion challenges. The combined effect of salt spray deposition in the splash zone and tidal changes accelerates metal corrosion, while existing monitoring systems cannot accurately identify corrosion hotspots or establish a dynamic correlation with the turbine's operating status. Furthermore, traditional protective measures such as impressed current cathodic protection and coating maintenance lack data support, often resulting in over-protection or under-protection.

[0005] These technical deficiencies have led to two major problems for wind turbine generators: first, insufficient precision in corrosion protection, which increases unnecessary maintenance costs and may overlook significant safety hazards; and second, a lack of data support for full life-cycle management, making it difficult to scientifically plan overhaul cycles and component replacement schedules. With the continuous increase in the capacity of individual wind turbine generators and the trend towards offshore development, the economic losses and safety risks caused by these problems are becoming increasingly prominent.

[0006] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0007] In view of this, the present invention provides a corrosion status monitoring system and protection method for wind turbine generator sets, which has the advantages of realizing multi-parameter collaborative monitoring and real-time data analysis, and improving the accuracy of corrosion protection.

[0008] In a first aspect, the present invention provides a corrosion status monitoring system for wind turbine generator sets, comprising:

[0009] Sensor modules are respectively deployed on the outer wall of the wind turbine tower, the outer surface of the foundation ring, and the internal space of the nacelle, for real-time acquisition and output of multi-source corrosion data; the sensor modules include a salt spray sensor for measuring the concentration of salt spray deposition in the atmosphere, a reference electrode for measuring the potential distribution on the structural surface, and a temperature and humidity sensor for acquiring ambient temperature and relative humidity.

[0010] The data processing module is communicatively connected to the sensor module and has a built-in corrosion analysis algorithm. It is used to receive and store the multi-source corrosion data, perform fusion analysis on the multi-source corrosion data based on the corrosion analysis algorithm, calculate the current corrosion rate, and predict the remaining protection life of key components.

[0011] The decision output module, connected to the data processing module, is used to generate a corresponding corrosion risk assessment report or output corresponding maintenance instructions based on the current corrosion rate and the predicted remaining protection life of key components.

[0012] In one optional embodiment, the salt spray sensor and the reference electrode are arranged in an array at equal intervals in the splash zone of the tower to collect salt spray deposition concentration and potential data at different locations in the splash zone, and generate a corrosion potential distribution map of the splash zone based on the salt spray deposition concentration and potential data corresponding to each location.

[0013] In one optional implementation, the data processing module is also communicatively connected to a wind turbine operation data acquisition unit to acquire the wind turbine's start-up and shutdown status and real-time power output data; the corrosion analysis algorithm is configured to combine the wind turbine's start-up and shutdown status and power output data to perform dynamic compensation and correction analysis on the multi-source corrosion data.

[0014] In one optional implementation, the decision output module communicates with the management platform of the remote operation and maintenance center through the wind farm's internal network to automatically transmit the generated maintenance instructions to the management platform and trigger the generation of corresponding maintenance work orders.

[0015] Secondly, the present invention also provides a method for corrosion protection of wind turbine generator sets, comprising the following steps:

[0016] S1. Multi-source corrosion data is acquired in real time by sensor modules deployed on the outer wall of the wind turbine tower, the outer surface of the foundation ring, and the interior space of the nacelle; the multi-source corrosion data includes the concentration of salt spray deposition in the atmosphere, the surface potential distribution of the structure, the ambient temperature, and the relative humidity.

[0017] S2. The multi-source corrosion data is processed using the corrosion analysis algorithm built into the data processing module to calculate the current corrosion rate and predict the remaining protection life of key components.

[0018] S3. Receive the current corrosion rate and the predicted remaining protection life of key components through the decision output module, generate a corresponding corrosion risk assessment report or output a corresponding maintenance instruction, and initiate corresponding protection operations or maintenance responses based on the corrosion risk assessment report or the maintenance instruction.

[0019] In an alternative implementation, in S3, the protective operation includes: automatically adjusting the output parameters of the impressed current cathodic protection system electrically connected to the wind turbine steel structure.

[0020] In an alternative implementation, in S3, the maintenance response includes: outputting a coating repair recommendation for the outer wall of the tower or the outer surface of the foundation ring; the coating repair recommendation includes a recommended coating type and a dry film thickness control range.

[0021] In one optional implementation, in S2, the prediction of the remaining protective life of the critical component is achieved by performing life regression prediction based on multi-source corrosion data and historical maintenance records of the component through a trained machine learning model.

[0022] In one optional implementation, the machine learning model is a gradient boosting decision tree regression model, whose input features include, but are not limited to: time-series variation data of salt spray deposition concentration, structural surface potential offset, ambient temperature and relative humidity.

[0023] In an optional implementation, the method further includes:

[0024] S4. After implementing protective operations or maintenance responses, re-acquire multi-source corrosion data from the sensor module, and generate corresponding feedback data based on the re-acquired multi-source corrosion data from the sensor module. Optimize the corrosion analysis algorithm using the feedback data. The feedback data includes one or more of the following: the trend of salt spray concentration change before and after implementing protective operations or maintenance responses; the range of potential value changes and the degree of improvement in distribution uniformity; the trend of changes in ambient temperature and relative humidity; the amount of change in corrosion rate; and the reassessment results of the remaining protective life of key components.

