Fan tower fatigue damage real-time monitoring and early warning system and evaluation method

Abstract

The present application relates to the technical field of wind power equipment structure health monitoring, including tower drum full life cycle digital identity and blockchain access module, global multi-source sensing data acquisition and edge preprocessing module, multi-load coupling fatigue damage quantitative calculation module, GRU neural network residual life prediction module, damage state grading linkage early warning module, full-process data chaining and trusted traceability module, multi-objective optimization operation and maintenance decision generation module, operation and maintenance data iteration and full-process closed-loop update module; including the following steps: S1, full-cycle digital identity filing and blockchain node access stage; S2, global multi-source data acquisition and edge preprocessing stage; S3, multi-load coupling fatigue damage quantitative calculation stage; S4, residual life accurate prediction stage; S5, damage state grading early warning and linkage control stage; S6, full-process data chaining and trusted traceability stage; S7, multi-objective optimization operation and maintenance decision generation stage; S8, operation and maintenance data iteration and closed-loop update stage.

Description

Technical Field

[0001] This invention relates to the field of wind power equipment structural health monitoring technology, and in particular to a real-time monitoring and early warning system and assessment method for fatigue damage of wind turbine towers. Background Technology

[0002] As the wind power industry continues to develop towards large capacity, high altitude, and deep sea, the wind turbine tower, as the core load-bearing support structure of the wind turbine, is subjected to a variety of complex alternating loads such as wind load, unit operation excitation, and environmental load throughout the entire service life. The continuous accumulation of fatigue damage can easily lead to safety accidents such as structural cracking, deformation, or even collapse, which directly affects the safe and stable operation and maintenance efficiency of the wind farm.

[0003] Current technologies for monitoring and managing fatigue damage to wind turbine towers are no longer adequate to meet the core requirements for tower structure safety control under the large-scale and deep-sea development of the wind power industry. There is an urgent need for an integrated technical solution to achieve full-process, high-precision, and high-reliability control of tower fatigue damage. Summary of the Invention

[0004] The purpose of this invention is to provide a real-time monitoring and early warning system and evaluation method for fatigue damage of wind turbine towers, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] The wind turbine tower fatigue damage real-time monitoring and early warning system includes a tower full life cycle digital identity and blockchain access module, a full-domain multi-source sensor data acquisition and edge preprocessing module, a multi-load coupled fatigue damage quantitative calculation module, a GRU neural network remaining life prediction module, a damage state graded linkage early warning module, a full-process data on-chain and trusted traceability module, a multi-objective optimization operation and maintenance decision generation module, and a post-operation and maintenance data iteration and full-process closed-loop update module.

[0007] As a further improvement to this technical solution: the tower full lifecycle digital identity and blockchain access module includes a monitoring entity qualification verification unit, a tower digital identity filing unit, a national cryptographic algorithm encryption unit, and a consortium blockchain node permission management unit. The monitoring entity qualification verification unit completes the entry and verification of qualification documents of relevant participating entities. The tower digital identity filing unit completes the entry and structured storage of basic information of a single wind turbine tower in all dimensions. The national cryptographic algorithm encryption unit completes the encryption processing of node identity information and file data. The consortium blockchain node permission management unit completes the identity authentication and node access of each participating entity based on the consortium blockchain architecture. It has a built-in RBAC and ABAC hybrid permission model to complete the fine-grained access permission configuration of different roles and entities. It synchronously connects to the underlying blockchain system to complete the on-chain storage of the initial file information of the tower and generate an immutable initial traceability ledger for the full lifecycle of the tower.

[0008] As a further improvement to this technical solution: the full-domain multi-source sensing data acquisition and edge preprocessing module includes a fiber optic strain sensor, a triaxial accelerometer, an infrared thermometer, a laser displacement sensor, a three-dimensional ultrasonic anemometer, a wall-climbing ultrasonic inspection robot, an industrial-grade edge gateway, a SCADA system interface unit, and a meteorological and oceanographic early warning platform interface unit. The infrared thermometer collects temperature data from the tower weld area, the laser displacement sensor collects tower foundation settlement data, the three-dimensional ultrasonic anemometer collects wind speed, wind direction, and turbulence intensity data, and the SCADA system interface unit synchronously acquires the wind turbine's operating data. The early warning platform docking unit receives early warning signals of extreme weather disasters in real time. Fiber optic strain sensors are arranged along the key sections of the top, middle and bottom of the tower, and collect strain data of each section of the tower at a fixed sampling frequency. Triaxial accelerometers are arranged along the key sections of the tower and collect vibration response data of each section of the tower at a fixed sampling frequency. A wall-climbing ultrasonic inspection robot completes full-area inspection of the weld seams on the inner wall of the tower at a fixed cycle, collects weld defect data, and identifies micro-crack sizes. An industrial-grade edge gateway is deployed in the wind turbine nacelle to complete noise reduction processing of the collected multi-source data, normalize the data, extract core feature values, and transmit encrypted data.

[0009] As a further improvement to this technical solution: the multi-load coupled fatigue damage quantitative calculation module includes a load spectrum decomposition unit, a multi-load coupling correction unit, a stiffness attenuation correction unit, a nonlinear fatigue accumulation calculation unit, and an improved particle swarm optimization solution unit. The load spectrum decomposition unit completes the cyclic counting and spectrum decomposition processing of multi-source load data; the nonlinear fatigue accumulation calculation unit completes the fatigue damage accumulation calculation based on bilinear Miner theory; the multi-load coupling correction unit completes the coupling effect calculation of different types of loads and generates multi-load coupling correction coefficients; the stiffness attenuation correction unit combines the service life of the tower and environmental corrosion parameters to generate stiffness attenuation coefficients; and the improved particle swarm optimization solution unit completes the iterative optimization solution of the damage calculation objective function and outputs the optimal damage accumulation calculation result. The core calculation of this module adopts the following objective function: In the formula, This represents the total cumulative fatigue damage to the tower. For the first Linear damage amount per load cycle, This is a multi-load coupling correction factor. This is the stiffness attenuation coefficient. As the environmental correction factor, this module uses the following rate update formula for iterative solution: In the formula, For the particle renewal rate, The inertia weights are linearly decreasing. and As a learning factor, and A random number between 0 and 1 This represents the historical best position for a single particle. This represents the particle's current position. This represents the global optimal position of the particle swarm.

[0010] As a further improvement to this technical solution: the GRU neural network remaining life prediction module includes a feature input unit, a GRU neural network calculation unit, and a result output unit. The feature input unit completes the input and standardization processing of the core feature parameters required for prediction. The result output unit completes the output of the tower's remaining life prediction value and damage development trend. The GRU neural network calculation unit sets up a two-layer GRU network structure, with a corresponding number of neurons in each layer. The unit has a built-in Dropout layer. The unit uses the AdamW optimizer to complete the model training, with the mean absolute percentage error as the loss function. After inputting the feature parameters of the corresponding dimension, it outputs the remaining life prediction value of the tower for a specified period in the future.

