Fan maintenance system based on digital twinborn annotation

Abstract

The invention relates to the technical field of fan testing and overhauling, and discloses a fan overhauling system based on digital twinborn labeling, which comprises a digital twinborn module, a precision parameter module, a testing and overhauling module, a labeling control module, an information acquisition module and an information labeling module, wherein the digital twinning module is used for constructing a three-dimensional twinning model of a fan; the precision parameter module is used for acquiring the precision parameter of each part in the three-dimensional twin model; the test and maintenance module is used for performing test and maintenance early warning; the labeling control module is used for dividing labeling precision grades; the information acquisition module is used for acquiring annotation information; and the information labeling module is used for carrying out information labeling on each part. According to the method, by integrating multi-dimensional data analysis and intelligent means, the intelligence and reasonability of information labeling of digital twin parts of the fan are remarkably improved, the operation and maintenance cost is reduced, and powerful data support is provided for testing and overhauling of the fan.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of fan testing and maintenance, and particularly relates to a fan maintenance system based on digital twin labeling. BACKGROUND

[0002] With the rapid development and technological progress of the wind energy industry, as the core equipment for clean energy production, the efficient, safe and stable operation of the wind turbine is crucial to the reliability of energy supply. However, the complexity of the wind turbine system and various uncertainties brought about by long-term operation, especially the wear and tear of parts, fault prediction and preventive maintenance, have become the key to the industry's continuous optimization of operating costs and improvement of overall efficiency. In recent years, with the deep integration of Internet of Things (IoT), big data analysis, artificial intelligence (AI) and digital twin (Digital Twin) technologies, the operation and maintenance management mode of the wind power industry is undergoing a profound transformation. Digital twin technology realizes the accurate simulation and real-time monitoring of the whole life cycle state of the wind turbine by establishing a dynamic mapping relationship between the wind turbine entity and its virtual model. The core value of this technology lies in its ability to build a highly realistic three-dimensional model based on rich operation data and reflect the performance indicators and health status of the wind turbine in real time.

[0003] Currently, in the field of wind turbine operation and maintenance, there is still room for development in the technology of labeling parts to assist in testing and maintenance of wind turbine parts. The precision labeling of parts usually needs to consider multiple dimensions, such as the importance, complexity, maintenance difficulty, maintenance frequency of the parts, etc. These factors together determine the level of labeling precision and the level of detail of the required labeling information. However, the existing technology still needs further research and improvement in quantifying these indicators and assigning reasonable weights, in order to more accurately guide the design, production and maintenance of parts. In addition, the completeness, timeliness and intelligence level of the labeling information of key parts in the existing wind turbine operation and maintenance information system also need to be improved. Ideally, different parts should be labeled with necessary static information (such as material, specifications, design life, etc.) and dynamic information (such as actual wear and tear, stress changes, performance degradation, etc.) according to their precision level. Especially in the labeling of high-precision key parts, introducing deep learning technology to analyze massive historical maintenance data can effectively predict the parts that may fail in the future and make early warning labels on the digital twin model, improving the forward-looking nature of maintenance work.

[0004] A fan maintenance system based on digital twin labeling is provided, which integrates multi-dimensional data analysis and intelligent means to improve the intelligence and rationality of fan testing and maintenance, and reduce operation and maintenance costs.

[0005] The information disclosed in this BACKGROUND section is only for the purpose of increasing the understanding of the general background of the application, and should not be taken as recognition or admission that this information constitutes prior art with respect to any patentably novel subject matter of the present application. SUMMARY

[0006] The technical problem to be solved by the present application is to overcome the defects of the prior art and provide a fan maintenance system based on digital twin labeling, which integrates multi-dimensional data analysis and intelligent means to improve the intelligence and rationality of fan testing and maintenance, and reduce operation and maintenance costs.

[0007] To solve the above technical problems, the present application provides the following technical solutions:

[0008] A fan maintenance system based on digital twin labeling, comprising a digital twin module, a precision parameter module, a testing and maintenance module, a labeling control module, an information acquisition module, and an information labeling module.

[0009] The digital twin module is used to construct a three-dimensional twin model of the fan based on digital twin technology.

[0010] The precision parameter module is used to obtain the precision parameters of each part in the three-dimensional twin model.

[0011] The testing and maintenance module is used to test and maintain each part based on the precision parameters.