[0025] As can be seen from the above, the wind turbine generator corrosion status monitoring system and protection method provided in this application include a sensor module to collect multi-source corrosion data in real time, a data processing module to perform fusion analysis and predict lifespan, and a decision output module to generate maintenance instructions. Through multi-source data collaborative monitoring and dynamic analysis, it solves the problems of single parameters and slow response of traditional monitoring methods, and has the advantages of realizing multi-parameter collaborative monitoring and real-time data analysis, and improving the accuracy of corrosion protection. Attached Figure Description

[0026] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0027] Figure 1 This is a schematic diagram of a corrosion status monitoring system for a wind turbine generator set according to an embodiment of the present invention;

[0028] Figure 2 This is a schematic flowchart of a corrosion protection method for wind turbine generators according to an embodiment of the present invention. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] In existing technologies, wind turbine generators operate in harsh environments with high salt spray and high humidity for extended periods, making their metal structures susceptible to corrosion. Traditional corrosion monitoring methods rely primarily on single sensors or periodic manual inspections, which suffer from limitations such as limited data acquisition dimensions, poor timeliness of analysis, insufficient accuracy of predictive models, and delayed maintenance response. For example, a single salt spray sensor cannot reflect changes in the electrochemical state of the metal surface, while long intervals between manual inspections make it difficult to capture dynamic corrosion processes in a timely manner, resulting in a lack of real-time data support for maintenance decisions.

[0031] To address the aforementioned issues, the applicant first analyzed the limitations of existing data acquisition and processing methods, finding that a single sensor can only acquire local or single-factor corrosion information, making it difficult to comprehensively assess the overall corrosion status under complex environments. Furthermore, considering the differences in corrosion mechanisms across different parts of a wind turbine generator (such as the outer wall of the tower, the outer surface of the foundation ring, and the interior of the nacelle), it is necessary to strategically deploy multiple types of sensors to cover critical areas. Based on this, the applicant proposes fusing and analyzing salt spray concentration, potential distribution, and temperature and humidity data, dynamically calculating the corrosion rate and predicting remaining lifespan through an algorithmic model. This transforms monitoring data into actionable maintenance instructions, achieving closed-loop management from data acquisition to decision output.

[0032] Therefore, as Figure 1 As shown, this invention proposes a corrosion status monitoring system for wind turbine generator sets, comprising:

[0033] The sensor modules are respectively deployed on the outer wall of the wind turbine tower, the outer surface of the foundation ring, and the internal space of the nacelle, for real-time acquisition and output of multi-source corrosion data. The sensor modules include a salt spray sensor for measuring the concentration of salt spray deposition in the atmosphere, a reference electrode for measuring the potential distribution on the structural surface, and a temperature and humidity sensor for acquiring ambient temperature and relative humidity.

[0034] The data processing module communicates with the sensor module and has a built-in corrosion analysis algorithm. It receives and stores multi-source corrosion data, performs fusion analysis on the multi-source corrosion data based on the corrosion analysis algorithm, calculates the current corrosion rate, and predicts the remaining protection life of key components.

[0035] The decision output module, connected to the data processing module, is used to generate corresponding corrosion risk assessment reports or output corresponding maintenance instructions based on the current corrosion rate and the predicted remaining protection life of key components.

[0036] The salt spray sensor quantifies the concentration of salt spray in the atmosphere by detecting the amount of salt spray particles deposited on the sensor surface. It can be implemented using a conductivity-based sensor, and its output data is used to assess the erosion intensity of the environmental corrosive medium. The reference electrode measures the electrochemical potential of the metal structure surface, typically using a copper / copper sulfate electrode. Potential distribution data is used to identify localized corrosion tendencies. The temperature and humidity sensor collects ambient temperature and relative humidity data, typically using a digital sensor. This data is used to correct the environmental impact factors in the corrosion rate calculation model. The corrosion analysis algorithm in the data processing module is a mathematical model that fuses multi-source corrosion data. It can be implemented using a linear polarization resistance method combined with a machine learning model to comprehensively calculate the corrosion rate and predict remaining life. The decision output module generates executable instructions based on the algorithm results. It can be connected to the maintenance system via a software interface to automate the transmission of maintenance instructions.

[0037] Specifically, sensor modules are deployed in corrosion-sensitive areas such as the outer wall of the tower, the outer surface of the foundation ring, and the interior of the engine room. These modules simultaneously collect multi-dimensional data on the environment and metal surfaces using salt spray sensors, reference electrodes, and temperature and humidity sensors. The data processing module receives this data and performs multi-source fusion using corrosion analysis algorithms. For example, it correlates salt spray concentration with temperature and humidity data to calculate the atmospheric corrosivity index, while simultaneously assessing the electrochemical corrosion tendency of the metal surface using potential distribution data. The algorithm model dynamically calculates the current corrosion rate and predicts the remaining protection life based on historical data and the initial state of the components. The decision output module generates risk assessment reports or maintenance instructions based on the calculation results. For example, when the predicted remaining life is lower than a preset threshold, it automatically triggers coating repair or cathodic protection system adjustment instructions.

[0038] Compared to existing technologies, traditional methods rely on single sensors or manual inspections, failing to achieve real-time acquisition and fusion analysis of multi-dimensional data, leading to incomplete corrosion assessments. This solution, however, utilizes the collaborative deployment of multiple sensor types and algorithm fusion to simultaneously capture environmental corrosive medium concentrations, metal surface electrochemical states, and temperature and humidity changes, significantly improving the comprehensiveness of corrosion status assessment. Furthermore, while existing technologies often rely on static empirical formulas for corrosion rate calculations, this solution employs dynamic data compensation and machine learning models to achieve accurate prediction of corrosion trends, providing a reliable basis for proactive maintenance.

[0039] Through the above technical solution, this application solves the problems of single data dimension and lagging analysis in traditional monitoring methods, and realizes real-time acquisition and fusion processing of multi-source corrosion data. By collaboratively analyzing salt spray concentration, potential distribution, and temperature and humidity data, the corrosion risk level of different locations can be accurately identified, and the remaining lifespan of key components can be dynamically predicted. Maintenance commands based on the algorithm output can directly trigger automated operation and maintenance, such as adjusting cathodic protection parameters or generating coating repair work orders, thereby shortening the maintenance response cycle and reducing the risk of structural failure due to corrosion.