[0011] As a further improvement to this technical solution: the damage status classification and linkage early warning module includes a damage status matching unit, an early warning level determination unit, a SCADA system linkage unit, and an early warning information push unit. The damage status matching unit completes the matching process between the damage assessment results and the early warning determination rules. The SCADA system linkage unit completes the command interaction with the wind turbine SCADA system under emergency conditions. The early warning information push unit completes the targeted push of early warning information of the corresponding level. The early warning level determination unit establishes a four-level early warning determination rule based on the degree of damage, the importance of the damage location, and the damage development rate. The first-level early warning corresponds to a minor damage state, the second-level early warning corresponds to a mild damage state, the third-level early warning corresponds to a moderate damage state, and the fourth-level early warning corresponds to a severe damage state. The unit combines the damage location weight to complete the final determination of the early warning level.

[0012] As a further improvement to this technical solution: the full-process data on-chain and trusted traceability module includes a data on-chain preprocessing unit, a PBFT consensus processing unit, a block data storage unit, and a smart contract execution unit. The data on-chain preprocessing unit completes the format standardization and hash value calculation of the data to be on-chain. The block data storage unit completes the distributed storage and ledger update of the block data. The PBFT consensus processing unit executes the three-stage consensus process of pre-preparation, preparation, and submission, completes the data legality verification between nodes, and achieves consensus confirmation by 2f+1 nodes. The smart contract execution unit has a built-in full-cycle traceability contract for the tower, completes the on-chain mapping of the tower's full-process data, and realizes full-process data traceability query based on the tower's unique digital identity.

[0013] As a further improvement to this technical solution: the multi-objective optimization operation and maintenance decision generation module includes a decision parameter input unit, a multi-objective optimization calculation unit, and an operation and maintenance scheme output unit. The decision parameter input unit completes the input processing of parameters such as early warning level, operating condition, operation and maintenance resources, and maintenance cost. The multi-objective optimization calculation unit completes the optimization solution of the operation and maintenance scheme based on the multi-objective genetic algorithm. The operation and maintenance scheme output unit completes the generation and output of the optimal operation and maintenance scheme.

[0014] As a further improvement to this technical solution: the post-operation and maintenance data iteration and full-process closed-loop update module includes an operation and maintenance data entry unit, a model parameter correction unit, and a damage status update unit. The operation and maintenance data entry unit completes the entry and storage of operation and maintenance records and re-inspection data. The model parameter correction unit completes the parameter correction of the damage calculation model and the life prediction model based on the actual re-inspection data. The damage status update unit completes the recalculation and update of the current damage status and remaining life of the tower, and starts the next round of monitoring and evaluation cycle.

[0015] A method for real-time monitoring and early warning assessment of fatigue damage in wind turbine towers, including the following steps:

[0016] S1. Full-cycle digital identity filing and blockchain node access stage: complete the qualification verification and blockchain identity authentication of each participating entity, establish a unique digital identity for each wind turbine tower, record and store the tower's full-dimensional basic information, configure fine-grained access permissions for each node, complete the blockchain on-chain storage of the tower's initial file information, and generate an immutable initial traceability ledger.

[0017] S2. In the full-domain multi-source data acquisition and edge preprocessing stage, tower structure status data, environmental and load data, unit operation data, and weld defect feature data are collected synchronously at a fixed sampling frequency. Early warning signals of extreme weather disasters are received in real time. Noise reduction, normalization, and core feature value extraction are performed on the collected multi-source data, and encrypted data transmission is completed.

[0018] S3, Quantitative calculation stage of multi-load coupling fatigue damage: Complete the cyclic counting and spectral decomposition processing of multi-source load data, complete the correction calculation of multi-load coupling effect and structural stiffness attenuation, complete the nonlinear fatigue damage accumulation calculation based on bilinear Miner theory, and output the total fatigue damage accumulation of the tower, damage development rate and damage distribution results of key sections through iterative optimization solution.

[0019] S4. In the accurate prediction stage of remaining life, the input and standardization of the core feature parameters for prediction are completed. The calculation is completed through a pre-trained GRU neural network, and the predicted value of the remaining life of the tower and the damage development trend for a specified period of time are output.

[0020] S5, Damage Status Grading Early Warning and Linkage Control Stage: Match the damage assessment results with the early warning judgment rules, determine the corresponding early warning level by combining the damage location weight, complete the linkage control with the wind turbine SCADA system under emergency conditions, and complete the targeted push of early warning information of the corresponding level.

[0021] S6, the full-process data on-chain and trusted traceability stage, completes the format standardization and hash value calculation of the data to be on-chain, completes the blockchain on-chain storage of the data through multi-node consensus verification, completes the update of the distributed ledger, executes smart contracts to complete the on-chain mapping of the entire process data of the tower, and supports full-process data traceability query based on the unique digital identity of the tower.

[0022] S7. Multi-objective optimization operation and maintenance decision generation stage: Complete the input processing of operation and maintenance decision-related parameters, complete the optimization solution of operation and maintenance plan based on multi-objective genetic algorithm, generate and output the optimal operation and maintenance plan, and push it to the corresponding responsible entity for execution;

[0023] S8. Post-operation and maintenance data iteration and closed-loop update stage: Complete the input and storage of operation and maintenance records and re-inspection data; based on the actual re-inspection data, complete the parameter correction of the damage calculation model and life prediction model; complete the recalculation and update of the current damage status and remaining life of the tower; and start the next round of monitoring and evaluation cycle.

[0024] Compared with the prior art, the beneficial effects of the present invention are:

[0025] 1. This invention enables multi-dimensional state perception of key sections across the entire wind turbine tower, fully covering various influencing factors such as tower structural response, service environment, and unit operation. This significantly improves the ability to identify early fatigue damage in the tower. Simultaneously, through nonlinear fatigue damage modeling with multi-load coupling, it accurately characterizes the damage evolution law during the actual service process of the tower, effectively improving the accuracy of tower fatigue damage assessment and remaining life prediction. Combined with a graded early warning mechanism and unit-linked management and control, it can achieve graded and precise management and control of tower damage risks, significantly reducing the probability of false alarms and missed alarms, effectively avoiding structural safety accidents caused by tower fatigue damage, and comprehensively ensuring the safe and stable operation of wind turbine units.

[0026] 2. This invention relies on consortium blockchain technology to achieve integrated management and reliable traceability of tower data throughout its entire lifecycle. It can accurately trace the causes of tower damage, providing reliable data support for structural design optimization, quality control, and responsibility identification. At the same time, it combines multi-objective optimization algorithms to generate adaptable operation and maintenance solutions, enabling the rational allocation of wind farm operation and maintenance resources, effectively reducing wind farm operation and maintenance costs and unplanned downtime losses. The solution is adaptable to different application scenarios such as onshore and offshore, as well as wind turbine towers of different structural types. It has strong scenario adaptability and engineering application value, and can provide reliable structural safety management and control technology support for the large-scale and deep-sea development of the wind power industry.

[0027] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it according to the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Specific embodiments of the present invention are given in detail below with reference to the accompanying drawings. Attached Figure Description

[0028] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0029] Figure 1 This is a schematic diagram of a real-time monitoring and early warning system for fatigue damage of wind turbine towers and an evaluation method. Detailed Implementation

[0030] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are for illustrative purposes only and are not intended to limit the scope of the invention. The invention is described more specifically in the following paragraphs by way of example with reference to the accompanying drawings. It should be noted that the drawings are in a very simplified form and use non-precise proportions, and are only used to facilitate and clarify the illustration of the embodiments of the present invention.

