Fan full life cycle archive management system and management method thereof

Abstract

The invention provides a fan full life cycle archive management system and a management method thereof, and the system comprises an Internet of Things data collection module which is installed on a key part of a wind turbine generator set and collects vibration, temperature and stress data; the full-life-cycle digital archive management module is used for recording full-process information of the fan from design to operation and maintenance based on a cloud database and a block chain technology, and ensuring data integrity and traceability; the AI predictive maintenance analysis module analyzes data and predicts faults by using deep learning and machine learning algorithms; the intelligent scheduling and maintenance decision support module is used for generating an optimal maintenance plan in combination with external factors; and the responsibility tracing and process optimization module is used for recording maintenance details to optimize the process and spare part management. The management method comprises the steps of data acquisition, storage, analysis, maintenance plan making, task execution and recording, effect evaluation, process optimization and the like. Through cooperative work of all the modules and implementation of the system method, accurate and efficient full-life-cycle management of the draught fan is achieved.

Description

Technical Field

[0001] The present invention relates to the field of wind farm equipment management, and in particular to a wind turbine full life cycle archive management system and a management method thereof. Background Art

[0002] As the global demand for renewable energy continues to grow, wind power generation, as a clean and renewable form of energy, has become increasingly important. However, as the core component of wind power generation systems, wind turbines face many challenges due to their complex structure and harsh operating environment, especially how to effectively manage and maintain them to ensure their long-term stable operation and maximum power generation efficiency. The maintenance of traditional wind turbines mostly relies on regular inspections and post-fault repairs. This passive maintenance mode is not only inefficient, but also fails to detect potential faults in a timely manner, often leading to unplanned downtime, increasing maintenance costs, and reducing overall power generation efficiency.

[0003] In recent years, with the development of advanced technologies such as the Internet of Things (IoT), big data, artificial intelligence (AI) and blockchain, new possibilities have been provided for the intelligent management and predictive maintenance of wind turbines. IoT technology makes it possible to monitor the operating status of wind turbines in real time; the combination of big data and AI algorithms can mine early signs of equipment failure from massive data and achieve predictive maintenance; and the application of blockchain technology has greatly enhanced the transparency, security and traceability of data, establishing a solid information foundation for the full life cycle management of wind turbines.

[0004] Although the above technologies have significant advantages in theory, in practice, how to systematically integrate these technologies to build a comprehensive management system that can not only monitor in real time and make intelligent predictions, but also optimize decision-making and responsibility tracing is still a technical problem that needs to be solved urgently. Therefore, developing an intelligent wind turbine management and predictive maintenance system that can cover the entire life cycle of wind turbines and integrate data collection, intelligent analysis, decision support, responsibility tracing and process optimization is of great significance to improving the operation and maintenance efficiency of the wind power industry, reducing costs and ensuring power supply stability. Summary of the invention

[0005] The main purpose of the present invention is to provide a wind turbine full life cycle archive management system and a management method thereof, so as to solve the problem that the maintenance of traditional wind turbines mostly relies on regular inspections and post-fault repairs. This passive maintenance mode is not only inefficient, but also unable to detect potential faults in a timely manner, often leading to unplanned downtime, increasing maintenance costs, and reducing overall power generation efficiency.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: a wind turbine full life cycle archive management system, the management system includes: an Internet of Things data acquisition module, which is configured in the key components of the wind turbine set and is used to collect vibration, temperature, and stress operating status data in real time; The full life cycle digital archive management module is a database system built on the cloud, which uses blockchain technology to record all information from design, manufacturing, installation to operation and maintenance to ensure data integrity and traceability; AI predictive maintenance analysis module, based on deep learning and machine learning algorithms, analyzes received data, identifies equipment status change trends and abnormal patterns, and predicts possible future failures; Intelligent scheduling and maintenance decision support module, which automatically generates the optimal maintenance plan based on external factors such as geographic location, weather conditions and grid demand, and provides it to the maintenance team; The responsibility tracing and process optimization module records all maintenance activity details and responsible persons, and uses data analysis to optimize maintenance processes and spare parts management strategies.

[0007] In a preferred embodiment, the acquisition module also includes at least one type of sensor, selected from vibration sensors, temperature sensors, and stress monitors, for accurately capturing changes in physical parameters during the operation of the wind turbine.

[0008] In the preferred solution, the full life cycle digital archive management module further includes a distributed ledger system based on blockchain to ensure that the creation, update and access of each device archive record are transparent and cannot be tampered with, and supports efficient query and data sharing.

[0009] In the preferred solution, the AI ​​predictive maintenance analysis module uses at least one machine learning model, including a support vector machine, a random forest or a neural network, to train the collected multi-dimensional data to achieve early identification and warning of faults.

[0010] In the preferred solution, the scheduling and maintenance decision support module integrates intelligent algorithms, which can automatically adjust the maintenance plan, taking into account maintenance costs, downtime impacts and spare parts supply conditions, in order to minimize operation and maintenance costs and maximize power generation efficiency.

[0011] In the preferred solution, the responsibility tracing and process optimization module is equipped with a data mining function to analyze the relationship between maintenance activities and failure occurrence, providing a basis for continuously improving maintenance strategies and improving operation and maintenance efficiency.

[0012] The preferred embodiment further includes a user-friendly mobile application that enables field engineers to receive maintenance task notifications, view detailed fault analysis reports, historical maintenance records, and operation guides.

[0013] In a preferred embodiment, the computer program product includes computer program code stored on a non-transitory computer-readable storage medium, and when executed by a computer, causes the computer to perform all the functions of the above-mentioned system.