[0040] In one optional implementation, the salt spray sensor and reference electrode are arranged in an equally spaced array in the splash zone of the tower to collect salt spray deposition concentration and potential data at different locations in the splash zone, and to generate a corrosion potential distribution map of the splash zone based on the salt spray deposition concentration and potential data corresponding to each location.

[0041] The splash zone refers to the area on the outer wall of the tower that is frequently in contact with seawater, typically located within the tidal range, and can be specifically defined by a height range of 0-5 meters above the ground. This area experiences significantly higher corrosion rates than other areas due to prolonged exposure to salt spray and wave splash.

[0042] The equidistant array configuration refers to arranging the salt spray sensors and reference electrodes at fixed intervals, specifically by placing a group of sensors at 1-meter intervals. This arrangement can uniformly cover the vertical space of the splash area, avoiding blind spots in data acquisition.

[0043] Among them, the salt spray sensor refers to the device used to detect the concentration of salt spray deposition in the atmosphere. Specifically, it can be implemented using a sensor based on the principle of conductivity measurement, and outputs concentration data by detecting the amount of salt spray particles deposited on the sensor surface.

[0044] The reference electrode is a device used to measure the surface potential of a metal structure. Specifically, it can be a copper / copper sulfate electrode or a silver / silver chloride electrode. It obtains potential distribution data by electrically connecting it to the surface of the tower.

[0045] Among them, the corrosion potential distribution map refers to a spatial visualization map reflecting the corrosion tendency at different locations in the splash zone. Specifically, it can be generated using the inverse distance weighted interpolation algorithm or the Kriging interpolation algorithm, which is achieved by performing spatial correlation analysis between the salt spray concentration and potential data at discrete points.

[0046] Specifically, after the salt spray sensor and reference electrode are arranged at equal intervals in the splash zone, they can systematically collect salt spray concentration and potential data at different heights and locations within the area. Since the corrosion degree in the splash zone is influenced by both salt spray deposition and electrochemical reactions, a mapping relationship between corrosion tendency and spatial location can be established by simultaneously acquiring salt spray concentration and potential data. During data processing, the salt spray concentration data at discrete points is used to assess the intensity of salt spray erosion, while the potential data is used to determine the active corrosion state of the metal structure. After spatial expansion of the discrete data using an interpolation algorithm, the generated corrosion potential distribution map can present the corrosion risk distribution in the splash zone in two-dimensional or three-dimensional form, thereby accurately identifying high-risk areas with abnormal salt spray concentration accumulation or significant potential shifts.

[0047] Compared to existing technologies, traditional methods typically involve randomly deploying sensors in the splash zone, resulting in uneven data acquisition density and spatial blind spots, failing to accurately reflect the corrosion distribution pattern. This proposed solution, however, combines equally spaced array deployment with spatial interpolation algorithms to achieve full-coverage monitoring of the corrosion state in the splash zone, significantly improving the accuracy of corrosion hotspot location.

[0048] Through the above technical solution, this application can accurately identify local areas within the splash zone where the salt spray deposition concentration is abnormally high or the potential distribution is unbalanced, thereby guiding maintenance personnel to prioritize coating repair or cathodic protection parameter adjustment at high-risk locations. This technical solution solves the monitoring blind spot problem caused by traditional random deployment methods, provides data support for the spatially targeted implementation of corrosion protection measures, and improves the spatial resolution and positioning accuracy of the corrosion monitoring system.

[0049] In one optional implementation, the data processing module is also communicatively connected to a wind turbine operation data acquisition unit to acquire the wind turbine's start-up and shutdown status and real-time power output data; the corrosion analysis algorithm is configured to combine the wind turbine's start-up and shutdown status and power output data to perform dynamic compensation and correction analysis on multi-source corrosion data.

[0050] The wind turbine operation data acquisition unit refers to the interface module that acquires equipment operating parameters in real time from the wind turbine generator control system. Specifically, it can be implemented using an industrial protocol converter that communicates with the SCADA system. Its function is to convert the wind turbine's start-up and shutdown status and power output data into input signals that the corrosion analysis algorithm can recognize, thus providing a data foundation for dynamic compensation. Dynamic compensation correction analysis refers to the calculation process of adjusting corrosion monitoring data in real time according to the wind turbine's operating status. Specifically, it can be implemented using a linear compensation factor based on power output data or an environmental weight adjustment algorithm based on start-up and shutdown status. Its function is to eliminate corrosion data measurement deviations caused by mechanical vibration, temperature fluctuations, or airflow changes by introducing equipment operating parameters.

[0051] Specifically, when the wind turbine is operating at high power, mechanical vibration and increased nacelle temperature can cause instantaneous fluctuations in the salt spray deposition concentration data collected by the salt spray sensor. In this case, the corrosion analysis algorithm reads real-time power output data and introduces a compensation coefficient positively correlated with the power value into the salt spray concentration calculation. For example, the measured salt spray concentration when the power exceeds the rated value is multiplied by a dynamic adjustment factor to correct for sensor signal distortion caused by vibration. When the wind turbine is shut down, the salt spray deposition pattern on the tower surface differs from that under operating conditions due to the lack of airflow disturbance. The algorithm identifies start-stop state switching signals and adjusts the environmental factor weighting coefficients in the potential distribution analysis model. For example, in the shutdown state, the influence weight of humidity on potential offset is increased, thereby more accurately reflecting the corrosion tendency under static conditions.