[0031] Please see Figure 1 In this embodiment of the invention, the wind turbine tower fatigue damage real-time monitoring and early warning system includes a tower full life cycle digital identity and blockchain access module, a full-domain multi-source sensor data acquisition and edge preprocessing module, a multi-load coupled fatigue damage quantitative calculation module, a GRU neural network remaining life prediction module, a damage state graded linkage early warning module, a full-process data on-chain and trusted traceability module, a multi-objective optimized operation and maintenance decision generation module, and a post-operation and maintenance data iteration and full-process closed-loop update module.

[0032] Specifically, the tower's full lifecycle digital identity and blockchain access module provides the system with a basic identity and data trust foundation; the full-domain multi-source sensor data acquisition and edge preprocessing module provides the system with raw data support; the multi-load coupled fatigue damage quantitative calculation module is the core computing unit of the system; the GRU neural network remaining life prediction module realizes the prediction of the tower's service life; the damage status graded linkage early warning module realizes the graded control and emergency response of damage risks; the full-process data on-chain and trusted traceability module realizes the immutability and traceability query of full-process data; the multi-objective optimized operation and maintenance decision generation module realizes the intelligent optimization of operation and maintenance plans; and the post-operation and maintenance data iteration and full-process closed-loop update module realizes the continuous iteration and cyclic monitoring of the system.

[0033] The tower full lifecycle digital identity and blockchain access module includes a monitoring entity qualification verification unit, a tower digital identity filing unit, a national cryptographic algorithm encryption unit, and a consortium blockchain node permission management unit. The monitoring entity qualification verification unit completes the entry and verification of qualification documents of relevant participating entities. The tower digital identity filing unit completes the entry and structured storage of basic information of a single wind turbine tower in all dimensions. The national cryptographic algorithm encryption unit completes the encryption processing of node identity information and file data. The consortium blockchain node permission management unit completes the identity authentication and node access of each participating entity based on the consortium blockchain architecture. It has a built-in RBAC and ABAC hybrid permission model to complete the fine-grained access permission configuration of different roles and entities. It synchronously connects to the underlying blockchain system to complete the on-chain storage of the initial file information of the tower and generate an immutable initial traceability ledger for the entire lifecycle of the tower.

[0034] Specifically, the monitoring entity qualification verification unit is used to input, identify and verify the qualification documents of system participants such as wind farm operators, tower manufacturers, installation units, operation and maintenance organizations, testing organizations, and regulatory departments, to ensure that the participants have the corresponding legal qualifications and professional licenses, and to protect the legality of system participants from the source;

[0035] The tower digital identity filing unit is used to generate a unique digital identity for each wind turbine tower, and to input and structure and store basic parameters such as tower structure type, height, wall thickness distribution, and material grade, manufacturing information such as weld number, non-destructive testing report, and material mechanical property parameters, installation records such as foundation type and installation accuracy data, and historical operation and maintenance records, forming a basic digital archive for the entire life cycle of the tower, providing basic data support for subsequent monitoring, evaluation, and traceability;

[0036] The national cryptographic algorithm encryption unit uses the national cryptographic SM2 algorithm to encrypt and sign the node identity information of the participating entities, and uses the national cryptographic SM4 algorithm to perform symmetric encryption processing on the tower archive data, so as to ensure the uniqueness of the node identity and the security of the transmission and storage of the archive data, and prevent the data from being tampered with or illegally stolen.

[0037] The consortium blockchain node permission management unit, based on the consortium blockchain architecture, completes the identity authentication of each participating entity and the access control of blockchain nodes. It has built-in RBAC role-based access control model and ABAC attribute-based access control model, which can complete fine-grained access permission configuration according to the role and attribute of the participating entity, ensuring that different entities can only access data within their corresponding permission scope. This unit also connects to the underlying blockchain system to complete the on-chain notarization of the approved initial archive information of the tower, generating an immutable initial traceability ledger for the entire life cycle of the tower, providing a reliable ledger foundation for subsequent full-process data traceability.

[0038] The full-domain multi-source sensor data acquisition and edge preprocessing module includes a fiber optic strain sensor, a triaxial accelerometer, an infrared thermometer, a laser displacement sensor, a 3D ultrasonic anemometer, a wall-climbing ultrasonic inspection robot, an industrial-grade edge gateway, a SCADA system interface unit, and a meteorological and marine early warning platform interface unit. The infrared thermometer collects temperature data from the tower weld area, the laser displacement sensor collects tower foundation settlement data, the 3D ultrasonic anemometer collects wind speed, direction, and turbulence intensity data, the SCADA system interface unit synchronously acquires wind turbine operating data, and the meteorological and marine early warning platform interface unit... The system receives early warning signals of extreme weather disasters. Fiber optic strain sensors are arranged along the key sections of the top, middle and bottom of the tower to collect strain data of each section of the tower at a fixed sampling frequency. Triaxial accelerometers are arranged along the key sections of the tower to collect vibration response data of each section of the tower at a fixed sampling frequency. A wall-climbing ultrasonic inspection robot completes full-area inspection of the weld seams on the inner wall of the tower at a fixed cycle, collects weld defect data, identifies micro-crack sizes, and an industrial-grade edge gateway is deployed in the wind turbine nacelle to complete noise reduction processing of multi-source data, normalization processing of data, extraction of core feature values, and transmission of encrypted data.

[0039] Specifically, infrared temperature sensors are deployed at key locations on the tower weld seam to collect real-time temperature data of the tower weld seam area at a fixed sampling frequency, providing temperature compensation parameters and auxiliary judgment basis for the damage assessment of the weld seam area;

[0040] Laser displacement sensors are placed at corresponding positions on the tower foundation and the bottom of the tower to collect foundation settlement data of the tower at a fixed sampling frequency, providing basic data for the calculation of additional bending moment of the tower and the correction of damage caused by foundation settlement.

[0041] A three-dimensional ultrasonic anemometer is installed on the top of the wind turbine nacelle. It collects real-time wind speed, wind direction and turbulence intensity data of the wind field at a fixed sampling frequency, providing core environmental data for wind load calculation and load spectrum decomposition.

[0042] The SCADA system interface unit connects with the wind turbine's SCADA system through a standard API interface, and synchronously acquires unit operating data such as wind turbine rotor speed, pitch angle, power generation, and nacelle vibration in real time, providing unit operating status data for the calculation of unit vibration load and damage coupling analysis;

[0043] The meteorological and marine early warning platform docking unit connects to the national meteorological and marine early warning platforms through a standard interface, and receives early warning signals for extreme meteorological disasters such as typhoons, strong gusts, blizzards, and storm surges in real time, providing pre-trigger conditions for the activation of the system's extreme operating mode;

[0044] Fiber optic strain sensors are uniformly arranged along the key stress sections at the top, middle and bottom of the tower. They collect real-time strain data of each section of the tower at a fixed sampling frequency of 50Hz. This data is the core structural response data for calculating tower fatigue damage and can directly reflect the stress state of the tower under alternating loads.