[0014] In a preferred embodiment, the method comprises: S1. Data collection steps: Install vibration sensors and stress monitors at the root of the wind turbine blades and key nodes of the wind turbine, install temperature sensors and vibration sensors at appropriate positions of the stator winding, bearings and housing of the generator, and install temperature sensors, vibration sensors and stress monitors at different positions of the gearbox, near each bearing seat and key stress-bearing structure. Each sensor collects the vibration, temperature and stress physical parameter change data of the wind turbine during operation according to the set sampling frequency. The IoT data collection module cleans and converts the collected raw data, and transmits it to the full life cycle digital archive management module in real time through wireless or wired transmission. S2. Data storage and management steps: After receiving the data, the full life cycle digital archive management module stores it in the cloud database according to the data source and type, and enters the fan design drawings, manufacturing process parameters, installation and commissioning records, and non-real-time operation data; Use a blockchain-based distributed ledger system to create a unique hash value identifier for each device archive record, record the timestamp of the data operation, operator information, and content before and after the modification, to ensure that data creation, update, and access are transparent, tamper-proof, and traceable, and only authorized users or nodes can operate the data; S3, data analysis and fault prediction step: The AI ​​predictive maintenance analysis module extracts the wind turbine operation data of a specific time period from the full life cycle digital archive management module and performs standardized processing on it; Select a support vector machine, random forest or neural network machine learning model, use historical normal and fault data to train the model, adjust model parameters, and enable the model to learn the data characteristics of normal and faulty status of the equipment; input the real-time collected and pre-processed data into the trained model, calculate the degree of deviation between the current status of the equipment and the normal status, and predict the type and time range of possible faults in the future; S4. Maintenance plan formulation step: The intelligent dispatching and maintenance decision support module receives the fault prediction results and interacts with the geographic information system, meteorological monitoring system, and power grid dispatching system to obtain the wind turbine's geographical location, weather conditions, and power grid demand information, combined with the company's internal maintenance resource situation; With the goal of minimizing operation and maintenance costs and maximizing power generation efficiency, the system meets the technical requirements of maintenance tasks and safety regulations, and uses integrated intelligent algorithms to automatically generate the optimal maintenance plan, including maintenance task arrangements, work processes, and power generation plan adjustment suggestions; S5. Maintenance task execution and recording steps: The intelligent scheduling and maintenance decision support module pushes the maintenance plan to the field engineer through the mobile application. The field engineer receives the notification and views the fault analysis report, historical maintenance records and operation guide. The field engineer brings tools and spare parts to the wind turbine site as required, performs maintenance operations according to the operation process, and records key operation steps and data. The responsibility tracing and process optimization module records the maintenance personnel information, operation time, content maintenance activity details and stores them in the full life cycle digital archive management module in association with the maintenance task. After completing the task, the field engineer feedbacks the maintenance results through the mobile application. S6. Maintenance effect evaluation and process optimization steps: Determine evaluation indicators based on maintenance task objectives, such as operational stability of equipment after repair, recovery of power generation efficiency, and whether maintenance costs meet expectations; collect operational data before and after maintenance, maintenance activity record data, and cost data from the full life cycle digital archive management module, and use data analysis tools to analyze and mine the relationship between maintenance activities and failures; optimize maintenance processes based on evaluation results and analysis conclusions, such as modifying maintenance operation specifications, adjusting maintenance cycles, and optimizing spare parts management strategies, and update the optimized maintenance processes and strategies to the full life cycle digital archive management module.