[0052] Compared to existing technologies, current corrosion monitoring systems typically rely solely on fixed environmental parameters for data analysis, neglecting the correlation between fan operating conditions and the dynamic corrosion process. This leads to decreased accuracy of monitoring data during equipment start-up, shutdown, or load changes. This solution integrates a fan operating data acquisition unit and designs a dynamic compensation and correction algorithm to achieve real-time correlation analysis between corrosion monitoring data and equipment operating conditions, thus resolving the measurement error problem caused by neglecting dynamic operating conditions in traditional methods.

[0053] Through the above technical solutions, this application can effectively eliminate the measurement deviation of corrosion data caused by the start-up and shutdown of the fan or power fluctuations, improve the acquisition accuracy of key parameters such as salt spray concentration and potential distribution, and enhance the adaptability of the corrosion rate calculation model to different operating conditions by dynamically adjusting the environmental factor weight coefficients, thereby providing a more reliable data basis for maintenance decisions.

[0054] In one alternative implementation, the decision output module communicates with the management platform of the remote operation and maintenance center through the wind farm's internal network to automatically transmit the generated maintenance instructions to the management platform and trigger the generation of corresponding maintenance work orders.

[0055] The internal network of the wind farm refers to the communication network deployed within the wind power plant. This can be implemented using Ethernet or wireless communication technologies, such as establishing a data transmission channel via fiber optic cables or 4G / 5G modules, to connect field equipment and remote systems and ensure real-time transmission of maintenance commands. The management platform refers to the operation and maintenance task scheduling system, which can be implemented using industrial-grade operation and maintenance management software. For example, it receives commands and automatically generates tasks through an integrated work order management interface, directly converting corrosion monitoring results into executable maintenance operations.

[0056] Specifically, when the data processing module calculates that the remaining protection life of a critical component is below a preset threshold, the decision output module encapsulates the maintenance instructions into a standardized data packet and sends it to the management platform of the remote operation and maintenance center via the wind farm's internal network. After parsing the data packet, the management platform calls the built-in work order generation interface to automatically create a maintenance task. The task content includes the maintenance location, priority, and operation guidelines. For example, when damage to the outer coating of the tower is detected, the system automatically generates a work order containing the coordinates of the repair location and a suggested coating type, and assigns it to the nearest maintenance team for execution. The entire process requires no manual intervention, avoiding delays or information distortion that may occur with traditional methods such as telephone notifications or paper work order delivery.

[0057] Compared to existing technologies, traditional methods rely on manual inspections followed by manual entry of maintenance requests, which carries risks of response delays and operational errors. This solution, however, automates the entire process from corrosion status identification to maintenance task assignment by establishing a data link between the monitoring system and the operations and maintenance platform, thus overcoming the efficiency bottleneck of manual data transfer.

[0058] Through the above technical solution, this application realizes the real-time transmission of maintenance instructions and automatic generation of work orders, which significantly shortens the processing cycle from corrosion early warning to maintenance response, reduces delays or errors caused by manual operation, and improves the operation and maintenance efficiency of corrosion protection for wind turbine generators.

[0059] In addition, such as Figure 2 As shown, this application also provides a method for corrosion protection of wind turbine generator sets, including the following steps:

[0060] Step S1: Real-time acquisition of multi-source corrosion data is achieved through sensor modules deployed on the outer wall of the wind turbine tower, the outer surface of the foundation ring, and the interior space of the nacelle. The multi-source corrosion data includes the concentration of salt spray deposition in the atmosphere, the surface potential distribution of the structure, and the ambient temperature and relative humidity.

[0061] S2. The corrosion analysis algorithm built into the data processing module is used to process multi-source corrosion data, calculate the current corrosion rate, and predict the remaining protection life of key components.

[0062] S3. Receive the current corrosion rate and the predicted remaining protection life of key components through the decision output module, generate the corresponding corrosion risk assessment report or output the corresponding maintenance instructions, and initiate the corresponding protection operation or maintenance response based on the corrosion risk assessment report or maintenance instructions.

[0063] The sensor modules are deployed on the outer wall of the tower, the outer surface of the foundation ring, and the interior space of the nacelle. This refers to installing the sensors in areas of high corrosion risk within the wind turbine generator set. This can be achieved using bolt fixing or magnetic installation methods to ensure close contact between the sensors and the monitored surfaces. Multi-source corrosion data includes salt spray deposition concentration, structural surface potential distribution, ambient temperature, and relative humidity. This can be achieved by simultaneously acquiring data using salt spray sensors, reference electrodes, and temperature and humidity sensors, providing a comprehensive reflection of the multi-dimensional characteristics of the corrosive environment. Corrosion analysis algorithms process the multi-source corrosion data by fusing and analyzing different types of data. This can be achieved using the linear polarization resistance method combined with a machine learning model to dynamically calculate corrosion rates and predict lifetime. Generating maintenance instructions and initiating protective operations translates the analysis results into executable maintenance actions. This can be achieved through preset logical rules or automatic control interfaces, reducing manual decision-making time.

[0064] Specifically, when deploying sensor modules on the outer wall of the tower, the outer surface of the foundation ring, and the interior of the nacelle, the salt spray sensor is installed on the windward side susceptible to salt spray deposition, for example, fixed to the outer wall of the tower two meters above the ground using a bracket; the reference electrode is arranged in a ring array on the outer surface of the foundation ring to ensure that the potential measurement covers the entire circumference; the temperature and humidity sensor is installed near the nacelle ventilation openings to avoid local temperature deviations. After receiving the sensor data, the data processing module first performs outlier filtering, such as removing peak salt spray concentrations caused by momentary sensor malfunctions; then it performs spatial correlation analysis between the salt spray concentration and potential distribution data, and calculates the electrochemical corrosion rate by combining the temperature and humidity data; finally, it predicts the remaining lifespan of the component based on historical corrosion rate curves. When the remaining lifespan is lower than a preset threshold, the decision output module automatically generates maintenance instructions, such as triggering cathodic protection system current adjustment or pushing a coating repair work order to the maintenance personnel's terminal.