[0045] The triaxial accelerometer is arranged along the key sections of the top, middle and bottom of the tower. It collects vibration response data of the X, Y and Z axes of each section of the tower at a fixed sampling frequency of 20Hz. It is used to analyze the vibration characteristics and natural frequency changes of the tower, and provides core data for determining the stiffness attenuation of the tower structure and decomposing the vibration load.

[0046] The wall-climbing ultrasonic testing robot can autonomously crawl along the inner wall of the tower and complete the full-area inspection of the weld seams on the inner wall of the tower once a week. It collects defect data of the weld seams through ultrasonic non-destructive testing technology, accurately identifies the size and distribution of microcracks in the weld seam area, and provides direct detection data for the actual verification of tower fatigue damage and early defect identification.

[0047] The industrial-grade edge gateway, deployed within the nacelle of each wind turbine, serves as the core carrier for edge computing within the module. Equipped with a multi-core processor and large-capacity memory, it can process multi-source data collected from various front-end sensors in real time. It uses wavelet transform algorithms to denoise the collected data, eliminating invalid data caused by environmental interference. The denoised data is then normalized to unify its dimensions and value range. Key feature values ​​such as strain peak value, vibration dominant frequency, and load spectrum characteristics are extracted, and redundant data is filtered out. Finally, the processed data is encrypted and transmitted to the central server via an encrypted link, reducing the computational burden on the central server and improving system response speed.

[0048] The multi-load coupled fatigue damage quantitative calculation module includes a load spectrum decomposition unit, a multi-load coupling correction unit, a stiffness attenuation correction unit, a nonlinear fatigue accumulation calculation unit, and an improved particle swarm optimization solution unit. The load spectrum decomposition unit performs cyclic counting and spectral decomposition processing of multi-source load data. The nonlinear fatigue accumulation calculation unit performs fatigue damage accumulation calculation based on bilinear Miner theory. The multi-load coupling correction unit calculates the coupling effect of different types of loads and generates multi-load coupling correction coefficients. The stiffness attenuation correction unit combines the tower's service life and environmental corrosion parameters to generate stiffness attenuation coefficients. The improved particle swarm optimization solution unit iteratively optimizes the damage calculation objective function and outputs the optimal damage accumulation calculation result. The core calculation of this module uses the following objective function: In the formula, This represents the total cumulative fatigue damage to the tower. For the first Linear damage amount per load cycle, This is a multi-load coupling correction factor. This is the stiffness attenuation coefficient. As the environmental correction factor, this module uses the following rate update formula for iterative solution: In the formula, For the particle renewal rate, The inertia weights are linearly decreasing. and As a learning factor, and A random number between 0 and 1 This represents the historical best position for a single particle. This represents the particle's current position. This represents the global optimal position of the particle swarm.

[0049] Specifically, the load spectrum decomposition unit, based on the improved rainflow counting method, performs cyclic counting processing on the collected multi-source load data such as wind load, wave load, and unit vibration load to complete the decomposition of the load spectrum and separate the core parameters such as amplitude, mean, and number of cycles of different types of loads, providing basic load data for subsequent fatigue damage calculation.

[0050] The nonlinear fatigue accumulation calculation unit, based on the bilinear Miner theory, completes the cumulative calculation of fatigue damage, replacing the traditional linear Miner accumulation rule. It can fully consider the damage acceleration effect caused by low load after high load, and is more in line with the fatigue damage accumulation law in the actual service process of the tower.

[0051] Multi-load coupling correction element is used to calculate the coupling effect between different types of loads and generate multi-load coupling correction coefficients. It can correct fatigue damage calculation results under multiple loads and solve the error problem caused by single load calculation.

[0052] The stiffness attenuation correction unit, combined with the actual service life of the tower and environmental corrosion parameters, generates a stiffness attenuation coefficient. It can correct the impact of the reduction in structural stiffness caused by weld performance degradation and environmental corrosion after long-term service of the tower on fatigue damage, and improve the accuracy of damage calculation.

[0053] An improved particle swarm optimization solution unit is used to perform iterative optimization of the damage calculation objective function. Through iterative optimization of the particle swarm, it outputs the optimal damage accumulation calculation result that meets the constraints, ensuring the convergence and accuracy of the calculation result.

[0054] The role of the core objective function in this section: This formula is the core calculation formula for the total fatigue damage accumulation of the tower. It can comprehensively consider four core factors: single load cyclic damage, multi-load coupling effect, structural stiffness decay, and environmental impact, to complete the accurate calculation of the total fatigue damage accumulation of the tower. The physical meaning of each parameter in the formula is marked in the formula. Through this formula, the degree of fatigue damage of the tower during service can be directly quantified, providing a core judgment basis for subsequent early warning classification and life prediction.

[0055] The purpose of this velocity update formula: This formula is the core iterative formula of the improved particle swarm optimization algorithm, used to update the velocity and position of particles during the iteration process, through inertia weights. The linear decrease of the equilibrium algorithm, the global optimization ability and local convergence ability, and the setting of learning factors and random numbers ensure that the particles can fully traverse the solution space and quickly converge to the global optimal solution, thus obtaining an accurate calculation result of the total fatigue damage accumulation of the tower.

[0056] The GRU neural network remaining life prediction module includes a feature input unit, a GRU neural network computation unit, and a result output unit. The feature input unit completes the input and standardization of the core feature parameters required for prediction. The result output unit completes the output of the tower's remaining life prediction value and damage development trend. The GRU neural network computation unit sets up a two-layer GRU network structure, with a corresponding number of neurons in each layer. The unit has a built-in Dropout layer. The unit uses the AdamW optimizer to complete the model training, with the mean absolute percentage error as the loss function. After inputting the feature parameters of the corresponding dimension, it outputs the tower's remaining life prediction value for a specified period in the future.

[0057] Specifically, the feature input unit is used to receive and process the 10 core feature parameters required for remaining life prediction, including peak strain of each section of the tower, vibration amplitude, wind speed, turbulence intensity, wave height and salt spray concentration in marine scenarios, service life of the tower, stiffness attenuation coefficient, and historical damage. It performs standardization processing on the input feature parameters, unifies the dimensions and value range of the features, and provides standardized data that meets the input requirements for neural network calculation.

[0058] The GRU neural network computing unit is the core computing carrier of the module. It adopts a two-layer cascaded GRU gated recurrent unit network structure, with 320 neurons in each layer, which can fully fit the nonlinear development law of tower fatigue damage. The unit has a built-in Dropout layer with a deactivation rate of 0.35, which can effectively avoid overfitting during model training. The unit uses the AdamW optimizer to complete the model training, using the mean absolute percentage error as the loss function, which can effectively reduce the prediction error of the model and improve the prediction accuracy. After inputting the standardized feature parameters, the unit can complete the forward propagation calculation and output the predicted value of the remaining life of the tower and the damage development trend data for a specified period of time in the future.

[0059] The result output unit receives the calculation results from the GRU neural network computing unit, completes the formatting and output of the results, and synchronously transmits the prediction results to the damage status grading linkage early warning module and the full-process data on-chain and trusted traceability module, providing data support for subsequent early warning grading and data storage.