[0015] In a preferred embodiment, the method comprises: The specific algorithm method of steps S1-S6 is: S11, data collection step: the collected vibration data sequence is , the temperature data series is , the stress data series is ,in is the number of sampling points; First, calculate the dynamic mean of the data. Taking vibration data as an example, the dynamic mean formula is: ; in It is The dynamic mean of the sampling points, is the dynamic mean of the previous sampling point, is the smoothing coefficient ( ), which is used to reflect the impact of current data values ​​and consider the trend of historical data during data collection, avoiding sudden changes in the mean value due to individual abnormal data points, so that the mean can better follow the dynamic trend of data changes; Then calculate the dynamic standard deviation of the data, the formula is: ; in It is The dynamic standard deviation of the sampling points, It is the dynamic standard deviation of the previous sampling point. The dynamic standard deviation can be used to measure the degree of dispersion of data and help determine the stability and abnormal fluctuation of data. S21, data storage and management steps: Data storage algorithm: set data block Including collection time , Data Source , Data Type And specific data values ; Using hash functions , where || represents a connection operation, It is a custom hash function, such as a polynomial-based hash function: ; in , are randomly generated coefficients ( ), is the degree of the polynomial, is a large prime number; the role of this hash function is to generate a unique identifier for the data block to ensure the integrity and traceability of the data. Any modification to the data will cause a huge change in the hash value, which can be easily detected; S32, data analysis and fault prediction steps: Data analysis and fault prediction algorithm: a hybrid algorithm based on multi-scale entropy MSE and support vector machine SVM is adopted; First, calculate the multi-scale entropy, assuming that the original data sequence , and coarse-grained them to obtain different scales sequence under ,in ; Multiscale entropy The calculation steps are as follows: For each scale sequence under , calculate its sample entropy ,in is the embedding dimension, is the similarity tolerance; the sample entropy formula is: ; in is satisfied of dimensional vector pair The ratio of the number of pairs to the total number of vector pairs ( ); Multi-scale entropy reflects the complexity and regularity of data at different time scales. The entropy values ​​of wind turbine data in normal operation have a certain characteristic range at different scales, and the entropy value will change in the fault state; Then the multi-scale entropy at different scales is combined into a feature vector , and input it into the support vector machine for fault classification and prediction; the support vector machine constructs the optimal classification hyperplane ,in is the normal vector, is a function that maps input data to a high-dimensional space, is the bias term, using the kernel function Convert the nonlinear problem in low-dimensional space into a linear problem in high-dimensional space, so as to classify the fault type and predict the time range of fault occurrence; S41. Maintenance plan formulation steps: Maintenance plan formulation algorithm: Set maintenance cost Including maintenance personnel labor costs , Spare parts replacement cost , Equipment downtime loss cost , power generation efficiency loss ; Maintenance personnel labor costs ,in is the number of maintenance personnel, It is The unit time wage of a maintenance worker, is the estimated working time; Spare parts replacement cost ,in Is the spare parts type, It is Unit price of spare parts, is the required quantity; Equipment downtime loss cost ,in is the rated power of the fan, is the estimated downtime, is the economic value of a unit of electricity; Power generation efficiency loss ,in is the time when the failure occurred, is to restore normal operating time, is the function of actual generated power over time; Constructing a multi-objective optimization function ,in and is the weight coefficient ( ), which indicates the importance of maintenance cost and power generation efficiency loss in the comprehensive goal; the multi-objective optimization problem is solved by intelligent optimization algorithms such as genetic algorithms to obtain the optimal maintenance plan parameters, such as maintenance personnel arrangement, spare parts allocation, and maintenance time; S51, maintenance task execution and recording steps: Maintenance task execution and recording algorithm: During the maintenance task execution process, record each operation step Execution time , Operation results And related data ; Constructing maintenance task execution vectors , store it in the full life cycle digital archive management module for subsequent maintenance effect evaluation and process optimization; S61. Maintenance effect evaluation and process optimization steps: Maintenance effect evaluation algorithm: Operation stability index after equipment repair ,in is the number of sampling points in the evaluation period, is the actual operation data. It is the reference data when the equipment is operating normally; the smaller the index is, the more stable the equipment is. Maintenance cost compliance index ,in is the actual maintenance cost incurred, is the planned maintenance cost, The closer , indicating that the maintenance cost control is better; Power generation efficiency recovery index ,in is the average power generation of the wind turbine after maintenance, is the average power generation before maintenance, Greater than It indicates that the power generation efficiency has been improved; Through comprehensive analysis of these indicators, the association rule Apriori mining algorithm is used to mine the relationship between maintenance activities and failure occurrence; Among them, the association rules between certain specific maintenance operation combinations and the probability of equipment failure occurring again within a certain period of time are found: ,in and It is a maintenance operation. is the probability of failure; based on these association rules and evaluation results, the maintenance process is optimized, such as adjusting the order of maintenance operations, increasing or decreasing the frequency of certain maintenance operations.

[0016] The present invention provides a wind turbine full life cycle archive management system and a management method thereof, and the beneficial effects of the present invention are: 1) Enhanced fault prediction accuracy: By integrating advanced IoT technology and AI algorithms, the system can monitor the status of wind turbines in real time, accurately identify potential faults, significantly improve the accuracy of predictive maintenance, reduce unexpected downtime, and ensure stable operation of wind farms.

[0017] 2) Improve operation and maintenance efficiency and cost-effectiveness: The intelligent scheduling and decision support module can dynamically adjust the maintenance plan according to actual conditions and optimize resource allocation. At the same time, it combines spare parts management and cost control strategies to minimize operation and maintenance costs and improve overall economic benefits.

[0018] 3) Ensure data security and transparency: The full life cycle digital archive management system built using blockchain technology ensures the immutability and high transparency of data, enhances the credibility of equipment information, facilitates tracking and auditing, and is conducive to responsibility definition and compliance management. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The present invention will be further described below in conjunction with the accompanying drawings and embodiments: Figure 1 This is a flow chart of the management method of the wind turbine full life cycle archive management system of the present invention; Figure 2 It is a fault handling flow chart of the present invention; Figure 3 It is the maintenance flow chart of the present invention. DETAILED DESCRIPTION

[0020] Example 1 like Figures 1 to 3 As shown, a wind turbine full life cycle archive management system, the management system includes: an Internet of Things data acquisition module, configured in the key components of the wind turbine set, for real-time collection of vibration, temperature, stress operating status data; The full life cycle digital archive management module is a database system built on the cloud, which uses blockchain technology to record all information from design, manufacturing, installation to operation and maintenance to ensure data integrity and traceability; AI predictive maintenance analysis module, based on deep learning and machine learning algorithms, analyzes received data, identifies equipment status change trends and abnormal patterns, and predicts possible future failures; Intelligent scheduling and maintenance decision support module, which automatically generates the optimal maintenance plan based on external factors such as geographic location, weather conditions and grid demand, and provides it to the maintenance team; The responsibility tracing and process optimization module records all maintenance activity details and responsible persons, and uses data analysis to optimize maintenance processes and spare parts management strategies.

[0021] In a preferred embodiment, the acquisition module also includes at least one type of sensor, selected from vibration sensors, temperature sensors, and stress monitors, for accurately capturing changes in physical parameters during the operation of the wind turbine.

[0022] In the preferred solution, the full life cycle digital archive management module further includes a distributed ledger system based on blockchain to ensure that the creation, update and access of each device archive record are transparent and cannot be tampered with, and supports efficient query and data sharing.

[0023] In the preferred solution, the AI ​​predictive maintenance analysis module uses at least one machine learning model, including a support vector machine, a random forest or a neural network, to train the collected multi-dimensional data to achieve early identification and warning of faults.

[0024] In the preferred solution, the scheduling and maintenance decision support module integrates intelligent algorithms, which can automatically adjust the maintenance plan, taking into account maintenance costs, downtime impacts and spare parts supply conditions, in order to minimize operation and maintenance costs and maximize power generation efficiency.

[0025] In the preferred solution, the responsibility tracing and process optimization module is equipped with a data mining function to analyze the relationship between maintenance activities and failure occurrence, providing a basis for continuously improving maintenance strategies and improving operation and maintenance efficiency.