[0065] Compared to existing technologies, which typically rely on single-type sensor data or periodic manual inspections (e.g., using only salt spray concentration to determine corrosion levels), this method fails to identify localized corrosion caused by uneven potential distribution. Our method, however, utilizes multi-source data fusion to simultaneously monitor salt spray deposition, electrochemical state, and environmental parameters. For example, it can identify corrosion risks even when salt spray concentration is within limits but potential distribution is abnormal, overcoming the limitations of traditional data. Existing technologies rely on human experience for maintenance decisions, such as developing maintenance plans based on quarterly inspection reports. Our method, however, uses real-time data-driven decision-making to trigger immediate responses to sudden increases in corrosion rates, such as automatically activating emergency protection after a surge in salt spray concentration during typhoon season.

[0066] Through the above technical solutions, this application achieves comprehensive monitoring and real-time response to the corrosion status of wind turbine generators, solving the problems of single data acquisition and delayed analysis in traditional methods. By fusing multi-source data, the coupling effect of salt spray penetration and electrochemical corrosion can be accurately identified, avoiding misjudgments caused by fluctuations in environmental parameters. Automatic generation of maintenance instructions reduces the delay of manual intervention, ensuring that protective measures can be taken in the early stages of corrosion. Dynamic life prediction provides data support for component replacement and maintenance planning, extending the service life of critical structures.

[0067] In one alternative implementation, in step S3, the protective operation includes: automatically adjusting the output parameters of the impressed current cathodic protection system electrically connected to the wind turbine steel structure.

[0068] Among them, the impressed current cathodic protection system refers to a system that applies current to the protected metal structure through an external power source to suppress electrochemical corrosion. Specifically, it can be implemented by using a potentiostat or a thyristor rectifier. By adjusting the output current or voltage, the potential of the metal structure is maintained within the protection range.

[0069] Automatic adjustment of output parameters refers to dynamically adjusting the intensity of the applied current or the voltage value based on real-time monitored corrosion data. Specifically, it can be achieved using a closed-loop control algorithm or a PID controller. It calculates the required protective current in real time by receiving potential distribution data and outputs a control signal.

[0070] Among them, dynamic matching of protection current intensity and current corrosion state refers to using salt spray concentration, potential offset and temperature and humidity data as input variables, and determining the optimal protection parameters through a preset compensation model or adaptive algorithm. Specifically, it can be implemented by fuzzy logic control or neural network model, and optimize the protection strategy in real time according to environmental changes.

[0071] Specifically, when the salt spray sensor detects an increase in salt spray concentration or the reference electrode detects a positive potential shift, the data processing module generates control commands based on a corrosion analysis algorithm. These commands are then transmitted to the potentiostat of the impressed current cathodic protection system via the analog output module. The potentiostat adjusts its output current according to the received commands, stabilizing the potential of the wind turbine's steel structure within a preset protection range. For example, when the potential is above -650mV, the system automatically increases the output current to enhance the protection effect; when the potential is below -1100mV, the current is reduced to avoid over-protection. The entire adjustment process requires no manual intervention and achieves closed-loop control through real-time data feedback.

[0072] Compared to existing technologies, traditional methods rely on manual periodic monitoring and adjustment of protection parameters, resulting in response lag and insufficient protection accuracy. Furthermore, fixed-parameter applied current systems cannot adapt to dynamic changes in environmental factors such as salt spray concentration and humidity, leading to unstable protection performance. This solution eliminates the delay of manual operation through real-time monitoring data and automatic linkage with the protection system, while a dynamic compensation model ensures precise matching of protection parameters with environmental conditions.

[0073] Through the above technical solutions, this application achieves efficient closed-loop control of the cathodic protection system, solves the problem of lag in response to manual adjustment, and improves the adjustment accuracy of protection parameters; by dynamically matching the protection current with the corrosion state, it effectively suppresses local corrosion and over-protection of steel structures in salt spray environment, and extends the service life of key components.

[0074] In one alternative implementation, in step S3, the maintenance response includes: outputting a coating repair recommendation for the outer wall of the tower or the outer surface of the foundation ring; the coating repair recommendation includes a recommended coating type and a dry film thickness control range.

[0075] Among them, the coating repair suggestion refers to the maintenance plan automatically generated based on the real-time collected data on salt spray concentration, temperature and humidity and corrosion rate. Specifically, it can be achieved by using a data matching algorithm to compare the current corrosion conditions with a preset coating performance database, and dynamically selecting coating materials that are suitable for high salt spray or high humidity environments.

[0076] The coating type refers to the type of anti-corrosion coating that is compatible with the current corrosive environment. Specifically, it can be achieved by using epoxy zinc-rich coating or polyurethane coating. The recommendation logic for different coatings is triggered according to the salt spray concentration threshold to ensure the material's corrosion resistance matches the environmental conditions.

[0077] The dry film thickness control range refers to the thickness range that needs to be achieved during coating construction. Specifically, the lower limit value can be calculated by combining the corrosion rate prediction results with the coating protection life model, while the upper limit value can be set with reference to the construction process limitations. The thickness range parameters are output by the algorithm to avoid protection failure due to excessive thinness or material waste due to excessive thickness.