[0060] The damage status classification and linkage early warning module includes a damage status matching unit, an early warning level determination unit, a SCADA system linkage unit, and an early warning information push unit. The damage status matching unit completes the matching process between the damage assessment results and the early warning determination rules. The SCADA system linkage unit completes the command interaction with the wind turbine SCADA system in emergency conditions. The early warning information push unit completes the targeted push of early warning information of the corresponding level. The early warning level determination unit establishes a four-level early warning determination rule based on the degree of damage, the importance of the damage location, and the damage development rate. Level 1 early warning corresponds to a minor damage state, Level 2 early warning corresponds to a slight damage state, Level 3 early warning corresponds to a moderate damage state, and Level 4 early warning corresponds to a severe damage state. The unit combines the damage location weight to complete the final determination of the early warning level.

[0061] Specifically, the damage state matching unit is used to receive the damage accumulation, damage development rate and damage location distribution results output by the multi-load coupled fatigue damage quantitative calculation module, and complete the matching process with the early warning judgment rules built into the module to extract the core parameters related to the early warning judgment and provide basic data for the early warning level judgment.

[0062] The early warning level determination unit is the core determination unit of the module. It establishes a four-level early warning determination rule based on three core dimensions: damage degree, importance of damage location, and damage development rate. Level 1 early warning corresponds to a minor damage state in non-critical wall panel areas with a damage degree of less than 8%; Level 2 early warning corresponds to a slight damage state in non-main weld areas with a damage degree greater than or equal to 8% but less than 20%; Level 3 early warning corresponds to a moderate damage state in main weld areas with a damage degree greater than or equal to 20% but less than 35%; and Level 4 early warning corresponds to a severe damage state with a damage degree greater than or equal to 35% or a damage development rate greater than or equal to 3% per month. The unit also incorporates damage location weights to complete the final determination of the early warning level. The weight for main weld locations is 1.0, for ordinary weld locations it is 0.7, and for wall panel locations it is 0.3. This fully considers the impact of different damage locations on the safety of the tower structure, improving the rationality of the early warning determination.

[0063] The SCADA system linkage unit establishes a command interaction link with the wind turbine SCADA system through a standard protocol. In emergency situations where the damage is determined to be Level 4 severe, it can issue an emergency shutdown command to the SCADA system to control the wind turbine to stop running, thereby preventing the tower fatigue damage from further expanding and causing a safety accident. At the same time, it can receive command execution feedback from the SCADA system to ensure the effectiveness of linkage control.

[0064] The early warning information push unit is used to push the corresponding early warning information and preliminary operation and maintenance suggestions to the wind farm operation and maintenance center and the corresponding responsible entities according to the final determined early warning level, ensuring that the early warning information can reach the relevant responsible personnel in a timely manner and provide guidance for subsequent operation and maintenance.

[0065] The end-to-end data on-chain and trusted traceability module includes a data on-chain preprocessing unit, a PBFT consensus processing unit, a block data storage unit, and a smart contract execution unit. The data on-chain preprocessing unit completes the format standardization and hash value calculation of the data to be on-chain. The block data storage unit completes the distributed storage and ledger update of the block data. The PBFT consensus processing unit executes the three-stage consensus process of pre-preparation, preparation, and submission, completes the data legality verification between nodes, and achieves consensus confirmation by 2f+1 nodes. The smart contract execution unit has a built-in full-cycle traceability contract for the tower, which completes the on-chain mapping of the tower's full-process data and realizes full-process data traceability query based on the tower's unique digital identity.

[0066] Specifically, the data on-chain preprocessing unit is used to receive the data to be uploaded to the chain from various modules of the system, including tower monitoring data, damage assessment results, remaining life prediction values, early warning information, unit linkage operation records, operation and maintenance records, etc. It completes the format standardization processing of the data to be uploaded to the chain, and calculates the unique hash value of the data to be uploaded to the chain through a hash algorithm, providing a foundation for subsequent consensus verification and data anti-tampering.

[0067] The PBFT consensus processing unit is the core of the module's consensus mechanism. It adopts a practical Byzantine fault-tolerant consensus mechanism and executes a three-stage consensus process of pre-preparation, preparation, and submission. First, the master node broadcasts the proposal of the data to be uploaded to the chain. Each slave node completes the integrity and signature legality verification of the proposed data. After the verification is successful, a preparation message is sent. When a node collects 2f+1 valid preparation messages, it sends a submission message. Finally, all nodes complete the consensus confirmation of the data, where f is the number of fault-tolerant nodes in the system. This consensus mechanism can support 33% of the nodes in the system to fail, ensuring the consistency and reliability of the data upload process.

[0068] The block data storage unit is used to complete the distributed storage of block data after consensus is passed. The block data includes a block header and a block body. The block header contains the previous block hash, Merkle root, timestamp, node signature, etc., and the block body contains the corresponding full-process data record. The unit also completes the synchronous update of the distributed ledger of each node to ensure that the ledger data of all consensus nodes is consistent and realizes the distributed and tamper-proof storage of data.

[0069] The smart contract execution unit incorporates a tower full-cycle traceability smart contract developed using the Solidity language. The contract establishes a mapping relationship between the tower's unique digital identity and the full-process data records, enabling on-chain mapping of the tower's full-process data. Simultaneously, the contract provides a full-process data traceability query interface based on the tower's unique digital identity. Through this interface, users can query the corresponding tower's historical data from manufacturing, installation, operation, monitoring, damage to repair with a single click, achieving accurate traceability of the cause of damage and determination of responsibility.

[0070] The multi-objective optimization operation and maintenance decision generation module includes a decision parameter input unit, a multi-objective optimization calculation unit, and an operation and maintenance scheme output unit. The decision parameter input unit completes the input processing of parameters such as early warning level, operating condition, operation and maintenance resources, and maintenance cost. The multi-objective optimization calculation unit completes the optimization solution of the operation and maintenance scheme based on the multi-objective genetic algorithm. The operation and maintenance scheme output unit completes the generation and output of the optimal operation and maintenance scheme.

[0071] Specifically, the decision parameter input unit is used to receive and process the core parameters required for operation and maintenance decisions, including the warning level, the current operating status of the wind turbine, the scheduling of operation and maintenance resources, and maintenance cost parameters, so as to provide basic input data for subsequent optimization calculations;

[0072] The multi-objective optimization calculation unit is the core calculation unit of the module. Based on the multi-objective genetic algorithm, it takes the safety of the tower structure, the maximization of power generation revenue, and the minimization of maintenance costs as the three major optimization objectives to complete the optimization solution of the operation and maintenance plan. It can comprehensively consider factors such as damage risk, the difference in revenue between peak and off-peak power generation, the scheduling capability of operation and maintenance resources, maintenance process and spare parts costs, and output the optimal operation and maintenance plan that meets the requirements of multi-objective optimization.

[0073] The operation and maintenance solution output unit is used to receive the optimal operation and maintenance solution after optimization calculation, complete the formatting and output of the solution, and the solution content includes core contents such as maintenance time, maintenance process, personnel configuration, spare parts requirements, and cost budget. At the same time, the solution is pushed to the corresponding responsible parties to provide clear guidance for the execution of operation and maintenance work.