[0026] The preferred embodiment further includes a user-friendly mobile application that enables field engineers to receive maintenance task notifications, view detailed fault analysis reports, historical maintenance records, and operation guides.

[0027] In a preferred embodiment, the computer program product includes computer program code stored on a non-transitory computer-readable storage medium, and when executed by a computer, causes the computer to perform all the functions of the above-mentioned system.

[0028] Example 2 Further illustrate with reference to Example 1, Figure 1-3 As shown, the method includes: S1. Data collection steps: Install vibration sensors and stress monitors at the root of the wind turbine blades and key nodes of the wind turbine, install temperature sensors and vibration sensors at appropriate positions of the stator winding, bearings and housing of the generator, and install temperature sensors, vibration sensors and stress monitors at different positions of the gearbox, near each bearing seat and key stress-bearing structure. Each sensor collects the vibration, temperature and stress physical parameter change data of the wind turbine during operation according to the set sampling frequency. The IoT data collection module cleans and converts the collected raw data, and transmits it to the full life cycle digital archive management module in real time through wireless or wired transmission. S2. Data storage and management steps: After receiving the data, the full life cycle digital archive management module stores it in the cloud database according to the data source and type, and enters the fan design drawings, manufacturing process parameters, installation and commissioning records, and non-real-time operation data; Use a blockchain-based distributed ledger system to create a unique hash value identifier for each device archive record, record the timestamp of the data operation, operator information, and content before and after the modification, to ensure that data creation, update, and access are transparent, tamper-proof, and traceable, and only authorized users or nodes can operate the data; S3, data analysis and fault prediction step: The AI ​​predictive maintenance analysis module extracts the wind turbine operation data of a specific time period from the full life cycle digital archive management module and performs standardized processing on it; Select a support vector machine, random forest or neural network machine learning model, use historical normal and fault data to train the model, adjust model parameters, and enable the model to learn the data characteristics of normal and faulty status of the equipment; input the real-time collected and pre-processed data into the trained model, calculate the degree of deviation between the current status of the equipment and the normal status, and predict the type and time range of possible faults in the future; S4. Maintenance plan formulation step: The intelligent dispatching and maintenance decision support module receives the fault prediction results and interacts with the geographic information system, meteorological monitoring system, and power grid dispatching system to obtain the wind turbine's geographical location, weather conditions, and power grid demand information, combined with the company's internal maintenance resource situation; With the goal of minimizing operation and maintenance costs and maximizing power generation efficiency, the system meets the technical requirements of maintenance tasks and safety regulations, and uses integrated intelligent algorithms to automatically generate the optimal maintenance plan, including maintenance task arrangements, work processes, and power generation plan adjustment suggestions; S5. Maintenance task execution and recording steps: The intelligent scheduling and maintenance decision support module pushes the maintenance plan to the field engineer through the mobile application. The field engineer receives the notification and views the fault analysis report, historical maintenance records and operation guide. The field engineer brings tools and spare parts to the wind turbine site as required, performs maintenance operations according to the operation process, and records key operation steps and data. The responsibility tracing and process optimization module records the maintenance personnel information, operation time, content maintenance activity details and stores them in the full life cycle digital archive management module in association with the maintenance task. After completing the task, the field engineer feedbacks the maintenance results through the mobile application. S6. Maintenance effect evaluation and process optimization steps: Determine evaluation indicators based on maintenance task objectives, such as operational stability of equipment after repair, recovery of power generation efficiency, and whether maintenance costs meet expectations; collect operational data before and after maintenance, maintenance activity record data, and cost data from the full life cycle digital archive management module, and use data analysis tools to analyze and mine the relationship between maintenance activities and failures; optimize maintenance processes based on evaluation results and analysis conclusions, such as modifying maintenance operation specifications, adjusting maintenance cycles, and optimizing spare parts management strategies, and update the optimized maintenance processes and strategies to the full life cycle digital archive management module.

