Offshore wind turbine generator operation state evaluation method

Abstract

The invention relates to the technical field of offshore wind turbine generators, and discloses an offshore wind turbine generator running state evaluation method. Comprising the steps of deploying a sensor network on an offshore wind turbine generator, carrying out data acquisition and preprocessing, detecting states of key components, carrying out overall state evaluation, early warning and decision support, and carrying out preventive overhaul and maintenance, summarization and recording. According to the operation state evaluation method for the offshore wind turbine generator set, comprehensive monitoring and scientific management of the offshore wind turbine generator set and auxiliary equipment are realized by integrating technologies of multi-source data acquisition, state monitoring and early warning and the like, and the limitation that information islands depend on artificial experience in a traditional operation and maintenance mode is broken through depending on an intelligent sensing network and big data analysis, so that the operation state evaluation method for the offshore wind turbine generator set is realized. The digital operation and maintenance management capability of the whole life cycle of the wind power plant equipment is established, and the fault early warning accuracy and the operation and maintenance efficiency are remarkably improved through intelligent monitoring.

Description

Technical Field

[0001] This invention relates to the field of offshore wind turbine technology, specifically to a method for evaluating the operating status of offshore wind turbines. Background Technology

[0002] Offshore wind turbines consist of a rotor (blades), nacelle (including core components such as generators and gearboxes), tower, foundation structure (such as monopiles, jacket foundations, or floating platforms), and submarine cables. The rotor's rotation drives the generator to produce electricity, which is then transmitted to the onshore power grid via submarine cables. Compared to onshore wind power, offshore wind power offers advantages such as higher wind speeds, lower turbulence intensity, and no need for land use, making it suitable for large-scale development.

[0003] Offshore wind power operates in the marine environment for extended periods, enduring far more severe challenges than onshore wind power. The sea areas where wind turbines are located are characterized by extreme temperature variations, high humidity, and salt spray corrosion, making the electrical equipment inside the turbines highly susceptible to failure under these harsh conditions. Traditional offshore wind power operation and maintenance (O&M) relies primarily on a combination of scheduled inspections and fault-based maintenance, which has revealed numerous shortcomings in practical applications. Scheduled inspections, conducted at fixed intervals, ignore the differences in actual equipment conditions, leading to over-maintenance of some equipment while failing to detect potential problems in others. When equipment malfunctions, factors such as sea conditions and the availability of maintenance vessels hinder the rapid arrival of repair personnel, resulting in prolonged downtime losses. Offshore wind turbine monitoring involves the real-time transmission and sharing of information from numerous subsystems, including power, communication, and environmental monitoring. It lacks data collaboration based on a single data source, real-time interaction of massive amounts of data, and dynamic visualization. Furthermore, traditional O&M relies excessively on human experience and judgment, lacking scientific predictive maintenance and intelligent decision support capabilities, making it difficult to meet the O&M needs of large-scale offshore wind farms. Therefore, a method for assessing the operational status of offshore wind turbines is needed to address these issues. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method for evaluating the operating status of offshore wind turbines, thus solving the problems mentioned in the background.

[0005] This invention provides the following technical solution: a method for evaluating the operating status of offshore wind turbines, comprising the following steps: Step S1: Deploy a sensor network on the offshore wind turbine. Step S2: Perform data acquisition and preprocessing; Step S3: Detect the status of key components; Step S4: Conduct an overall status assessment; Step S5: Early Warning and Decision Support; Step S6: Preventive inspection and maintenance; Step S7: Summarize and record.