[0074] The post-operation and maintenance data iteration and full-process closed-loop update module includes an operation and maintenance data entry unit, a model parameter correction unit, and a damage status update unit. The operation and maintenance data entry unit completes the entry and storage of operation and maintenance records and re-inspection data. The model parameter correction unit completes the parameter correction of the damage calculation model and the life prediction model based on the actual re-inspection data. The damage status update unit completes the recalculation and update of the current damage status and remaining life of the tower, and starts the next round of monitoring and evaluation cycle.

[0075] Specifically, the operation and maintenance data entry unit is used to receive operation and maintenance records, defect re-inspection data, and component replacement information uploaded by the operation and maintenance organization, complete the data review, entry and structured storage, and simultaneously update the data to the tower's full life cycle digital archive, providing actual verification data for subsequent model correction and damage status updates.

[0076] The model parameter correction unit is used to correct the parameters of the multi-load coupled fatigue damage calculation model and the GRU neural network remaining life prediction model based on the actual damage data of the tower obtained from the operation and maintenance re-inspection, thereby reducing the error between the model calculation results and the actual damage state and continuously improving the calculation accuracy and prediction accuracy of the model.

[0077] The damage status update unit is used to recalculate the current damage status and remaining life of the tower based on the corrected model parameters and the actual data after re-inspection, complete the update of damage status data in the tower digital archive and blockchain ledger, and at the same time issue the start command of the next round of monitoring and evaluation cycle to the system, so as to realize continuous dynamic monitoring and closed-loop management of the tower throughout its entire life cycle.

[0078] A method for real-time monitoring and early warning assessment of fatigue damage in wind turbine towers, including the following steps:

[0079] S1. Full-cycle digital identity filing and blockchain node access stage: complete the qualification verification and blockchain identity authentication of each participating entity, establish a unique digital identity for each wind turbine tower, record and store the tower's full-dimensional basic information, configure fine-grained access permissions for each node, complete the blockchain on-chain storage of the tower's initial file information, and generate an immutable initial traceability ledger.

[0080] S2. In the full-domain multi-source data acquisition and edge preprocessing stage, tower structure status data, environmental and load data, unit operation data, and weld defect feature data are collected synchronously at a fixed sampling frequency. Early warning signals of extreme weather disasters are received in real time. Noise reduction, normalization, and core feature value extraction are performed on the collected multi-source data, and encrypted data transmission is completed.

[0081] S3, Quantitative calculation stage of multi-load coupling fatigue damage: Complete the cyclic counting and spectral decomposition processing of multi-source load data, complete the correction calculation of multi-load coupling effect and structural stiffness attenuation, complete the nonlinear fatigue damage accumulation calculation based on bilinear Miner theory, and output the total fatigue damage accumulation of the tower, damage development rate and damage distribution results of key sections through iterative optimization solution.

[0082] S4. In the accurate prediction stage of remaining life, the input and standardization of the core feature parameters for prediction are completed. The calculation is completed through a pre-trained GRU neural network, and the predicted value of the remaining life of the tower and the damage development trend for a specified period of time are output.

[0083] S5, Damage Status Grading Early Warning and Linkage Control Stage: Match the damage assessment results with the early warning judgment rules, determine the corresponding early warning level by combining the damage location weight, complete the linkage control with the wind turbine SCADA system under emergency conditions, and complete the targeted push of early warning information of the corresponding level.

[0084] S6, the full-process data on-chain and trusted traceability stage, completes the format standardization and hash value calculation of the data to be on-chain, completes the blockchain on-chain storage of the data through multi-node consensus verification, completes the update of the distributed ledger, executes smart contracts to complete the on-chain mapping of the entire process data of the tower, and supports full-process data traceability query based on the unique digital identity of the tower.

[0085] S7. Multi-objective optimization operation and maintenance decision generation stage: Complete the input processing of operation and maintenance decision-related parameters, complete the optimization solution of operation and maintenance plan based on multi-objective genetic algorithm, generate and output the optimal operation and maintenance plan, and push it to the corresponding responsible entity for execution;

[0086] S8. Post-operation and maintenance data iteration and closed-loop update stage: Complete the input and storage of operation and maintenance records and re-inspection data; Based on the actual re-inspection data, complete the parameter correction of the damage calculation model and life prediction model; Complete the recalculation and update of the current damage status and remaining life of the tower; Start the next round of monitoring and evaluation cycle.

[0087] Specifically, step S1 corresponds to the digital identity and blockchain access module of the tower throughout its entire lifecycle. It completes the preliminary preparations for system operation, establishes the tower's unique digital identity and trusted traceability ledger, and provides a foundation for monitoring, evaluation, and traceability throughout the entire process.

[0088] Step S2 corresponds to the system's full-domain multi-source sensor data acquisition and edge preprocessing module, which completes the full-dimensional data acquisition and preprocessing required for tower fatigue damage assessment, providing raw data support for subsequent damage calculation and life prediction.

[0089] Step S3 corresponds to the multi-load coupled fatigue damage quantitative calculation module of the system, which completes the accurate quantitative calculation of tower fatigue damage, outputs the core parameters of the tower's damage state, and provides the core judgment basis for subsequent early warning and life prediction.

[0090] Step S4 corresponds to the GRU neural network remaining life prediction module of the system, which completes the accurate prediction of the remaining service life of the tower, predicts the future development trend of tower damage, and provides forward-looking data support for the formulation of operation and maintenance decisions.

[0091] Step S5 corresponds to the damage status classification and linkage early warning module of the system, which completes the classification and early warning of tower damage risk and linkage control of emergency conditions, so as to realize the timely handling and prevention of tower safety risks.

[0092] Step S6 corresponds to the system's full-process data on-chain and trusted traceability module, which completes the tamper-proof storage and traceability management of the full-process data, ensuring the credibility of the full-process data and providing a credible basis for tracing the cause of damage and determining responsibility.

[0093] Step S7 corresponds to the system's multi-objective optimization operation and maintenance decision generation module, which completes the intelligent optimization and output of operation and maintenance plans, providing scientific and reasonable guidance for the execution of operation and maintenance work;

[0094] Step S8 corresponds to the system's post-operation and maintenance data iteration and full-process closed-loop update module. It completes the post-operation and maintenance data update and model correction, and starts the next round of monitoring and evaluation cycle to achieve continuous dynamic monitoring and closed-loop management of the tower throughout its entire life cycle.