[0012] The labeling control module is used to divide each part in the three-dimensional twin model into labeling precision levels based on the precision parameters, and calculate the labeling information update period of each part.

[0013] The information collection module is configured to determine a set of labeling information of each part based on the labeling accuracy level, and collect labeling information contained in the set of labeling information.

[0014] The information labeling module is configured to label information of each part of the three-dimensional twin model based on the set of labeling information, and update the information labeling of each part based on the labeling information of each part in a periodic manner.

[0015] As a preferred scheme of the fan maintenance system based on digital twin labeling, the accuracy parameters include importance score, average repair time, economic value ratio, expected life, average maintenance cycle, and failure risk score of each part.

[0016] The accuracy parameter module is configured with a neural network model, and the failure risk score is calculated as follows: the probability of each failure of the fan is predicted based on the neural network model, and the failure risk score of each part is calculated based on the probability of each failure, that is, the probability of each failure involving any part A is counted and summed to obtain the failure risk score of part A, and if the failure risk score is greater than 1, the failure risk score is set to 1.

[0017] The neural network model includes an input layer, a hidden layer, and an output layer, the input of the neural network model is fan operation parameters, working environment parameters, and historical failure data, and the output is the occurrence probability of each failure of the fan, the fan operation parameters include vibration amplitude, fan temperature, speed, and power, and the working environment parameters include environmental temperature, humidity, wind speed, salt concentration, and corrosion grade.

[0018] As a preferred scheme of the fan maintenance system based on digital twin labeling, the labeling control module divides the labeling accuracy level of each part in the three-dimensional twin model as follows:

[0019] The accuracy index of each part is calculated based on the accuracy parameters, and the labeling accuracy level of each part is divided based on the accuracy index, specifically as follows:

[0020] A first accuracy index threshold and a second accuracy index threshold are set.

[0021] If the accuracy index of any part A is less than the first accuracy index threshold, the labeling accuracy level of part A is low.

[0022] If the accuracy index of any part is not less than the first accuracy index threshold and not greater than the second accuracy index threshold, the labeling accuracy level of part A is medium.

[0023] If the accuracy index of any part is greater than the second accuracy index threshold, then the annotation accuracy level of part A is high.

[0024] As a preferred embodiment of the wind turbine maintenance system based on digital twin annotation described in this invention, the annotation control module uses the following method to classify the annotation accuracy level for each part in the three-dimensional twin model:

[0025] Based on the accuracy parameters, the accuracy index of each part is calculated, and the accuracy level of each part is assigned based on the accuracy index, as follows:

[0026] Set the first precision index threshold and the second precision index threshold;

[0027] If the accuracy index of any part A is less than the first accuracy index threshold, then the annotation accuracy level of part A is low.

[0028] If the accuracy index of any part is not less than the first accuracy index threshold and not greater than the second accuracy index threshold, then the labeling accuracy level of part A is medium.

[0029] If the accuracy index of any part is greater than the second accuracy index threshold, then the annotation accuracy level of part A is high.

[0030] As a preferred embodiment of the wind turbine maintenance system based on digital twin annotation described in this invention, the annotation control module calculates the annotation information update cycle for each part using the following formula:

[0031]

[0032] Among them, U i U represents the update cycle of the annotation information for the i-th part in the 3D twin model; B This indicates the update cycle of basic annotation information; a, λ, and τ are all weighting coefficients.

[0033] As a preferred embodiment of the wind turbine maintenance system based on digital twin annotation described in this invention, the annotation information set for each part determined by the information acquisition module is as follows:

[0034] The annotation information set of any part with a low annotation accuracy level is called the low accuracy set. The annotation information includes part number, part name, model and specifications, installation position and orientation, connection relationship, expected life and basic maintenance records.

[0035] The set of annotation information for any part with an annotation accuracy level of 1 is the medium accuracy set, which includes the annotation information contained in the low accuracy set, detailed maintenance records, health status, and warning thresholds.

[0036] The annotation information set of any part with a high annotation accuracy level is the high-precision set, which includes the annotation information contained in the medium-precision set, fault prediction information, workload data, electromagnetic compatibility, environmental adaptability, 3D model and CAD drawings, redundancy backup status, and performance degradation curve.