[0029] In a preferred embodiment, the method comprises: The specific algorithm method of steps S1-S6 is: S11, data collection step: the collected vibration data sequence is , the temperature data series is , the stress data series is ,in is the number of sampling points; First, calculate the dynamic mean of the data. Taking vibration data as an example, the dynamic mean formula is: ; in It is The dynamic mean of the sampling points, is the dynamic mean of the previous sampling point, is the smoothing coefficient ( ), which is used to reflect the impact of current data values ​​and consider the trend of historical data during data collection, avoiding sudden changes in the mean value due to individual abnormal data points, so that the mean can better follow the dynamic trend of data changes; Then calculate the dynamic standard deviation of the data, the formula is: ; in It is The dynamic standard deviation of the sampling points, It is the dynamic standard deviation of the previous sampling point. The dynamic standard deviation can be used to measure the degree of dispersion of data and help determine the stability and abnormal fluctuation of data. S21, data storage and management steps: Data storage algorithm: set data block Including collection time , Data Source , Data Type And specific data values ; Using hash functions , where || represents a connection operation, It is a custom hash function, such as a polynomial-based hash function: ; in , are randomly generated coefficients ( ), is the degree of the polynomial, is a large prime number; the role of this hash function is to generate a unique identifier for the data block to ensure the integrity and traceability of the data. Any modification to the data will cause a huge change in the hash value, which can be easily detected; S32, data analysis and fault prediction steps: Data analysis and fault prediction algorithm: a hybrid algorithm based on multi-scale entropy MSE and support vector machine SVM is adopted; First, calculate the multi-scale entropy, assuming that the original data sequence , and coarse-grained them to obtain different scales sequence under ,in ; Multiscale entropy The calculation steps are as follows: For each scale sequence under , calculate its sample entropy ,in is the embedding dimension, is the similarity tolerance; the sample entropy formula is: ; in is satisfied of dimensional vector pair The ratio of the number of pairs to the total number of vector pairs ( ); Multi-scale entropy reflects the complexity and regularity of data at different time scales. The entropy values ​​of wind turbine data in normal operation have a certain characteristic range at different scales, and the entropy value will change in the fault state; Then the multi-scale entropy at different scales is combined into a feature vector , and input it into the support vector machine for fault classification and prediction; the support vector machine constructs the optimal classification hyperplane ,in is the normal vector, is a function that maps input data to a high-dimensional space, is the bias term, using the kernel function Convert the nonlinear problem in low-dimensional space into a linear problem in high-dimensional space, so as to classify the fault type and predict the time range of fault occurrence; S41. Maintenance plan formulation steps: Maintenance plan formulation algorithm: Set maintenance cost Including maintenance personnel labor costs , Spare parts replacement cost , Equipment downtime loss cost , power generation efficiency loss ; Maintenance personnel labor costs ,in is the number of maintenance personnel, It is The unit time wage of a maintenance worker, is the estimated working time; Spare parts replacement cost ,in Is the spare parts type, It is Unit price of spare parts, is the required quantity; Equipment downtime loss cost ,in is the rated power of the fan, is the estimated downtime, is the economic value of a unit of electricity; Power generation efficiency loss ,in is the time when the failure occurred, is to restore normal operating time, is the function of actual generated power over time; Constructing a multi-objective optimization function ,in and is the weight coefficient ( ), which indicates the importance of maintenance cost and power generation efficiency loss in the comprehensive goal; the multi-objective optimization problem is solved by intelligent optimization algorithms such as genetic algorithms to obtain the optimal maintenance plan parameters, such as maintenance personnel arrangement, spare parts allocation, and maintenance time; S51, maintenance task execution and recording steps: Maintenance task execution and recording algorithm: During the maintenance task execution process, record each operation step Execution time , Operation results And related data ; Constructing maintenance task execution vectors , store it in the full life cycle digital archive management module for subsequent maintenance effect evaluation and process optimization; S61. Maintenance effect evaluation and process optimization steps: Maintenance effect evaluation algorithm: Operation stability index after equipment repair ,in is the number of sampling points in the evaluation period, is the actual operation data. It is the reference data when the equipment is operating normally; the smaller the index is, the more stable the equipment is. Maintenance cost compliance index ,in is the actual maintenance cost incurred, is the planned maintenance cost, The closer , indicating that the maintenance cost control is better; Power generation efficiency recovery index ,in is the average power generation of the wind turbine after maintenance, is the average power generation before maintenance, Greater than It indicates that the power generation efficiency has been improved; Through comprehensive analysis of these indicators, the association rule Apriori mining algorithm is used to mine the relationship between maintenance activities and failure occurrence; Among them, the association rules between certain specific maintenance operation combinations and the probability of equipment failure occurring again within a certain period of time are found: ,in and It is a maintenance operation. is the probability of failure; based on these association rules and evaluation results, the maintenance process is optimized, such as adjusting the order of maintenance operations, increasing or decreasing the frequency of certain maintenance operations.

[0030] Example 3 Further illustrate with reference to Example 2, Figure 1-3 As shown, Data collection step (S1): Install vibration sensors and stress monitors at the root of the wind turbine blades and key node positions of the wind turbine, install temperature sensors and vibration sensors at appropriate positions of the stator winding, bearings and housing of the generator, and install temperature sensors, vibration sensors and stress monitors at different positions of the gearbox, near each bearing seat and key force-bearing structure. Each sensor collects the vibration, temperature and stress physical parameter change data of the wind turbine during operation according to the set sampling frequency (for example, collecting data every 5 milliseconds).

[0031] Taking vibration data as an example, the collected vibration data sequence is First, calculate the dynamic mean of the data. The dynamic mean formula is: ,in (This value can be adjusted according to actual data fluctuations). ,when hour, , and so on to calculate the dynamic mean of each sampling point. Then calculate the dynamic standard deviation, the formula is , assuming the initial standard deviation The dynamic standard deviation of each sampling point is calculated based on the previous data points. The IoT data collection module cleans the collected raw data (removes obviously abnormal spike data, such as those exceeding the mean value). After the data is converted into the correct format, it is transmitted to the full life cycle digital archive management module in real time via 4G wireless transmission.

[0032] (ii) Data storage and management steps (S2): After receiving the data, the full life cycle digital archive management module stores the data in the cloud database according to the data source (wind turbine blades, generators, gearboxes, etc.) and type (vibration, temperature, stress), and enters non-real-time operating data such as the wind turbine's design drawings, manufacturing process parameters, and installation and commissioning records.

[0033] Set data block Including collection time , Data Source , Data Type And specific data values Using a hash function ,in (here , , , , , , these coefficients can be adjusted according to data security and uniqueness requirements). For example, for a data block ,in (Indicates data collected at 10:30:05 on December 15, 2024), , , ,but , calculate the hash value , create a unique identifier for the data, ensure that data creation, update and access are transparent, tamper-proof and traceable, and only authorized users or nodes can operate the data.

[0034] (III) Data analysis and fault prediction step (S3): The AI ​​predictive maintenance analysis module extracts the wind turbine operation data of the past week from the full life cycle digital archive management module and performs standardization on it (normalizing the data to the $[0,1]$ interval).