[0006] Preferably, in step S1, based on the characteristics and fault mechanisms of different equipment in the offshore wind turbine, differentiated monitoring schemes are deployed at different locations of the wind turbine to form a multi-dimensional data acquisition system. The sensor deployment scheme includes: Step S11, Deployment of blade monitoring sensors: Strain gauges and vibration sensors are installed on multiple key sections of the blade to capture load and vibration information at different locations. The key sections are selected based on the areas of the blade that are subjected to greater stress and are prone to failure during operation, and are determined through finite element analysis. At the same time, high-definition cameras are installed at appropriate locations around the blade to monitor the clearance. Step S12, Deployment of transmission chain monitoring sensors: Vibration sensors and temperature sensors are installed on the spindle, gearbox, and generator to monitor their vibration amplitude, abnormal oil temperature, and speed fluctuation parameters. The installation position of the vibration sensors is determined according to the vibration characteristics of the components. Step S13, Deployment of bolt monitoring sensors: An ultrasonic sensor is installed on the bolt. Based on the principle of acoustoelasticity, the axial force of the bolt is measured by measuring the rate of change of ultrasonic wave propagation time. The sensor installation position should ensure that it can accurately reflect the stress state of the bolt. Step S14, Deployment of monitoring sensors for support structure: Strain and displacement sensors were installed around the tower and foundation. A Gaussian mixture model was used to analyze the health degradation trend of the tower and foundation scour support structure. The sensor installation locations were determined based on the stress characteristics and scour patterns of the support structure.

[0007] Preferably, in step S2, data acquisition and preprocessing includes: Step S21, Multi-source data fusion: It integrates heterogeneous data from multiple sources, including SCADA, CMS, vibration sensors, temperature sensors, and strain gauges. To address the issue of different data formats across systems, it employs a data conversion middleware to unify and convert the data into a standard format, covering key parameters such as wind speed, rotational speed, power, temperature, pressure, and vibration. Step S22, Data Cleaning and Standardization: The raw data is noise filtered by using median filtering to remove impulse noise and wavelet thresholding to remove Gaussian noise. Outlier removal is performed by using the 3σ principle to identify and remove abnormal data. For data loss, linear interpolation or spline interpolation is used to complete the data. After eliminating sensor errors and data loss, the data is standardized using the Z-score method. Step S23, Feature Extraction and Dimensionality Reduction: Feature parameters are extracted using time-domain analysis, frequency-domain analysis, and time-frequency analysis. Principal component analysis and dimensionality reduction techniques such as isometric ISOMAP are used to reduce data redundancy and highlight key features.

[0008] Preferably, in step S3, the detection of the status of critical components includes: Step S31, Leaf monitoring includes: Step S311, Load and Vibration Analysis: The dynamic characteristics of the blade are captured by multi-section load monitoring and broadband vibration response, and the blade condition is determined by combining material strain / load threshold analysis and time series data trend analysis. Step S312, Clearance Monitoring: Use a high-definition camera to measure the distance between the blade tip and the tower in real time, and combine audio and video monitoring to identify blade icing and crack damage; Step S32, transmission chain monitoring includes: Step S321, Vibration and Temperature Analysis: Monitor the vibration amplitude, abnormal oil temperature, and speed fluctuation parameters of the spindle, gearbox, and generator; combine the LightGBM model to screen features and construct degradation trend indicators. Step S322, Oil abrasive analysis: The wear degree of the gearbox is detected by the content of metal particles in the oil; Step S33, Bolt monitoring: Based on the principle of acoustoelasticity, the axial force of bolts is measured using the rate of change of ultrasonic propagation time, and the load distribution on the flange surface is inverted using a finite element model. Step S34, Support Structure Monitoring: A Gaussian mixture model is used to analyze the health status decline trend of the tower and foundation scour support structure. An early warning is triggered when the log-likelihood probability of a new observation exceeds a threshold.

[0009] Preferably, in step S4, to address the challenges of complex and variable offshore wind power environments and the difficulty of evaluating the operation and maintenance of offshore wind turbines, the following methods are used to improve the accuracy of offshore wind turbine operation status prediction: Step S41: Use the random forest algorithm to predict wind power output. By setting parameters such as the number of trees in the forest and the feature selection method, train the random forest model and input wind speed, wind direction and meteorological data to predict wind power output. Step S42: Use the fuzzy C-means clustering algorithm to classify the risk level of wind turbine operation, set the number of cluster centers and fuzzy factor parameters, and classify the wind turbine operation status into four levels: healthy, sub-healthy, qualified and abnormal based on the wind turbine full-state analysis and fault diagnosis and early warning results. Step S43: Construct an offshore wind turbine operating state transition model based on the Markov chain principle, determine the state transition probability matrix, obtain the transition probability between different states by analyzing historical operating state data, and predict the future operating state of the wind turbine.