[0095] The method of use and working principle of this invention are as follows:

[0096] Usage: First, complete the qualification verification and blockchain identity authentication of all participating entities to establish a unique digital identity for each wind turbine tower and record all-dimensional basic information. After configuring corresponding access permissions, complete the initial file's blockchain-based notarization and generate an immutable initial traceability ledger. Subsequently, collect multi-dimensional data such as tower structural status, environmental load, unit operation, and weld defects at a fixed frequency. Simultaneously, receive extreme weather warning signals in real time. After noise reduction, normalization, and core feature extraction of the collected multi-source data, encrypt and transmit it. Then, perform cyclic counting and spectral decomposition on the processed load data, and perform correction calculations based on multi-load coupling effects and structural stiffness attenuation. Based on nonlinear fatigue accumulation theory, perform quantitative calculations of tower fatigue damage, outputting the tower damage accumulation, damage development rate, and damage distribution results. Afterward, input standardized core feature parameters, and use a pre-trained neural network to predict the tower's remaining service life, outputting the predicted remaining service life value. The damage development trend is analyzed, and the damage assessment results are matched with the early warning judgment rules. The early warning level is determined by combining the damage location weight. In emergency situations, the wind turbine control system is linked to complete safety management. At the same time, the corresponding early warning information is pushed to the relevant responsible parties. During the process, the core data of the entire process is standardized in format and hash value is calculated. After multi-node consensus verification, the data is stored on the blockchain and the distributed ledger is updated. The traceability query of the entire process data of the tower is realized through smart contracts. Then, the multi-objective optimization solution of the operation and maintenance plan is completed by combining the early warning level, operating conditions, operation and maintenance resources and cost parameters. The optimal operation and maintenance plan is generated and pushed to the corresponding subject for execution. After the operation and maintenance work is completed, the maintenance record and re-inspection data are entered. Based on the actual detection data, the parameters of the damage calculation and life prediction model are corrected. The current damage status and remaining life of the tower are recalculated and updated. The next round of monitoring and assessment cycle is started simultaneously to realize continuous dynamic monitoring and closed-loop management of the entire life cycle of the tower.

[0097] Working Principle: The core logic revolves around the comprehensive perception, accurate assessment, intelligent early warning, and reliable traceability of fatigue damage throughout the entire lifecycle of wind turbine towers. Through a comprehensive perception network composed of multiple types of sensors, it achieves simultaneous acquisition and edge preprocessing of multi-dimensional data, including the structural state of the entire tower cross-section, service environment, and unit operation. This provides comprehensive basic data support for damage assessment. Then, through a fatigue damage calculation model coupled with multiple loads, it comprehensively considers the coupling effect of various load types, structural stiffness attenuation, and environmental influences. Using nonlinear damage accumulation theory and intelligent optimization algorithms, it achieves accurate quantification of tower fatigue damage. Simultaneously, it fits the tower fatigue damage based on a gated recurrent unit neural network. By studying the nonlinear evolution of damage, accurate prediction of the remaining service life of the tower is achieved. Then, a graded early warning mechanism is established by combining the degree of damage, the importance of the damage location, and the damage development rate, so as to realize graded management and control of tower damage risk and unit emergency response. At the same time, relying on the alliance blockchain technology and smart contracts, the immutable storage and reliable traceability of data of the tower from manufacturing, installation, operation to maintenance are realized. Finally, by combining the damage status, operating benefits and maintenance costs, the maintenance plan is optimized in multiple objectives, forming a closed-loop management and control of the entire process from data collection, damage assessment, life prediction, risk warning to maintenance decision-making, and continuously iterating and optimizing the accuracy and reliability of tower structural health management.

[0098] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any way. Those skilled in the art can readily implement the present invention based on the description and drawings above. However, any modifications, alterations, and variations made by those skilled in the art without departing from the scope of the present invention using the disclosed technical content are equivalent embodiments of the present invention. Furthermore, any modifications, alterations, and variations made to the above embodiments based on the essential technology of the present invention are still within the protection scope of the present invention.

Claims

1. A real-time monitoring and early warning system for fatigue damage of wind turbine towers, characterized in that, It includes a tower full lifecycle digital identity and blockchain access module, a full-domain multi-source sensor data acquisition and edge preprocessing module, a multi-load coupled fatigue damage quantitative calculation module, a GRU neural network remaining life prediction module, a damage state graded linkage early warning module, a full-process data on-chain and trusted traceability module, a multi-objective optimization operation and maintenance decision generation module, and a post-operation and maintenance data iteration and full-process closed-loop update module.

2. The real-time monitoring and early warning system for fatigue damage of wind turbine towers according to claim 1, characterized in that, The tower full lifecycle digital identity and blockchain access module includes a monitoring entity qualification verification unit, a tower digital identity filing unit, a national cryptographic algorithm encryption unit, and a consortium blockchain node permission management unit. The monitoring entity qualification verification unit completes the entry and verification of qualification documents of relevant participating entities. The tower digital identity filing unit completes the entry and structured storage of basic information of a single wind turbine tower in all dimensions. The national cryptographic algorithm encryption unit completes the encryption processing of node identity information and file data. The consortium blockchain node permission management unit completes the identity authentication and node access of each participating entity based on the consortium blockchain architecture. It has a built-in RBAC and ABAC hybrid permission model to complete the fine-grained access permission configuration of different roles and entities. It synchronously connects to the underlying blockchain system to complete the on-chain storage of the initial file information of the tower and generate an immutable initial traceability ledger for the entire lifecycle of the tower.

3. The real-time monitoring and early warning system for fatigue damage of wind turbine towers according to claim 1, characterized in that, The comprehensive multi-source sensor data acquisition and edge preprocessing module includes a fiber optic strain sensor, a triaxial accelerometer, an infrared thermometer, a laser displacement sensor, a 3D ultrasonic anemometer, a wall-climbing ultrasonic inspection robot, an industrial-grade edge gateway, a SCADA system interface unit, and a meteorological and marine early warning platform interface unit. The infrared thermometer collects temperature data from the tower weld area, the laser displacement sensor collects tower foundation settlement data, the 3D ultrasonic anemometer collects wind speed, direction, and turbulence intensity data, the SCADA system interface unit synchronously acquires wind turbine operating data, and the meteorological and marine early warning platform interface unit... Real-time reception of early warning signals for extreme weather disasters; fiber optic strain sensors are arranged along key sections at the top, middle and bottom of the tower, collecting strain data of each section of the tower at a fixed sampling frequency; triaxial accelerometers are arranged along key sections of the tower, collecting vibration response data of each section of the tower at a fixed sampling frequency; a wall-climbing ultrasonic inspection robot completes full-area inspection of the weld seams on the inner wall of the tower at a fixed cycle, collecting weld defect data and identifying micro-crack sizes; an industrial-grade edge gateway is deployed inside the wind turbine nacelle to complete noise reduction processing of multi-source data, normalization processing of data, extraction of core feature values, and transmission of encrypted data.

4. The real-time monitoring and early warning system for fatigue damage of wind turbine towers according to claim 1, characterized in that, The multi-load coupled fatigue damage quantitative calculation module includes a load spectrum decomposition unit, a multi-load coupling correction unit, a stiffness attenuation correction unit, a nonlinear fatigue accumulation calculation unit, and an improved particle swarm optimization solution unit. The load spectrum decomposition unit performs cyclic counting and spectral decomposition processing of multi-source load data. The nonlinear fatigue accumulation calculation unit performs fatigue damage accumulation calculation based on bilinear Miner theory. The multi-load coupling correction unit calculates the coupling effect of different types of loads and generates multi-load coupling correction coefficients. The stiffness attenuation correction unit combines the tower's service life and environmental corrosion parameters to generate stiffness attenuation coefficients. The improved particle swarm optimization solution unit iteratively optimizes the damage calculation objective function and outputs the optimal damage accumulation calculation result. The core calculation of this module uses the following objective function: In the formula, This represents the total cumulative fatigue damage to the tower. For the first Linear damage amount per load cycle, This is a correction factor for multi-load coupling. This is the stiffness attenuation coefficient. As the environmental correction factor, this module uses the following rate update formula for iterative solution: In the formula, For the particle renewal rate, The inertia weights are linearly decreasing. and As a learning factor, and A random number between 0 and 1 This represents the historical best position for a single particle. This represents the particle's current position. This represents the global optimal position of the particle swarm.