[0035] A hybrid algorithm based on multi-scale entropy (MSE) and support vector machine (SVM) is used. First, the multi-scale entropy is calculated. Suppose the original data sequence , and coarse-grained them to obtain different scales sequence under ,in For each scale sequence under , calculate its sample entropy ,in , (These parameters can be adjusted according to data characteristics and fault detection accuracy). The sample entropy formula is: For example, for scale sequence ,calculate and Then we get the sample entropy Then the multi-scale entropy at different scales is combined into a feature vector (Here take scale, the scale value can be selected according to the actual situation). The input is sent to the support vector machine for fault classification and prediction. The support vector machine constructs the optimal classification hyperplane , using radial basis function As the kernel function, (Adjustable), using kernel functions to transform nonlinear problems in low-dimensional space into linear problems in high-dimensional space, thereby achieving classification of fault types (such as fan blade cracks, generator overheating, gearbox gear wear, etc.) and predicting the time range of fault occurrence (such as failure is expected to occur within the next 48 hours).

[0036] (IV) Maintenance plan formulation step (S4): The intelligent dispatching and maintenance decision support module receives the fault prediction results and interacts with the geographic information system, meteorological monitoring system, and power grid dispatching system to obtain the wind turbine's geographic location (longitude ,latitude ), weather conditions (there will be light rain in the next three days, with wind speeds between 3 and 5 m / s), grid demand information (currently in a low electricity consumption period, demand for the wind turbine's power generation is relatively low), combined with the company's internal maintenance resources (there are 3 maintenance personnel with high, medium and low skill levels, and the spare parts inventory includes 2 sets of wind turbine blade spare parts, 3 sets of generator bearing spare parts, etc.).

[0037] Maintenance cost Including maintenance personnel labor costs , Spare parts replacement cost , Equipment downtime loss cost , power generation efficiency loss . Maintenance personnel labor costs , assuming that the hourly wage of a senior maintenance worker is Yuan, Intermediate Maintenance Personnel Yuan, low-level maintenance personnel The estimated working hours are Hour, Hour, Hours, then Yuan. Spare parts replacement cost .

[0038] Replacement price of a fan blade spare part Yuan, need to replace 1, generator bearing spare parts price Yuan, need to replace 2, then RMB. Cost of equipment downtime loss , fan rated power , estimated downtime Hours, economic value per unit of electricity Yuan / kWh, then Power generation efficiency loss , assuming the average power generation before maintenance , estimated average power generation after maintenance , maintenance time interval For 10 hours, (Here the negative sign indicates an improvement in power generation efficiency). Constructing a multi-objective optimization function .

[0039] set up , ,but Different maintenance plans (such as different maintenance personnel combinations, spare parts usage plans, maintenance schedules, etc.) are optimized through genetic algorithms to minimize The optimal maintenance plan was obtained by arranging senior and mid-level maintenance personnel to start maintenance work at 9 a.m. the next day. They would first replace the spare parts of the wind turbine blades and then check the condition of the generator bearings. The maintenance time was expected to be 8 hours. Corresponding suggestions for adjusting the power generation plan were also formulated (such as appropriately increasing the power generation of surrounding wind turbines during the maintenance period to maintain the stability of the power grid).

[0040] (V) Maintenance task execution and recording steps (S5): The intelligent scheduling and maintenance decision support module pushes the maintenance plan to the field engineer through the mobile application. The field engineer receives the notification and views the fault analysis report (detailed explanation of the fault prediction based on the abnormal multi-scale entropy of the fan blade vibration, with a similarity of 80% compared with the historical fault data), historical maintenance records (the last time the fan blade spare parts were replaced was a year ago, and it ran stably for 300 days after the maintenance) and operation guide (detailed fan blade spare parts replacement steps and safety precautions). The field engineer brings the tools and spare parts to the fan site as required, carries out maintenance operations according to the operation process, and records each operation step. Execution time , Operation results And related data For example, the operation step "Remove the old blade" takes Minutes, operation results , related data . Construct maintenance task execution vector , the responsibility tracing and process optimization module stores it in the full life cycle digital archive management module. After completing the task, the on-site engineer will feedback the maintenance results through the mobile application (the wind blades are replaced successfully, the equipment operates normally, and the vibration, temperature, and stress data return to the normal range).

[0041] (VI) Maintenance effect evaluation and process optimization step (S6): Determine the evaluation indicators according to the maintenance task objectives and the operating stability indicators of the equipment after repair , assuming that within one week after the maintenance vibration data points, The average value of vibration data during normal operation is $0.03$, and the calculation result is (The smaller the value, the more stable the operation.) Maintenance cost compliance index , the actual maintenance cost Planned maintenance cost Yuan, then (Close to 1 indicates good cost control). Power generation efficiency recovery index , , ,but (greater than 1 indicates improved power generation efficiency).

[0042] The Apriori algorithm is used to mine the association between maintenance activities and failures. For example, the association rule between "unbalanced skill levels of maintenance personnel and insufficient spare parts inventory" and "equipment failures again within three months" is found with a confidence level of 60%. Based on these evaluation results and association rules, the maintenance process is optimized, such as adjusting the maintenance personnel training plan, improving the level of low-skilled personnel, optimizing the spare parts inventory management strategy, increasing the minimum inventory quantity of commonly used spare parts, etc. The optimized maintenance process and strategy are updated to the full life cycle digital archive management module to provide a reference for subsequent maintenance management work, and also provide experience for the maintenance management of other wind turbines.

[0043] In actual applications, first, sensors are installed on key components of the wind turbine according to the data collection step (S1) to collect and transmit data. The full life cycle digital archive management module (S2) receives data and stores and manages it. The AI ​​predictive maintenance analysis module (S3) regularly performs data analysis and fault prediction. When a fault is predicted, the intelligent scheduling and maintenance decision support module (S4) formulates and pushes a maintenance plan. On-site engineers perform tasks and feedback the results according to the maintenance task execution and recording steps (S5). Finally, the maintenance process and strategy are continuously improved through the maintenance effect evaluation and process optimization steps (S6), so as to achieve efficient management of the wind turbine throughout its life cycle, improve the reliability of wind turbine operation, reduce operation and maintenance costs, and improve power generation efficiency.