[0010] Preferably, in step S4, the results of the wind turbine's full-state analysis and fault diagnosis and early warning are fully combined, and the probability of a fault is judged by comparing the indicators of healthy, sub-healthy, qualified, and abnormal.

[0011] Preferably, in step S5, in the early warning and decision support, the predictive maintenance decision support system provides intelligent decision suggestions for operation and maintenance strategy formulation based on the equipment status assessment results and combined with multi-dimensional information such as weather forecasts and maintenance resources. The predictive maintenance decision support system uses a Bayesian network to construct a maintenance decision model and uses an expert knowledge base to perform reasoning analysis on various fault symptoms.

[0012] Preferably, the predictive maintenance decision support system is designed for the characteristics of offshore wind farms. It constructs a task management system that includes modules such as scheduled inspections, technical upgrades, and scheduling. Based on weather windows and constraints on personnel and spare parts, the system automatically generates the optimal maintenance plan.

[0013] Preferably, in step S6, the construction personnel bring spare parts and construction tools to the site and carry out standardized construction in accordance with the standardized construction process. The construction process includes safety inspection before construction, equipment shutdown operation, disassembly and replacement of faulty parts, installation and debugging of new parts, and equipment start-up and operation test after construction.

[0014] Preferably, in step S7, after the construction is completed, the staff effectively records and summarizes the operation and maintenance process, including equipment failure status, maintenance measures, information on replaced spare parts, construction time and personnel, and determines the comprehensive operation and maintenance cost based on the composition of operation and maintenance costs.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This offshore wind turbine operation status assessment method classifies the operation status of offshore wind turbines into levels, namely healthy, sub-healthy, qualified, and abnormal indicators. The probability of failure is judged by comparing the healthy, sub-healthy, qualified, and abnormal indicators. After the system determines the level of the wind turbine operation status, the predictive maintenance decision support system constructs a task management system that includes modules such as scheduled inspection, technical transformation, and scheduling, based on the characteristics of offshore wind farms. Based on weather windows and constraints of personnel and spare parts, the system automatically generates the optimal maintenance plan.

[0016] 2. The offshore wind turbine operation status assessment method deploys differentiated monitoring schemes at different locations of the wind turbine based on equipment characteristics and fault mechanisms, forming a multi-dimensional data acquisition system. Through the deployment of a multi-source sensor network, it achieves status monitoring of the entire wind turbine chain. The coordinated configuration of various sensors effectively overcomes the limitations of single monitoring methods and significantly improves the comprehensiveness and accuracy of status monitoring. Attached Figure Description

[0017] Figure 1 This is a flowchart of the offshore wind turbine operation status assessment method of the present invention; Figure 2 This is a flowchart of the sensor deployment scheme of the present invention; Figure 3 This is a flowchart of the data acquisition and preprocessing process of the present invention; Figure 4 This is a flowchart illustrating the detection process for the status of key components in this invention. Figure 5 This is a flowchart of the blade monitoring and drivetrain monitoring of the present invention; Figure 6 This is a flowchart for the overall status assessment of the present invention. Detailed Implementation

[0018] 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, and 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.

[0019] Please see Figure 1 The method for assessing the operational status of offshore wind turbines includes the following steps: Step S1: Deploy a sensor network on the offshore wind turbine. Step S2: Perform data acquisition and preprocessing; Step S3: Detect the status of key components; Step S4: Conduct an overall status assessment; Step S5: Early Warning and Decision Support; Step S6: Preventive inspection and maintenance; Step S7: Summarize and record.