[0037] As a preferred embodiment of the wind turbine maintenance system based on digital twin annotation described in this invention, the method for updating the information annotation of each part by the information annotation module is as follows: record the time of information annotation for each part; when the time interval between any part A in the three-dimensional twin model and the last information annotation reaches the annotation information update cycle of part A, collect the annotation information contained in the annotation information set, update the annotation information set of part A, and re-annotate the information of part A in the three-dimensional twin model based on the updated annotation information set.

[0038] As a preferred embodiment of the wind turbine maintenance system based on digital twin annotation described in this invention, the method by which the testing and maintenance module performs testing and maintenance early warning for each component is as follows:

[0039] The system continuously obtains the fault risk score of each component and the probability of occurrence of each type of fault in the wind turbine from the accuracy parameter module. The accuracy parameter module is configured with the fault risk threshold of each component and the probability risk threshold of each type of fault in the wind turbine. If the fault risk score of any component is higher than the corresponding fault risk threshold or the probability of occurrence of any type of fault in the wind turbine is higher than the corresponding probability risk threshold, a test and maintenance warning is sent to the management personnel, and all components involved in the test and maintenance warning are marked as risk components.

[0040] Adjust the labeling accuracy level of the risky parts to high, reset and update the labeling information set of the risky parts, recalculate the labeling information update cycle of the risky parts, and re-label the information of the risky parts.

[0041] If any part has completed testing and maintenance, the labeling accuracy level of the part that has completed testing and maintenance will be adjusted to low, the labeling information set of the part that has completed maintenance will be reset and updated, the labeling information update cycle of the part that has completed maintenance will be recalculated, and the information labeling of the part that has completed maintenance will be re-applied.

[0042] As a preferred embodiment of the wind turbine maintenance system based on digital twin annotation described in this invention, the test and maintenance module is equipped with a priority calculation submodule, which is used to sort the maintenance priorities of risky parts; and pushes a list of the sorted maintenance priorities of risky parts to a preset terminal, in which the estimated repair time of each risky part is marked.

[0043] As a preferred embodiment of the wind turbine maintenance system based on digital twin annotation described in this invention, the system further includes a visualization interaction module for dynamically displaying at least one of the following contents on the three-dimensional twin model:

[0044] Different levels of annotation precision are distinguished by color gradients;

[0045] When a click operation is received on any part in the 3D twin model, a floating window pops up displaying the annotation information set of the clicked part;

[0046] The locations of parts that trigger test and maintenance warnings are marked in real time.

[0047] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0048] This invention, by precisely classifying the labeling accuracy of components and considering factors such as their importance, complexity, and ease of maintenance, can provide targeted reference information for testing and maintenance, avoiding over-maintenance or under-maintenance, thereby improving operational efficiency and resource utilization. The model meticulously displays the minute details of core components while also ensuring effective information allocation, simplifying the display of non-critical parts. This helps maintenance personnel quickly pinpoint the problem and formulate targeted testing and maintenance strategies.

[0049] By analyzing historical maintenance data using deep learning algorithms, the location and timing of potential failures in key wind turbine components can be accurately predicted. Early warning information can be marked on the digital twin model in advance, promoting a shift from reactive maintenance to proactive preventative testing and maintenance. Integrating real-time environmental factors, such as temperature, humidity, wind speed, salinity, and corrosion levels, the annotation information of components in the model is dynamically adjusted, making maintenance work more closely resemble actual conditions. This ensures that, under specific environmental conditions, priority is given to testing and maintaining components that are more susceptible to damage and prone to problems. Attached Figure Description

[0050] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0051] Figure 1 A schematic diagram of the structure of the wind turbine maintenance system based on digital twin annotation provided by the present invention;

[0052] Figure 2 This is a schematic diagram of the structure of the neural network model for predicting wind turbine faults provided by the present invention. Detailed Implementation

[0053] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0054] Currently, wind turbine testing and maintenance primarily rely on physical parameters, such as operating status data and environmental parameters. This limits the accuracy and speed of testing and maintenance, and makes it impossible to effectively predict turbine failures. The inventors of this application have creatively devised a digital twin technology. On one hand, the physical parameters of the wind turbine are represented by a digital twin. On the other hand, the three-dimensional digital twin model further incorporates precision parameters, including the importance score of each component, average repair time, economic value ratio, expected lifespan, average maintenance cycle, and failure risk score. These parameters cannot be directly represented in reality. Therefore, this application, combined with digital twin technology, can further consider these non-realistically collectable parameter data during processing and calculation, thereby enabling wind turbine testing and maintenance.