[0044] The above embodiments are only preferred technical solutions of the present invention and should not be regarded as limiting the present invention. The protection scope of the present invention shall be the technical solutions recorded in the claims, including equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present invention.

Claims

1. A wind turbine full life cycle archive management system, characterized by: The management system includes: IoT data acquisition module, which is configured in the key components of the wind turbine and is used to collect vibration, temperature, and stress operating status data in real time; The full life cycle digital archive management module is a database system built on the cloud, which uses blockchain technology to record all information from design, manufacturing, installation to operation and maintenance to ensure data integrity and traceability; AI predictive maintenance analysis module, based on deep learning and machine learning algorithms, analyzes received data, identifies equipment status change trends and abnormal patterns, and predicts possible future failures; Intelligent scheduling and maintenance decision support module, which automatically generates the optimal maintenance plan based on external factors such as geographic location, weather conditions and grid demand, and provides it to the maintenance team; The responsibility tracing and process optimization module records all maintenance activity details and responsible persons, and uses data analysis to optimize maintenance processes and spare parts management strategies.

2. According to claim 1, a wind turbine full life cycle archive management system is characterized by: The acquisition module also includes at least one type of sensor, selected from vibration sensors, temperature sensors, and stress monitors, for accurately capturing changes in physical parameters during the operation of the wind turbine.

3. According to claim 1, a wind turbine full life cycle archive management system is characterized by: The full life cycle digital archive management module further includes a blockchain-based distributed ledger system to ensure that the creation, update and access of each device archive record are transparent and cannot be tampered with, and supports efficient query and data sharing.

4. According to claim 1, a wind turbine full life cycle archive management system is characterized by: The AI ​​predictive maintenance analysis module uses at least one machine learning model, including support vector machine, random forest or neural network, to train the collected multi-dimensional data to achieve early identification and warning of faults.

5. According to claim 1, a wind turbine full life cycle archive management system is characterized by: The scheduling and maintenance decision support module integrates intelligent algorithms and can automatically adjust the maintenance plan, taking into account maintenance costs, downtime impact and spare parts supply status to minimize operation and maintenance costs and maximize power generation efficiency.

6. According to claim 1, a wind turbine full life cycle archive management system is characterized by: The responsibility tracing and process optimization module is equipped with data mining functions to analyze the relationship between maintenance activities and failure occurrence, providing a basis for continuously improving maintenance strategies and improving operation and maintenance efficiency.

7. According to claim 1, a wind turbine full life cycle archive management system is characterized by: Further included is a user-friendly mobile application that enables field engineers to receive maintenance task notifications, view detailed failure analysis reports, historical repair records, and operation guides.

8. According to claim 1, a wind turbine full life cycle archive management system is characterized by: The computer program product includes computer program code stored on a non-transitory computer-readable storage medium, and when executed by a computer, causes the computer to perform all the functions of the above-mentioned system.

9. A management method for a wind turbine full life cycle archive management system according to any one of claims 1 to 8, characterized in that: The method includes: S1. Data collection steps: Install vibration sensors and stress monitors at the root of the wind turbine blades and key nodes of the wind turbine, install temperature sensors and vibration sensors at appropriate positions of the stator winding, bearings and housing of the generator, and install temperature sensors, vibration sensors and stress monitors at different positions of the gearbox, near each bearing seat and key stress-bearing structure. Each sensor collects the vibration, temperature and stress physical parameter change data of the wind turbine during operation according to the set sampling frequency. The IoT data collection module cleans and converts the collected raw data, and transmits it to the full life cycle digital archive management module in real time through wireless or wired transmission. S2. Data storage and management steps: After receiving the data, the full life cycle digital archive management module stores it in the cloud database according to the data source and type, and enters the fan design drawings, manufacturing process parameters, installation and commissioning records, and non-real-time operation data; Use a blockchain-based distributed ledger system to create a unique hash value identifier for each device archive record, record the timestamp of the data operation, operator information, and content before and after the modification, to ensure that data creation, update, and access are transparent, tamper-proof, and traceable, and only authorized users or nodes can operate the data; S3, data analysis and fault prediction step: The AI ​​predictive maintenance analysis module extracts the wind turbine operation data of a specific time period from the full life cycle digital archive management module and performs standardized processing on it; Select a support vector machine, random forest or neural network machine learning model, use historical normal and fault data to train the model, adjust model parameters, and enable the model to learn the data characteristics of normal and faulty status of the equipment; input the real-time collected and pre-processed data into the trained model, calculate the degree of deviation between the current status of the equipment and the normal status, and predict the type and time range of possible faults in the future; S4. Maintenance plan formulation step: The intelligent dispatching and maintenance decision support module receives the fault prediction results and interacts with the geographic information system, meteorological monitoring system, and power grid dispatching system to obtain the wind turbine's geographical location, weather conditions, and power grid demand information, combined with the company's internal maintenance resource situation; With the goal of minimizing operation and maintenance costs and maximizing power generation efficiency, the system meets the technical requirements of maintenance tasks and safety regulations, and uses integrated intelligent algorithms to automatically generate the optimal maintenance plan, including maintenance task arrangements, work processes, and power generation plan adjustment suggestions; S5. Maintenance task execution and recording steps: The intelligent scheduling and maintenance decision support module pushes the maintenance plan to the field engineer through the mobile application. The field engineer receives the notification and views the fault analysis report, historical maintenance records and operation guide. The field engineer brings tools and spare parts to the wind turbine site as required, performs maintenance operations according to the operation process, and records key operation steps and data. The responsibility tracing and process optimization module records the maintenance personnel information, operation time, content maintenance activity details and stores them in the full life cycle digital archive management module in association with the maintenance task. After completing the task, the field engineer feedbacks the maintenance results through the mobile application. S6. Maintenance effect evaluation and process optimization steps: Determine evaluation indicators based on maintenance task objectives, such as operational stability of equipment after repair, recovery of power generation efficiency, and whether maintenance costs meet expectations; collect operational data before and after maintenance, maintenance activity record data, and cost data from the full life cycle digital archive management module, and use data analysis tools to analyze and mine the relationship between maintenance activities and failures; optimize maintenance processes based on evaluation results and analysis conclusions, such as modifying maintenance operation specifications, adjusting maintenance cycles, and optimizing spare parts management strategies, and update the optimized maintenance processes and strategies to the full life cycle digital archive management module.