[0020] In step S1, based on the characteristics of different equipment in the offshore wind turbine (such as the aerodynamic characteristics of the blades, the mechanical transmission characteristics of the drive train, and the stress characteristics of the bolts) and the failure mechanisms (such as fatigue fracture of the blades, wear failure of the drive train, and loosening failure of the bolts), differentiated monitoring schemes are deployed at different locations of the wind turbine to form a multi-dimensional data acquisition system. The sensor deployment scheme includes: Step S11, Deployment of blade monitoring sensors: Strain gauges and vibration sensors are installed at multiple key sections of the blade (such as the blade root, blade middle, and blade tip) to capture load and vibration information of the blade at different locations. The key sections are selected based on the areas of the blade that are subjected to greater stress and are prone to failure during operation, and are determined through finite element analysis. At the same time, high-definition cameras are installed at appropriate locations around the blade to monitor the clearance. Step S12, Deployment of transmission chain monitoring sensors: Vibration sensors and temperature sensors are installed on the spindle, gearbox, and generator to monitor their vibration amplitude, abnormal oil temperature, and speed fluctuation parameters. The installation position of the vibration sensor is determined according to the vibration characteristics of the component to ensure that fault characteristic signals can be accurately captured. Step S13, Deployment of bolt monitoring sensors: An ultrasonic sensor is installed on the bolt. Based on the principle of acoustoelasticity, the axial force of the bolt is measured by measuring the rate of change of ultrasonic wave propagation time. The sensor installation position should ensure that it can accurately reflect the stress state of the bolt. Step S14, Deployment of monitoring sensors for support structure: Strain and displacement sensors were installed around the tower and foundation. A Gaussian mixture model was used to analyze the health degradation trend of the tower and foundation scour support structure. The sensor installation locations were determined based on the stress characteristics and scour patterns of the support structure. By deploying a multi-source sensor network, the entire chain of wind turbine condition monitoring is achieved. The coordinated configuration of various sensors effectively overcomes the limitations of a single monitoring method and significantly improves the comprehensiveness and accuracy of condition monitoring.

[0021] In step S2, data acquisition and preprocessing include: Step S21, Multi-source data fusion: Integrating multi-source heterogeneous data from SCADA (Supervisory and Data Acquisition System), CMS (Condition Monitoring System), vibration sensors, temperature sensors, and strain gauges, and addressing the issue of different data formats from different systems, a data conversion middleware is used to uniformly convert the data into a standard format, such as JSON, covering key parameters of wind speed, rotational speed, power, temperature, pressure, and vibration, ensuring data integrity and consistency; Step S22, Data Cleaning and Standardization: The raw data is noise filtered by using median filtering to remove impulse noise and wavelet thresholding to remove Gaussian noise. Outlier removal is performed by using the 3σ principle to identify and remove abnormal data. For data loss, linear interpolation or spline interpolation is used to complete the data. After eliminating sensor errors and data loss, Z-score standardization is used to unify the data units and improve the accuracy of subsequent analysis. Step S23, Feature Extraction and Dimensionality Reduction: Feature parameters are extracted using time-domain analysis (calculating statistics such as mean, variance, and peak value), frequency-domain analysis (converting time-domain signals to frequency-domain signals using FFT), and time-frequency analysis (analyzing the time-frequency characteristics of signals using STFT). Principal component analysis (PCA) is then used to reduce dimensionality by calculating the correlation coefficient matrix, eigenvalues, and eigenvectors of the data. Finally, isometric ISOMAP dimensionality reduction is employed, involving steps such as constructing neighborhood graphs, calculating geodesic distances, and performing multidimensional scaling analysis to reduce data redundancy and highlight key features.