[0055] Furthermore, this application embeds a neural network model into the precision parameter module, thereby leveraging digital twin and neural network technologies to directly incorporate non-realistically collectible parameter data into the digital twin's digital body. This provides a new application form of digital twins and neural networks, and is well-suited for scenarios such as wind turbine testing and maintenance that require a combination of numerous real and non-realistic parameters. The core concept of this application will be explained in detail below.

[0056] This embodiment introduces a wind turbine maintenance system based on digital twin annotation, referring to... Figure 1 The system includes a digital twin module, a precision parameter module, a testing and maintenance module, a labeling control module, an information acquisition module, and an information labeling module.

[0057] The digital twin module is used to build a three-dimensional twin model of the wind turbine based on digital twin technology;

[0058] This module utilizes digital twin technology and real-time data synchronization via the Internet of Things to construct a high-precision digital twin model of the wind turbine equipment. This model should comprehensively display all physical characteristics, operating status, and working environment information of the wind turbine, ensuring consistency and real-time performance between the model and the physical equipment.

[0059] The accuracy parameter module is used to obtain the accuracy parameters of each part in the three-dimensional twin model;

[0060] The accuracy parameters include the importance score of each part, average repair time, economic value ratio, expected life, average maintenance cycle, and failure risk score.

[0061] The accuracy parameter module is configured with a neural network model; the calculation method of the fault risk score is as follows: predict the probability of each fault of the wind turbine based on the neural network model, and calculate the fault risk score of each part based on the probability of each fault. The method is to count the probability of each fault involved in any part A and sum them to obtain the fault risk score of part A. If the fault risk score is greater than 1, then the fault risk score is set to 1.

[0062] Reference Figure 2 The neural network model includes an input layer, a hidden layer, and an output layer. The inputs to the neural network model are the fan operating parameters, working environment parameters, and historical fault data, and the output is the probability of occurrence of each type of fan fault. The fan operating parameters include vibration amplitude, fan temperature, speed, and power. The working environment parameters include ambient temperature, humidity, wind speed, salt concentration, and corrosion level.

[0063] The annotation control module is used to classify the annotation accuracy level of each part in the 3D twin model based on the accuracy parameters, and to calculate the annotation information update cycle of each part;

[0064] The method for classifying annotation accuracy levels is as follows:

[0065] The accuracy index for each part is calculated using the following formula:

[0066]

[0067] Among them, I i This represents the accuracy index of the i-th part in the 3D twin model; the value of i ranges from 1, 2, ..., n, where n is the number of parts in the 3D twin model.

[0068] S i This represents the importance score of the i-th part in the 3D twin model, assigned by experts. For example, the importance score range for any part can be set from 1 to 10, with a higher score indicating greater importance. A score of 10 represents a critical, core component, the loss of which would cause the wind turbine to malfunction; a score of 1 indicates an auxiliary or non-critical component. The importance score reflects the basic importance of a component in its design and operation, and is directly proportional to its importance.

[0069] T iThis represents the average repair time of the i-th part in the 3D twin model; it is calculated based on historical data. Average repair time refers to the time required from detecting a fault to restoring normal operation. If the average repair time of a component is long, the system downtime will be longer when a fault occurs, resulting in greater economic or performance losses.

[0070] V i This represents the economic value ratio of the i-th part in the three-dimensional twin model; it is expressed as the ratio of the cost of this part to the total cost of the wind turbine; high-value components contribute significantly to the overall value of the equipment, and are therefore positively correlated with their importance.

[0071] L i This represents the expected lifespan of the i-th part in the 3D twin model; it is obtained from reference data provided by the manufacturer; theoretically, the longer the expected lifespan of a part, the lower the probability of failure.

[0072] F i This represents the average maintenance cycle of the i-th part in the 3D twin model. It is obtained by counting the number of maintenance visits for this part over a certain period and dividing by the duration of that period. A shorter maintenance cycle indicates that the part frequently requires maintenance or replacement, which usually reflects poor reliability or high vulnerability of the part. Frequent maintenance not only increases operating costs but may also lead to more unnecessary downtime, affecting the stability of the overall system.

[0073] R i This represents the failure risk score of the i-th part in the 3D twin model; α, β, γ, and δ are all weighting coefficients, set empirically; ε is a constant used to ensure that the denominator is not zero.