5. The real-time monitoring and early warning system for fatigue damage of wind turbine towers according to claim 1, characterized in that, The GRU neural network remaining life prediction module includes a feature input unit, a GRU neural network calculation unit, and a result output unit. The feature input unit completes the input and standardization processing of the core feature parameters required for prediction. The result output unit completes the output of the tower's remaining life prediction value and damage development trend. The GRU neural network calculation unit sets up a two-layer GRU network structure, with a corresponding number of neurons in each layer. The unit has a built-in Dropout layer and uses the AdamW optimizer to complete the model training, using the mean absolute percentage error as the loss function. After inputting the feature parameters of the corresponding dimension, it outputs the tower's remaining life prediction value for a specified period in the future.

6. The real-time monitoring and early warning system for fatigue damage of wind turbine towers according to claim 1, characterized in that, The damage status classification and linkage early warning module includes a damage status matching unit, an early warning level determination unit, a SCADA system linkage unit, and an early warning information push unit. The damage status matching unit completes the matching process between the damage assessment results and the early warning determination rules. The SCADA system linkage unit completes the command interaction with the wind turbine SCADA system in emergency conditions. The early warning information push unit completes the targeted push of early warning information of the corresponding level. The early warning level determination unit establishes a four-level early warning determination rule based on the degree of damage, the importance of the damage location, and the damage development rate. Level 1 early warning corresponds to a minor damage state, Level 2 early warning corresponds to a slight damage state, Level 3 early warning corresponds to a moderate damage state, and Level 4 early warning corresponds to a severe damage state. The unit combines the damage location weight to complete the final determination of the early warning level.

7. The real-time monitoring and early warning system for fatigue damage of wind turbine towers according to claim 1, characterized in that, The end-to-end data on-chain and trusted traceability module includes a data on-chain preprocessing unit, a PBFT consensus processing unit, a block data storage unit, and a smart contract execution unit. The data on-chain preprocessing unit completes the format standardization and hash value calculation of the data to be on-chain. The block data storage unit completes the distributed storage and ledger update of the block data. The PBFT consensus processing unit executes the three-stage consensus process of pre-preparation, preparation, and submission, completes the data legality verification between nodes, and achieves consensus confirmation by 2f+1 nodes. The smart contract execution unit has a built-in full-cycle traceability contract for the tower, completes the on-chain mapping of the tower's full-process data, and realizes end-to-end data traceability query based on the tower's unique digital identity.

8. The real-time monitoring and early warning system for fatigue damage of wind turbine towers according to claim 1, characterized in that, The multi-objective optimization operation and maintenance decision generation module includes a decision parameter input unit, a multi-objective optimization calculation unit, and an operation and maintenance scheme output unit. The decision parameter input unit completes the input processing of parameters such as early warning level, operating condition, operation and maintenance resources, and maintenance cost. The multi-objective optimization calculation unit completes the optimization solution of the operation and maintenance scheme based on the multi-objective genetic algorithm. The operation and maintenance scheme output unit completes the generation and output of the optimal operation and maintenance scheme.

9. The real-time monitoring and early warning system for fatigue damage of wind turbine towers according to claim 1, characterized in that, The post-operation and maintenance data iteration and full-process closed-loop update module includes an operation and maintenance data entry unit, a model parameter correction unit, and a damage status update unit. The operation and maintenance data entry unit completes the entry and storage of operation and maintenance records and re-inspection data. The model parameter correction unit completes the parameter correction of the damage calculation model and the life prediction model based on the actual re-inspection data. The damage status update unit completes the recalculation and update of the current damage status and remaining life of the tower, and starts the next round of monitoring and evaluation cycle.

10. A method for real-time monitoring and early warning assessment of fatigue damage in wind turbine towers, applied to the real-time monitoring and early warning system for fatigue damage in wind turbine towers as described in any one of claims 1-9, characterized in that, Includes the following steps: S1. Full-cycle digital identity filing and blockchain node access stage: complete the qualification verification and blockchain identity authentication of each participating entity, establish a unique digital identity for each wind turbine tower, record and store the tower's full-dimensional basic information, configure fine-grained access permissions for each node, complete the blockchain on-chain storage of the tower's initial file information, and generate an immutable initial traceability ledger. S2. In the full-domain multi-source data acquisition and edge preprocessing stage, tower structure status data, environmental and load data, unit operation data, and weld defect feature data are collected synchronously at a fixed sampling frequency. Early warning signals of extreme weather disasters are received in real time. Noise reduction, normalization, and core feature value extraction are performed on the collected multi-source data, and encrypted data transmission is completed. S3, Quantitative calculation stage of multi-load coupling fatigue damage: Complete the cyclic counting and spectral decomposition processing of multi-source load data, complete the correction calculation of multi-load coupling effect and structural stiffness attenuation, complete the nonlinear fatigue damage accumulation calculation based on bilinear Miner theory, and output the total fatigue damage accumulation of the tower, damage development rate and damage distribution results of key sections through iterative optimization solution. S4. In the accurate prediction stage of remaining life, the input and standardization of the core feature parameters for prediction are completed. The calculation is completed through a pre-trained GRU neural network, and the predicted value of the remaining life of the tower and the damage development trend for a specified period of time are output. S5, Damage Status Grading Early Warning and Linkage Control Stage: Match the damage assessment results with the early warning judgment rules, determine the corresponding early warning level by combining the damage location weight, complete the linkage control with the wind turbine SCADA system under emergency conditions, and complete the targeted push of early warning information of the corresponding level. S6, the full-process data on-chain and trusted traceability stage, completes the format standardization and hash value calculation of the data to be on-chain, completes the blockchain on-chain storage of the data through multi-node consensus verification, completes the update of the distributed ledger, executes smart contracts to complete the on-chain mapping of the entire process data of the tower, and supports full-process data traceability query based on the unique digital identity of the tower. S7. Multi-objective optimization operation and maintenance decision generation stage: Complete the input processing of operation and maintenance decision-related parameters, complete the optimization solution of operation and maintenance plan based on multi-objective genetic algorithm, generate and output the optimal operation and maintenance plan, and push it to the corresponding responsible entity for execution; S8. Post-operation and maintenance data iteration and closed-loop update stage: Complete the input and storage of operation and maintenance records and re-inspection data; based on the actual re-inspection data, complete the parameter correction of the damage calculation model and life prediction model; complete the recalculation and update of the current damage status and remaining life of the tower; and start the next round of monitoring and evaluation cycle.

Images