[0022] In step S3, the detection of the status of key components includes: Step S31, Leaf monitoring includes: Step S311, Load and Vibration Analysis: By monitoring the load on multiple sections, the stress distribution of the blade at different sections is obtained. Combined with material strain / load threshold analysis, when the stress exceeds the threshold, it is determined that the blade may be at risk of overload. Wideband vibration response (0.01Hz - 5000Hz) is used to capture the dynamic characteristics of the blade. By analyzing the trend of time series data, such as calculating the autocorrelation function and power spectral density of the vibration signal, it is determined whether the blade condition is normal. Step S312, Clearance Monitoring: Utilize a high-definition camera to measure the distance between the blade tip and the tower in real time. Employ image processing algorithms (such as edge detection and distance measurement algorithms) to accurately calculate the distance value, achieving centimeter-level measurement accuracy. Combined with audio and video monitoring, use deep learning algorithms (such as convolutional neural networks) to identify damage features such as blade icing and cracks. Issue timely warnings when damage is detected. Step S32, transmission chain monitoring includes: Step S321, Vibration and Temperature Analysis: Monitor parameters such as vibration amplitude, abnormal oil temperature, and speed fluctuation of the spindle, gearbox, and generator. Use the LightGBM model to screen features. By setting feature importance thresholds, select features that are sensitive to fault diagnosis and construct degradation trend indicators, such as calculating the root mean square value of vibration signals over time to achieve early warning. Step S322, Oil abrasive analysis: The wear degree of the gearbox is detected by the content of metal particles in the oil. Ferrography analysis is used to prepare ferrography slides of oil samples. The morphology, size and number of metal particles are observed under a microscope. The wear state of the gearbox is judged based on the particle characteristics. Step S33, Bolt monitoring: Based on the principle of acoustoelasticity, the axial force of bolts is measured using the rate of change of ultrasonic propagation time. The measurement formula is: Δt = kΔF, where Δt is the rate of change of ultrasonic propagation time, ΔF is the change in bolt axial force, and k is a constant related to the material and bolt geometry. Combined with the finite element model, the load distribution on the flange surface is inverted. By establishing a three-dimensional finite element model of the bolt-flange connection and inputting the measured bolt axial force data, the stress distribution on the flange surface is obtained, thus realizing early warning of bolt loosening risk. Step S34, Support Structure Monitoring: A Gaussian mixture model was used to analyze the health degradation trend of supporting structures such as towers and foundation scour. The model was constructed by collecting strain and displacement data of the supporting structures at different time points. An early warning was triggered when the log-likelihood probability of a new observation exceeded a set threshold (determined through statistical analysis of historical data).

[0023] In step S4, to address the challenges of complex and variable offshore wind power environments and the difficulty in assessing the operation and maintenance of offshore wind turbines, the following methods are used to improve the accuracy of offshore wind turbine operation status prediction: Step S41: Use the random forest algorithm to predict wind power output. By setting parameters such as the number of trees in the forest and the feature selection method, train the random forest model and input wind speed, wind direction and meteorological data to predict wind power output. Step S42: Use the fuzzy C-means clustering algorithm to classify the risk level of wind turbine operation, set the number of cluster centers and fuzzy factor parameters, and classify the wind turbine operation status into four levels: healthy, sub-healthy, qualified and abnormal based on the wind turbine full-state analysis and fault diagnosis and early warning results. Step S43: Construct an offshore wind turbine operating state transition model based on the Markov chain principle, determine the state transition probability matrix, obtain the transition probability between different states by analyzing historical operating state data, predict the future operating state of the wind turbine, and fully combine the wind turbine full state analysis and fault diagnosis and early warning results to judge the probability of failure by comparing with healthy, sub-healthy, qualified and abnormal indicators.

[0024] In step S5, the predictive maintenance decision support system provides intelligent decision-making suggestions for the formulation of operation and maintenance strategies based on the equipment condition assessment results and combined with multi-dimensional information such as weather forecasts and maintenance resources. A maintenance decision model is constructed using Bayesian networks to determine the nodes in the network (such as equipment status, weather conditions, maintenance resources, etc.) and the probabilistic relationships between nodes. Through an expert knowledge base (containing various fault symptoms, solutions, and historical operation and maintenance data), various fault symptoms are reasoned and analyzed to generate solutions and issue them to the field in the form of work orders. To address the characteristics of offshore wind farms, a task management system was constructed that includes modules for scheduled inspections, technical upgrades, and scheduling. Based on weather windows (time periods for maintenance operations determined by weather forecasts) and constraints on personnel and spare parts (such as personnel skill levels and spare parts inventory), a genetic algorithm was used to automatically generate the optimal maintenance plan. By setting parameters such as population size, crossover probability, and mutation probability, the system can automatically generate the optimal maintenance plan.