[0074] Each part is assigned a labeling accuracy level based on the aforementioned accuracy index;

[0075] The first precision index threshold and the second precision index threshold are set based on the experience of those skilled in the art;

[0076] If the accuracy index of any part A is less than the first accuracy index threshold, then the annotation accuracy level of part A is low.

[0077] If the accuracy index of any part is not less than the first accuracy index threshold and not greater than the second accuracy index threshold, then the labeling accuracy level of part A is medium.

[0078] If the accuracy index of any part is greater than the second accuracy index threshold, then the annotation accuracy level of part A is high.

[0079] The formula for calculating the update cycle of the annotation information for each part is as follows:

[0080]

[0081] Among them, U i U represents the update cycle of the annotation information for the i-th part in the 3D twin model; B This indicates the update cycle of basic annotation information and is a reference value for updating the annotation information of a part. It is set by technical personnel in this field based on the fan model and industry knowledge; a, λ, and τ are all weighting coefficients, set based on experience.

[0082] The information acquisition module is used to determine the annotation information set for each part based on the annotation accuracy level, and to acquire the annotation information contained in the annotation information set; specifically as follows:

[0083] The annotation information set of any part with a low annotation accuracy level is called the low accuracy set. The annotation information includes part number, part name, model and specifications, installation position and orientation, connection relationship, expected life and basic maintenance records.

[0084] The part number serves as a unique identifier for the part, facilitating tracking and retrieval. The model specification includes detailed parameters such as dimensions, weight, material, and strength. The connection relationships indicate the connection structure and mechanical transmission between this part and other components. The basic maintenance record shows the date of the last maintenance.

[0085] Common parts with low marking accuracy include auxiliary structural parts, fasteners, and seals;

[0086] The set of annotation information for any part with an annotation accuracy level of 1 is the medium accuracy set, which includes the annotation information contained in the low accuracy set, detailed maintenance records, health status, and warning thresholds.

[0087] The detailed maintenance records include the type and result of the last maintenance and the time of the next maintenance; the health status includes current wear and tear, newness and age, damage status and other status information; the warning thresholds include the safety thresholds for each health status parameter.

[0088] Common parts with a marking accuracy level of medium include gearbox components, pitch system components, and electrical connectors;

[0089] The annotation information set of any part with a high annotation accuracy level is the high-precision set, which includes the annotation information contained in the medium-precision set, fault prediction information, workload data, electromagnetic compatibility, environmental adaptability, 3D model and CAD drawings, redundancy backup status, and performance degradation curve.

[0090] The fault prediction information is based on a neural network model to predict fault risk. Workload data includes the torque, pressure, temperature, and vibration levels experienced by the component. Electromagnetic compatibility (EMC) data applies to electrical components, including EMC test results and electromagnetic environment requirements. Environmental adaptability includes information on corrosion resistance, explosion-proof capabilities, and waterproof ratings. Redundancy status indicates whether the component is equipped with a redundant system and its status. The performance degradation curve is a trend curve showing the performance decline of the component over time, generated from historical data.

[0091] Common parts with high annotation accuracy include spindles, blades, generators, and control system components.

[0092] This labeled information is displayed intuitively through a digital twin model and updated regularly, assisting managers in making effective maintenance decisions and improving the overall operational efficiency and safety of the equipment.

[0093] The information annotation module is used to annotate each part of the 3D twin model based on the annotation information set; and to update the information annotation of each part based on the annotation information update cycle of each part.

[0094] The method for information annotation is as follows: Corresponding information tags or visual interfaces are embedded on the 3D model at the locations of various components, and detailed component information is displayed through mouse hovering, clicking, and other methods. Simultaneously, it connects with the backend database to achieve real-time information updates and queries.

[0095] The method for updating the information annotation of each part is as follows: record the time when the information is annotated for each part; when the time interval between any part A in the 3D twin model and the last information annotation reaches the annotation information update cycle of part A, collect the annotation information contained in the annotation information set, update the annotation information set of part A, and re-annotate the information of part A in the 3D twin model based on the updated annotation information set.