10. A management method for a wind turbine full life cycle archive management system according to claim 9, characterized in that: The method includes: The specific algorithm method of steps S1-S6 is: S11, data collection step: the collected vibration data sequence is , the temperature data series is , the stress data series is ,in is the number of sampling points; First, calculate the dynamic mean of the data. Taking vibration data as an example, the dynamic mean formula is: ； in It is The dynamic mean of the sampling points, is the dynamic mean of the previous sampling point, is the smoothing coefficient ( ), which is used to reflect the impact of current data values ​​and consider the trend of historical data during data collection, avoiding sudden changes in the mean value due to individual abnormal data points, so that the mean can better follow the dynamic trend of data changes; Then calculate the dynamic standard deviation of the data, the formula is: ； in It is The dynamic standard deviation of the sampling points, It is the dynamic standard deviation of the previous sampling point. The dynamic standard deviation can be used to measure the degree of dispersion of data and help determine the stability and abnormal fluctuation of data. S21, data storage and management steps: Data storage algorithm: set data block Including collection time , Data Source , Data Type And specific data values ; Using hash functions , where || represents a connection operation, It is a custom hash function, such as a polynomial-based hash function: ； in , are randomly generated coefficients ( ), is the degree of the polynomial, is a large prime number; the role of this hash function is to generate a unique identifier for the data block to ensure the integrity and traceability of the data. Any modification to the data will cause a huge change in the hash value, which can be easily detected; S32, data analysis and fault prediction steps: Data analysis and fault prediction algorithm: a hybrid algorithm based on multi-scale entropy MSE and support vector machine SVM is adopted; First, calculate the multi-scale entropy, assuming that the original data sequence , and then coarse-grained it to obtain different scales sequence under ,in ; Multiscale entropy The calculation steps are as follows: For each scale sequence under , calculate its sample entropy ,in is the embedding dimension, is the similarity tolerance; the sample entropy formula is: ； in is satisfied of dimensional vector pair The ratio of the number of pairs to the total number of vector pairs ( ); Multi-scale entropy reflects the complexity and regularity of data at different time scales. The entropy values ​​of wind turbine data in normal operation have a certain characteristic range at different scales, and the entropy value will change in the fault state; Then the multi-scale entropy at different scales is combined into a feature vector , and input it into the support vector machine for fault classification and prediction; the support vector machine constructs the optimal classification hyperplane ,in is the normal vector, is a function that maps input data to a high-dimensional space, is the bias term, using the kernel function Convert the nonlinear problem in low-dimensional space into a linear problem in high-dimensional space, so as to classify the fault type and predict the time range of fault occurrence; S41. Maintenance plan formulation steps: Maintenance plan formulation algorithm: Set maintenance cost Including maintenance personnel labor costs , Spare parts replacement cost , Equipment downtime loss cost , power generation efficiency loss ; Maintenance personnel labor costs ,in is the number of maintenance personnel, It is The unit time wage of a maintenance worker, is the estimated working time; Spare parts replacement cost ,in Is the spare parts type, It is Unit price of spare parts, is the required quantity; Cost of equipment downtime ,in is the rated power of the fan, is the estimated downtime, is the economic value of a unit of electricity; Power generation efficiency loss ,in is the time when the failure occurred, is to restore normal operating time, is the function of actual generated power over time; Constructing a multi-objective optimization function ,in and is the weight coefficient ( ), which indicates the importance of maintenance cost and power generation efficiency loss in the comprehensive goal; the multi-objective optimization problem is solved by intelligent optimization algorithms such as genetic algorithms to obtain the optimal maintenance plan parameters, such as maintenance personnel arrangement, spare parts allocation, and maintenance time; S51, maintenance task execution and recording steps: Maintenance task execution and recording algorithm: During the maintenance task execution process, record each operation step Execution time , Operation results And related data ; Constructing maintenance task execution vectors , store it in the full life cycle digital archive management module for subsequent maintenance effect evaluation and process optimization; S61. Maintenance effect evaluation and process optimization steps: Maintenance effect evaluation algorithm: Operation stability index after equipment repair ,in is the number of sampling points in the evaluation period, is the actual operation data. It is the reference data when the equipment is operating normally; the smaller the index is, the more stable the equipment is. Maintenance cost compliance index ,in is the actual maintenance cost incurred, is the planned maintenance cost, The closer , indicating that the maintenance cost control is better; Power generation efficiency recovery index ,in is the average power generation of the wind turbine after maintenance, is the average power generation before maintenance, Greater than It indicates that the power generation efficiency has been improved; Through comprehensive analysis of these indicators, the association rule Apriori mining algorithm is used to mine the relationship between maintenance activities and failure occurrence; Among them, the association rules between certain specific maintenance operation combinations and the probability of equipment failure occurring again within a certain period of time are found: ,in and It is a maintenance operation. is the probability of failure; based on these association rules and evaluation results, the maintenance process is optimized, such as adjusting the order of maintenance operations, increasing or decreasing the frequency of certain maintenance operations.

Images