[0025] In step S6, construction personnel, carrying spare parts and construction tools, rush to the site and carry out standardized construction according to the standardized construction process. The construction process includes pre-construction safety inspection, equipment shutdown operation, disassembly and replacement of faulty parts, installation and debugging of new parts, and post-construction equipment start-up and operation testing. In step S7, after the construction is completed, the staff effectively records and summarizes the operation and maintenance process, including equipment failure status, maintenance measures, information on replaced spare parts, construction time and personnel. At the same time, the comprehensive operation and maintenance cost is determined based on the composition of operation and maintenance costs, providing a reference for subsequent operation and maintenance decisions.

[0026] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for assessing the operational status of offshore wind turbines, including the following steps: Step S1: Deploy a sensor network on the offshore wind turbine. Step S2: Perform data acquisition and preprocessing; Step S3: Detect the status of key components; Step S4: Conduct an overall status assessment; Step S5: Early Warning and Decision Support; Step S6: Preventive inspection and maintenance; Step S7: Summary and Recording.

2. The method for evaluating the operating status of offshore wind turbines according to claim 1, characterized in that, In step S1, based on the characteristics and fault mechanisms of different equipment in offshore wind turbines, differentiated monitoring schemes are deployed at different locations of the wind turbines to form a multi-dimensional data acquisition system. The sensor deployment scheme includes: Step S11, Deployment of blade monitoring sensors: Strain gauges and vibration sensors are installed on multiple key sections of the blade to capture load and vibration information at different locations. The key sections are selected based on the areas of the blade that are subjected to greater stress and are prone to failure during operation, and are determined through finite element analysis. At the same time, high-definition cameras are installed at appropriate locations around the blade to monitor the clearance. Step S12, Deployment of transmission chain monitoring sensors: Vibration sensors and temperature sensors are installed on the spindle, gearbox, and generator to monitor their vibration amplitude, abnormal oil temperature, and speed fluctuation parameters. The installation position of the vibration sensors is determined according to the vibration characteristics of the components. Step S13, Deployment of bolt monitoring sensors: An ultrasonic sensor is installed on the bolt. Based on the principle of acoustoelasticity, the axial force of the bolt is measured by measuring the rate of change of ultrasonic wave propagation time. The sensor installation position should ensure that it can accurately reflect the stress state of the bolt. Step S14, Deployment of monitoring sensors for support structure: Strain and displacement sensors were installed around the tower and foundation. A Gaussian mixture model was used to analyze the health degradation trend of the tower and foundation scour support structure. The sensor installation locations were determined based on the stress characteristics and scour patterns of the support structure.

3. The method for evaluating the operating status of offshore wind turbines according to claim 1, characterized in that, In step S2, data acquisition and preprocessing include: Step S21, Multi-source data fusion: It integrates heterogeneous data from multiple sources, including SCADA, CMS, vibration sensors, temperature sensors, and strain gauges. To address the issue of different data formats across systems, it employs a data conversion middleware to unify and convert the data into a standard format, covering key parameters such as wind speed, rotational speed, power, temperature, pressure, and vibration. Step S22, Data Cleaning and Standardization: The raw data is noise filtered by using median filtering to remove impulse noise and wavelet thresholding to remove Gaussian noise. Outlier removal is performed by using the 3σ principle to identify and remove abnormal data. For data loss, linear interpolation or spline interpolation is used to complete the data. After eliminating sensor errors and data loss, the data is standardized using the Z-score method. Step S23, Feature Extraction and Dimensionality Reduction: Feature parameters are extracted using time-domain analysis, frequency-domain analysis, and time-frequency analysis. Principal component analysis and dimensionality reduction techniques such as isometric ISOMAP are used to reduce data redundancy and highlight key features.