[0096] The testing and maintenance module is used to provide early warnings for testing and maintenance of each part based on the aforementioned accuracy parameters; the method is as follows:

[0097] The system continuously obtains the fault risk score of each component and the probability of occurrence of each type of fault in the wind turbine from the accuracy parameter module. The accuracy parameter module is configured with the fault risk threshold of each component and the probability risk threshold of each type of fault in the wind turbine. If the fault risk score of any component is higher than the corresponding fault risk threshold or the probability of occurrence of any type of fault in the wind turbine is higher than the corresponding probability risk threshold, a test and maintenance warning is sent to the management personnel, and all components involved in the test and maintenance warning are marked as risk components.

[0098] The annotation accuracy level of the risky parts is adjusted to high, the annotation information set of the risky parts is reset and updated, the annotation information update cycle of the risky parts is recalculated, and the information of the risky parts is re-annotated. In this way, when a part has a high-risk failure, the annotation accuracy of the part on the digital twin model can be improved, showing a more detailed internal structure and subtle changes, so as to more accurately locate and solve the problem.

[0099] If any part completes testing and maintenance, the annotation accuracy level of the part is adjusted to low, the annotation information set of the part is reset and updated, the annotation information update cycle of the part is recalculated, and the information is re-annotated for the part. When testing and maintaining any part, the annotation information provided by this solution can greatly improve the targeting and efficiency of the testing and maintenance.

[0100] The testing and maintenance module is equipped with a priority calculation submodule, which is used to sort the maintenance priorities of risky parts; and pushes a list of risky parts sorted by maintenance priority to a preset terminal, in which the estimated repair time of each risky part is marked.

[0101] The system also includes a visualization and interaction module for dynamically displaying at least one of the following on the 3D twin model:

[0102] Different levels of annotation precision are distinguished by color gradients;

[0103] When a click operation is received on any part in the 3D twin model, a floating window pops up displaying the annotation information set of the clicked part;

[0104] The locations of parts that trigger test and maintenance warnings are marked in real time.

[0105] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0106] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A wind turbine maintenance system based on digital twin annotation, characterized in that: It includes a digital twin module, a precision parameter module, a testing and maintenance module, a labeling control module, an information acquisition module, and an information labeling module; The digital twin module is used to build a three-dimensional twin model of the wind turbine based on digital twin technology; The accuracy parameter module is used to obtain the accuracy parameters of each part in the three-dimensional twin model; The testing and maintenance module is used to provide testing and maintenance warnings for each part based on the accuracy parameters. The annotation control module is used to classify the annotation accuracy level of each part in the 3D twin model based on the accuracy parameters, and to calculate the annotation information update cycle of each part; The information acquisition module is used to determine the annotation information set for each part based on the annotation accuracy level, and to acquire the annotation information contained in the annotation information set; The information annotation module is used to annotate each part of the 3D twin model based on the annotation information set; and to update the information annotation of each part based on the annotation information update cycle of each part.

2. The wind turbine maintenance system based on digital twin annotation as described in claim 1, characterized in that: The accuracy parameters include the importance score of each part, average repair time, economic value ratio, expected life, average maintenance cycle, and failure risk score. The accuracy parameter module is configured with a neural network model; the calculation method of the fault risk score is as follows: predict the probability of each fault of the wind turbine based on the neural network model, and calculate the fault risk score of each part based on the probability of each fault. The method is to count the probability of each fault involved in any part A and sum them to obtain the fault risk score of part A. If the fault risk score is greater than 1, then the fault risk score is set to 1. The neural network model includes an input layer, a hidden layer, and an output layer. The inputs to the neural network model are the fan operating parameters, working environment parameters, and historical fault data, and the output is the probability of occurrence of each type of fan fault. The fan operating parameters include vibration amplitude, fan temperature, speed, and power. The working environment parameters include ambient temperature, humidity, wind speed, salt concentration, and corrosion level.

3. The wind turbine maintenance system based on digital twin annotation as described in claim 2, characterized in that: The annotation control module uses the following method to classify the annotation accuracy level for each part in the 3D twin model: Based on the accuracy parameters, the accuracy index of each part is calculated, and the accuracy level of each part is assigned based on the accuracy index, as follows: Set the first precision index threshold and the second precision index threshold; If the accuracy index of any part A is less than the first accuracy index threshold, then the annotation accuracy level of part A is low. If the accuracy index of any part is not less than the first accuracy index threshold and not greater than the second accuracy index threshold, then the labeling accuracy level of part A is medium. If the accuracy index of any part is greater than the second accuracy index threshold, then the annotation accuracy level of part A is high.