[0078] Specifically, when the sensor module detects that the salt spray concentration on the outer wall of the tower exceeds a preset threshold, the data processing module calls a corrosion analysis algorithm to calculate the current corrosion rate and matches the coating type based on historical maintenance records. For example, when the salt spray concentration reaches 0.5 mg / m³... 3The algorithm automatically selects epoxy zinc-rich coatings with stronger salt spray resistance as the recommended type. Simultaneously, based on the corrosion rate prediction results, if the annual corrosion rate is 0.1 mm, the lower limit of the dry film thickness is calculated to be 400 micrometers, and the upper limit is set to 500 micrometers based on the spraying process limitations, resulting in a final output thickness control range of 400-500 micrometers. This process replaces manual experience-based judgment with dynamic data-driven approaches, achieving standardized generation of maintenance recommendations.

[0079] Compared to existing technologies, current coating maintenance solutions rely on manual inspections and experience-based selection of coating type and thickness, which suffers from high subjectivity, slow response, and poor material compatibility. This solution, through automatic matching of real-time corrosion data with a coating performance database, accurately recommends coating types suitable for the current environment. It also dynamically calculates the thickness range based on corrosion rate, generating data-driven maintenance instructions that significantly improve the targeted nature and standardized application of coating repairs.

[0080] Through the above technical solutions, this application solves the problem of low maintenance efficiency caused by manual judgment, realizes dynamic adaptation between coating type and corrosion conditions, avoids the risk of secondary corrosion caused by improper material selection, and reduces material waste during construction and maintenance costs by standardizing the thickness control range.

[0081] In one optional implementation, in step S2, the prediction of the remaining protective life of the critical component is achieved by using a trained machine learning model to perform life regression prediction based on multi-source corrosion data and historical maintenance records of the component.

[0082] Among them, multi-source corrosion data refers to the salt spray deposition concentration, structural surface potential distribution, ambient temperature and relative humidity data collected by salt spray sensors, reference electrodes and temperature and humidity sensors. Specifically, it can be implemented using a time-series data acquisition and storage module to comprehensively reflect the coupled influence of dynamic environmental factors on the corrosion process.

[0083] Among them, the component historical maintenance record refers to the time nodes and specific measures of maintenance operations such as coating repair and cathodic protection parameter adjustment. This can be realized through the database storage and retrieval module, and is used to quantify the contribution of maintenance intervention to corrosion inhibition.

[0084] The trained machine learning model refers to a data-driven model built on the gradient boosting decision tree regression algorithm. Specifically, it can be implemented using an open-source machine learning framework to capture the non-linear relationship between multi-source data and maintenance records.

[0085] Among them, lifetime regression prediction refers to the prediction task with the remaining protection lifetime as a continuous numerical output. Specifically, it can be achieved through the regression model training and inference module, which is used to generate a quantitative basis for maintenance decisions in the time dimension.

[0086] Specifically, time-series data on salt spray deposition concentration are input into the model to characterize the continuous corrosive effect of salt spray on metal surfaces; surface potential shifts are input to reflect real-time changes in electrochemical corrosion tendency; and ambient temperature and relative humidity data are input to assess the accelerating effect of humid and hot environments on corrosion rates. Simultaneously, historical maintenance records of components are converted into feature vectors and input into the model, such as the number and time intervals of coating repairs, and the magnitude and frequency of cathodic protection current adjustments, enabling the model to learn the inhibitory effects of maintenance measures on the corrosion process. The model uses a gradient boosting decision tree algorithm to perform multi-level nonlinear combinations of the above features, establishing a mapping relationship from input features to remaining lifetime, and finally outputting a predicted value for the remaining protection lifetime.

[0087] Compared to existing technologies, traditional methods typically use empirical formulas to calculate remaining life, considering only a single corrosion factor and ignoring the impact of maintenance interventions, leading to predictions that deviate from actual operating conditions. This solution, by introducing a machine learning model, can simultaneously process the time-series features of multi-source corrosion data and the event-based features of maintenance records, effectively capturing the complex interactions between environmental factors, electrochemical states, and human interventions, thus significantly improving prediction accuracy.

[0088] Through the above technical solution, this application solves the prediction bias problem caused by the incompleteness of data and the lack of maintenance information in traditional empirical models, and realizes dynamic quantitative assessment of remaining service life. The temporal variation characteristics of salt spray concentration and potential data are accurately correlated with the corrosion rate evolution trend, and the feature encoding of historical maintenance records enables the model to identify the long-term effect differences of different maintenance strategies, thereby automatically correcting the service life extension effect of maintenance measures in the prediction. The resulting remaining service life prediction value can be directly used to formulate differentiated maintenance cycle planning to avoid over-maintenance or under-maintenance.

[0089] In one alternative implementation, the machine learning model is a gradient boosting decision tree regression model, whose input features include, but are not limited to: time-series variation data of salt spray deposition concentration, structural surface potential offset, ambient temperature and relative humidity.

[0090] The gradient boosting decision tree regression model is an ensemble learning algorithm that iteratively generates multiple decision trees and sums the results for prediction. Specifically, it can be implemented using the GBDT framework from the open-source machine learning library, improving overall prediction accuracy by combining weak learners. This model can effectively handle the nonlinear relationships between multi-source corrosion data, overcoming the dependence of traditional empirical formulas on linear assumptions.

[0091] Among them, the time-series variation data of salt spray deposition concentration refers to the dynamic change information of salt spray deposition amount continuously recorded over time. Specifically, this can be achieved by periodically collecting and storing time-series data using salt spray sensors, which is used to reflect the cumulative effect of salt spray concentration fluctuations on the corrosion process. This feature enhances the model's ability to predict long-term corrosion trends by capturing the temporal changes of dynamic environmental factors.