4. The method for evaluating the operating status of offshore wind turbines according to claim 1, characterized in that, In step S3, the detection of the status of critical components includes: Step S31, Leaf monitoring includes: Step S311, Load and Vibration Analysis: The dynamic characteristics of the blade are captured by multi-section load monitoring and broadband vibration response, and the blade condition is determined by combining material strain / load threshold analysis and time series data trend analysis. Step S312, Clearance Monitoring: Use a high-definition camera to measure the distance between the blade tip and the tower in real time, and combine audio and video monitoring to identify blade icing and crack damage; Step S32, transmission chain monitoring includes: Step S321, Vibration and Temperature Analysis: Monitor the vibration amplitude, abnormal oil temperature, and speed fluctuation parameters of the spindle, gearbox, and generator. Combine the LightGBM model to screen features and construct degradation trend indicators. Step S322, Oil abrasive analysis: The wear degree of the gearbox is detected by the content of metal particles in the oil; Step S33, Bolt monitoring: Based on the principle of acoustoelasticity, the axial force of bolts is measured using the rate of change of ultrasonic propagation time, and the load distribution on the flange surface is inverted using a finite element model. Step S34, Support Structure Monitoring: A Gaussian mixture model is used to analyze the health status decline trend of the tower and foundation scour support structure. An early warning is triggered when the log-likelihood probability of a new observation exceeds a threshold.

5. The method for evaluating the operating status of offshore wind turbines according to claim 1, characterized in that, In step S4, to address the challenges of complex and variable offshore wind power environments and the difficulty of evaluating the operation and maintenance of offshore wind turbines, the following methods are used to improve the accuracy of offshore wind turbine operation status prediction: Step S41: Use the random forest algorithm to predict wind power output. By setting parameters such as the number of trees in the forest and the feature selection method, train the random forest model and input wind speed, wind direction and meteorological data to predict wind power output. Step S42: Use the fuzzy C-means clustering algorithm to classify the risk level of wind turbine operation, set the number of cluster centers and fuzzy factor parameters, and classify the wind turbine operation status into four levels: healthy, sub-healthy, qualified and abnormal based on the wind turbine full-state analysis and fault diagnosis and early warning results. Step S43: Construct an offshore wind turbine operating state transition model based on the Markov chain principle, determine the state transition probability matrix, obtain the transition probability between different states by analyzing historical operating state data, and predict the future operating state of the wind turbine.

6. The method for evaluating the operating status of offshore wind turbines according to claim 5, characterized in that, In step S4, the results of the wind turbine's full-state analysis and fault diagnosis and early warning are fully combined, and the probability of a fault is judged by comparing the indicators of healthy, sub-healthy, qualified, and abnormal.

7. The method for evaluating the operating status of offshore wind turbines according to claim 1, characterized in that, In step S5, in the early warning and decision support section, the predictive maintenance decision support system provides intelligent decision suggestions for operation and maintenance strategy formulation based on the equipment status assessment results and combined with multi-dimensional information such as weather forecasts and maintenance resources. The predictive maintenance decision support system uses a Bayesian network to construct a maintenance decision model and uses an expert knowledge base to reason and analyze various fault symptoms.

8. The method for evaluating the operating status of offshore wind turbines according to claim 7, characterized in that, The predictive maintenance decision support system is designed for the characteristics of offshore wind farms. It constructs a task management system that includes modules for scheduled inspections, technical upgrades, and scheduling. Based on weather windows and constraints such as personnel and spare parts, the system automatically generates the optimal maintenance plan.

9. The method for evaluating the operating status of offshore wind turbines according to claim 8, characterized in that, In step S6, the construction personnel bring spare parts and construction tools to the site and carry out standardized construction in accordance with the standardized construction process. The construction process includes safety inspection before construction, equipment shutdown operation, disassembly and replacement of faulty parts, installation and debugging of new parts, and equipment start-up and operation test after construction.

10. The method for evaluating the operating status of offshore wind turbines according to claim 1, characterized in that, In step S7, after the construction is completed, the staff will effectively record and summarize the operation and maintenance process, including equipment failure, maintenance measures, information on replaced spare parts, construction time and personnel. At the same time, the comprehensive operation and maintenance cost will be determined based on the composition of operation and maintenance costs.

Images