4. The wind turbine maintenance system based on digital twin annotation as described in claim 3, characterized in that: The formula for calculating the accuracy index is as follows: Among them, I i This represents the accuracy index of the i-th part in the 3D twin model; the value of i ranges from 1, 2, ..., n, where n is the number of parts in the 3D twin model. S i This represents the importance score of the i-th part in the 3D twin model; T i This represents the average repair time of the i-th part in the 3D twin model; V i This represents the economic value ratio of the i-th part in the three-dimensional twin model; L i This represents the expected lifespan of the i-th part in the 3D twin model; F i This represents the average maintenance cycle of the i-th part in the three-dimensional twin model; R i This represents the failure risk score of the i-th part in the 3D twin model; α, β, γ, and δ are all weighting coefficients; ε is a constant.

5. A wind turbine maintenance system based on digital twin annotation as described in claim 4, characterized in that: The formula used by the annotation control module to calculate the annotation information update cycle for each part is as follows: Among them, U i U represents the update cycle of the annotation information for the i-th part in the 3D twin model; B This indicates the update cycle of basic annotation information; a, λ, and τ are all weighting coefficients.

6. The wind turbine maintenance system based on digital twin annotation as described in claim 5, characterized in that: The specific annotation information set for each part determined by the information acquisition module is as follows: The annotation information set of any part with a low annotation accuracy level is called the low accuracy set. The annotation information includes part number, part name, model and specifications, installation position and orientation, connection relationship, expected life and basic maintenance records. The set of annotation information for any part with an annotation accuracy level of 1 is the medium accuracy set, which includes the annotation information contained in the low accuracy set, detailed maintenance records, health status, and warning thresholds. The annotation information set of any part with a high annotation accuracy level is the high-precision set, which includes the annotation information contained in the medium-precision set, fault prediction information, workload data, electromagnetic compatibility, environmental adaptability, 3D model and CAD drawings, redundancy backup status, and performance degradation curve.

7. A wind turbine maintenance system based on digital twin annotation as described in claim 6, characterized in that: The method for updating the information annotation of each part by the information annotation module is as follows: record the time when information annotation is performed for each part; when the time interval between any part A in the 3D twin model and the last information annotation reaches the annotation information update cycle of part A, collect the annotation information contained in the annotation information set, update the annotation information set of part A, and re-annotate the information of part A in the 3D twin model based on the updated annotation information set.

8. A wind turbine maintenance system based on digital twin annotation as described in claim 7, characterized in that: The method by which the test and maintenance module provides test and maintenance early warnings for each part is as follows: The system continuously obtains the fault risk score of each component and the probability of occurrence of each type of fault in the wind turbine from the accuracy parameter module. The accuracy parameter module is configured with the fault risk threshold of each component and the probability risk threshold of each type of fault in the wind turbine. If the fault risk score of any component is higher than the corresponding fault risk threshold or the probability of occurrence of any type of fault in the wind turbine is higher than the corresponding probability risk threshold, a test and maintenance warning is sent to the management personnel, and all components involved in the test and maintenance warning are marked as risk components. Adjust the labeling accuracy level of the risky parts to high, reset and update the labeling information set of the risky parts, recalculate the labeling information update cycle of the risky parts, and re-label the information of the risky parts. If any part has completed testing and maintenance, the labeling accuracy level of the part that has completed testing and maintenance will be adjusted to low, the labeling information set of the part that has completed maintenance will be reset and updated, the labeling information update cycle of the part that has completed maintenance will be recalculated, and the information labeling of the part that has completed maintenance will be re-applied.

9. A wind turbine maintenance system based on digital twin annotation as described in claim 8, characterized in that: The testing and maintenance module is equipped with a priority calculation submodule, which is used to sort the maintenance priorities of risky parts; and pushes a list of risky parts sorted by maintenance priority to a preset terminal, in which the estimated repair time of each risky part is marked.

10. A wind turbine maintenance system based on digital twin annotation as described in any one of claims 1-9, characterized in that: The system also includes a visualization and interaction module for dynamically displaying at least one of the following on the 3D twin model: Different levels of annotation precision are distinguished by color gradients; When a click operation is received on any part in the 3D twin model, a floating window pops up displaying the annotation information set of the clicked part; The locations of parts that trigger test and maintenance warnings are marked in real time.

Images