[0092] The surface potential offset refers to the deviation of the metal surface potential from its initial state. Specifically, it can be calculated by measuring the potential with a reference electrode and then determining the difference between the measured potential and the baseline value. This characteristic characterizes the degree of change in the metal's electrochemical corrosion tendency. It is directly related to the corrosion reaction kinetics, providing a quantitative indicator of the electrochemical corrosion state for the model.

[0093] Among them, ambient temperature and relative humidity refer to real-time environmental parameters collected by temperature and humidity sensors. Specifically, this can be achieved by periodically measuring and transmitting data using digital sensors, and is used to construct the synergistic relationship between temperature and humidity on corrosion rate. This feature, by quantifying the influence of environmental conditions on corrosion reaction rate, improves the model's physical correlation with corrosion mechanism.

[0094] Specifically, the gradient boosting decision tree regression model, by integrating the prediction results of multiple decision trees, can handle the complex nonlinear interactions between salt spray concentration, potential shift, temperature, and humidity. During model training, the time-series variation data of salt spray deposition concentration is decomposed into sliding window statistics, such as the average, standard deviation, and trend slope of the past 30 days, to characterize the short-term fluctuations and long-term cumulative effects of salt spray deposition. The structural surface potential shift is calculated by comparing the current potential with the reference potential at initial installation, quantifying the change in the polarization state of the metal structure and reflecting localized corrosion activity. Ambient temperature and relative humidity are input as continuous variables, contributing to the corrosion environment parameter matrix along with the salt spray data. The model automatically filters key influencing factors through feature importance analysis and performs supervised learning based on historical maintenance records and actual lifespan data, ultimately outputting a predicted value for the remaining protective life.

[0095] In some specific implementations, the time-series variation data of salt spray deposition concentration can be further processed into an hourly average sequence, and a 24-hour moving average curve can be extracted as an input feature. The surface potential offset can be obtained by calculating the difference between two adjacent measurements at the same monitoring point, and the rate of potential change can be used as a supplementary feature. Ambient temperature and relative humidity can be combined into a temperature-humidity joint index; for example, a temperature-humidity covariance matrix can be used to characterize the comprehensive impact of environmental conditions. The hyperparameters of the gradient boosting decision tree regression model can be optimized through grid search; for example, the maximum tree depth can be set to 5-8 layers, and the learning rate can be controlled within the range of 0.1-0.3 to balance model complexity and training efficiency.

[0096] Compared to existing technologies, traditional methods often employ linear regression models or empirical formulas to predict remaining lifetime, considering only the static relationship between a single environmental parameter and the corrosion rate, and failing to handle the nonlinear coupling effects between multi-source data. This proposed solution, however, utilizes a gradient-boosted decision tree regression model to automatically learn the synergistic effects of dynamic fluctuations in salt spray concentration, changes in potential shift, and environmental parameters, while simultaneously capturing the cumulative effects of corrosion processes using time-series data. Compared to a single decision tree model, the ensemble learning method significantly reduces the risk of overfitting and improves the generalization ability to different geographical environments and operating conditions.

[0097] Through the above technical solution, this application solves the problem of insufficient accuracy caused by the reliance on empirical formulas in the prediction of remaining service life in existing technologies, and realizes dynamic fusion analysis of multi-dimensional corrosion data. The model accurately quantifies the long-term impact of salt spray deposition through time-series data on salt spray concentration changes, directly reflects the metal corrosion activity by combining potential shift, and improves the physical correlation of corrosion mechanisms using environmental parameters, ultimately outputting more reliable prediction results for remaining protection life. Simultaneously, the model has online learning capabilities, continuously optimizing prediction accuracy through feedback data and adapting to the corrosion evolution patterns under different environmental conditions.

[0098] In one optional implementation, the method of this application further includes:

[0099] S4. After implementing protective operations or maintenance responses, re-acquire multi-source corrosion data from the sensor modules, and generate corresponding feedback data based on the re-acquired multi-source corrosion data from the sensor modules. Optimize the corrosion analysis algorithm using the feedback data. The feedback data includes one or more of the following: the trend of salt spray concentration changes before and after implementing protective operations or maintenance responses; the range of potential value changes and the degree of improvement in distribution uniformity; the trend of changes in ambient temperature and relative humidity; the amount of change in corrosion rate; and the reassessment results of the remaining protective life of key components.

[0100] Among them, re-collecting multi-source corrosion data refers to acquiring salt spray concentration, potential distribution, temperature and humidity data again through sensor modules after completing protective operations or maintenance responses. Specifically, this can be achieved by using timed triggering or event triggering, such as starting data acquisition within a preset time period after maintenance operations are completed, to reflect the actual effect of protective measures.

[0101] Among them, generating feedback data refers to comparing and analyzing the newly collected data with the historical data before maintenance, and extracting indicators such as the trend of salt spray concentration change and the degree of improvement in potential distribution uniformity. Specifically, this can be achieved through difference calculation, trend fitting or statistical analysis methods, and is used to quantitatively evaluate the degree of improvement of the corrosion state by protective measures.

[0102] Among them, optimizing the algorithm using feedback data refers to inputting feedback data into the corrosion analysis algorithm and adjusting the algorithm parameters or model weights. Specifically, this can be achieved using online learning, incremental training, or parameter adaptive calibration methods. For example, the weight parameters of the machine learning model can be updated using the gradient descent method, so that the algorithm can adapt to the state after new environmental conditions and protective measures.

[0103] Specifically, after coating repair or adjustment of cathodic protection parameters, the sensor module re-collects data such as salt spray concentration and potential distribution. For example, data is continuously collected within 24 hours after maintenance, generating feedback indicators such as the percentage reduction in salt spray concentration and the reduction in the standard deviation of potential. This data is input into the corrosion analysis algorithm. For instance, by comparing the change in corrosion rate before and after maintenance, the weight coefficients of the salt spray concentration feature in the gradient boosting decision tree model are adjusted, thereby optimizing the remaining life prediction model. This forms a closed-loop control mechanism of monitoring-decision-feedback-optimization, enabling the corrosion analysis algorithm to dynamically adjust according to the actual protection effect, overcoming the prediction bias caused by environmental changes or differences in protection measures in traditional static models.

[0104] Compared to existing technologies, current corrosion protection methods are typically based on fixed models or empirical formulas, lacking a mechanism for dynamic optimization based on actual maintenance results. This leads to a decrease in predictive accuracy after changes in environmental conditions or adjustments to protective measures. This solution introduces a feedback data closed-loop optimization mechanism, enabling the corrosion analysis algorithm to adapt to new data distributions in real time, thus improving the adaptability and long-term reliability of the prediction model.

[0105] Through the above technical solution, this application solves the prediction bias problem caused by the lack of real-time adjustment capability in existing corrosion analysis models. It achieves dynamic optimization of algorithm parameters based on actual protection effects, improving the accuracy of corrosion rate calculation and remaining life prediction. Furthermore, through a closed-loop feedback mechanism, the system can automatically adapt to the influence of different environmental conditions and protection measures, thereby providing more reliable data support for subsequent maintenance decisions and extending the effective protection life of critical components.

[0106] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A corrosion condition monitoring system for wind turbine generator sets, characterized in that, include: Sensor modules are respectively deployed on the outer wall of the wind turbine tower, the outer surface of the foundation ring, and the internal space of the nacelle, for real-time acquisition and output of multi-source corrosion data; the sensor modules include a salt spray sensor for measuring the concentration of salt spray deposition in the atmosphere, a reference electrode for measuring the potential distribution on the structural surface, and a temperature and humidity sensor for acquiring ambient temperature and relative humidity. The data processing module is communicatively connected to the sensor module and has a built-in corrosion analysis algorithm. It is used to receive and store the multi-source corrosion data, perform fusion analysis on the multi-source corrosion data based on the corrosion analysis algorithm, calculate the current corrosion rate, and predict the remaining protection life of key components. The data processing module is also communicatively connected to a wind turbine operation data acquisition unit to acquire the wind turbine's start-up and shutdown status and real-time power output data; the corrosion analysis algorithm is configured to combine the wind turbine's start-up and shutdown status and power output data to perform dynamic compensation and correction analysis on the multi-source corrosion data. The decision output module, connected to the data processing module, is used to generate a corresponding corrosion risk assessment report or output corresponding maintenance instructions based on the current corrosion rate and the predicted remaining protection life of key components.

2. The system according to claim 1, characterized in that, The salt spray sensor and the reference electrode are arranged in an array at equal intervals in the splash zone of the tower to collect salt spray deposition concentration and potential data at different locations in the splash zone, and generate a corrosion potential distribution map of the splash zone based on the salt spray deposition concentration and potential data corresponding to each location.

3. The system according to claim 1, characterized in that, The decision output module communicates with the management platform of the remote operation and maintenance center through the wind farm's internal network to automatically transmit the generated maintenance instructions to the management platform and trigger the generation of corresponding maintenance work orders.

4. A method for corrosion protection of wind turbine generator sets using the system described in any one of claims 1 to 3, characterized in that, Includes the following steps: S1. Multi-source corrosion data is acquired in real time by sensor modules deployed on the outer wall of the wind turbine tower, the outer surface of the foundation ring, and the interior space of the nacelle; the multi-source corrosion data includes the concentration of salt spray deposition in the atmosphere, the surface potential distribution of the structure, the ambient temperature, and the relative humidity. S2. The multi-source corrosion data is processed using the corrosion analysis algorithm built into the data processing module to calculate the current corrosion rate and predict the remaining protection life of key components. S3. Receive the current corrosion rate and the predicted remaining protection life of key components through the decision output module, generate a corresponding corrosion risk assessment report or output a corresponding maintenance instruction, and initiate corresponding protection operations or maintenance responses based on the corrosion risk assessment report or the maintenance instruction.

5. The method according to claim 4, characterized in that, In S3, the protective operation includes: automatically adjusting the output parameters of the impressed current cathodic protection system electrically connected to the wind turbine steel structure.

6. The method according to claim 4, characterized in that, In S3, the maintenance response includes: outputting a coating repair recommendation for the outer wall of the tower or the outer surface of the foundation ring; the coating repair recommendation includes a recommended coating type and a dry film thickness control range.

7. The method according to claim 4, characterized in that, In S2, the prediction of the remaining protective life of key components is achieved by using a trained machine learning model to perform life regression prediction based on multi-source corrosion data and historical maintenance records of the components.

8. The method according to claim 7, characterized in that, The machine learning model is a gradient boosting decision tree regression model, and its input features include, but are not limited to: time-series variation data of salt spray deposition concentration, structural surface potential offset, ambient temperature and relative humidity.

9. The method according to claim 4, characterized in that, The method further includes: S4. After implementing protective operations or maintenance responses, re-acquire multi-source corrosion data from the sensor module, and generate corresponding feedback data based on the re-acquired multi-source corrosion data from the sensor module. Optimize the corrosion analysis algorithm using the feedback data. The feedback data includes one or more of the following: the trend of salt spray concentration change before and after implementing protective operations or maintenance responses; the range of potential value changes and the degree of improvement in distribution uniformity; the trend of changes in ambient temperature and relative humidity; the amount of change in corrosion rate; and the reassessment results of the remaining protective life of key components.

Images