Wind power tower and wind field level diagnosis system and method based on digital twinborn technology

Abstract

The invention belongs to the technical field of wind power tower structure monitoring and diagnosis, and relates to a wind power tower and wind field level digital twin diagnosis system and method. The invention discloses a wind power tower diagnosis system based on a digital twinborn technology. The system comprises a sensor module, a customized monitoring system, a data processing algorithm module, a digital twinborn model and a diagnosis and prediction module. According to the system, key parts are monitored and covered in a targeted mode, load reconstruction and model updating are achieved through monitoring data, therefore, structural response is effectively recognized, performance degradation of tower global connection structure details is diagnosed and predicted online, and the health state of a whole wind field is deduced through limited measuring point data. In addition, the digital twinborn diagnosis monitoring system has real-time monitoring and adjusting functions for active control and semi-active control systems of the mixed tower type. Based on twin data and computational analysis, the system may provide source data to an active or semi-active control system to assist in structural vibration control.

Description

Technical Field

[0001] The present invention belongs to the technical field of wind power tower structure monitoring and diagnosis, and relates to a wind power tower and wind farm-level digital twin diagnosis and measurement system and method. Background Art

[0002] In the development of wind power, the development of wind farms in low wind speed areas such as plains faces many challenges. The wind speed in these areas is relatively low. In order to better utilize the high wind speed resources at high altitudes, it is necessary to increase the installation height of wind turbines. Traditional cylindrical towers will face problems such as large material usage and increased costs when the height increases, and the stability of a single truss structure at high wind speeds is also insufficient. Therefore, the use of a new truss-tube combined tower has become an effective solution to increase the installation height of wind turbines. The truss-tube combined tower combines the advantages of cylindrical towers and trusses, which can not only ensure the stability of the tower, but also reduce material costs. However, at the same time, this type of new structure has the characteristics of higher investment costs, significantly increased number of easily degraded details, and significantly increased difficulty in operation and maintenance. It is urgent to achieve timely and effective control of its structural integrity. At the same time, with the increase in tower height and unit capacity, the dynamic response of the tower structure under wind load becomes more prominent, and the customized application of structural control systems has undoubtedly become a necessary measure for "reducing costs and increasing efficiency". However, the application effectiveness of the above-mentioned combined tower-mechanical damping coupling system and the completeness of the control system perception source data are still insufficient, and targeted measures are urgently needed to strengthen them.

[0003] For the above-mentioned new truss-tube combined tower, the existing conventional structural monitoring system has problems such as incomplete monitoring and inaccurate diagnosis when dealing with complex environmental loads and structural performance degradation during long-term operation. In particular, it is difficult for traditional monitoring systems to cover the key parts of wind turbine towers, resulting in the inability to detect and deal with structural performance degradation in a timely manner. In addition, the stress conditions and performance degradation of different towers in the same wind farm are both correlated and different, making it difficult to fully understand the health status of the entire wind farm through limited measurement point data. Current technologies mainly focus on the design optimization and local monitoring of wind turbine towers, and lack an overall monitoring and diagnosis solution for truss-tube hybrid structures and wind farm-level correlation analysis. Therefore, it is necessary to propose a technology for wind farm-level analysis of the performance degradation of truss-tube combined towers to achieve more comprehensive monitoring and diagnosis and improve the overall operation efficiency and reliability of wind farms. Summary of the invention

[0004] In response to the problems of incomplete monitoring and inaccurate diagnosis in the prior art, the present invention proposes a truss-tube hybrid wind power tower and wind farm-level diagnostic system and method based on digital twin technology. Through targeted monitoring to cover key parts, the monitoring data is used to achieve load reconstruction and model update, thereby effectively identifying structural response, online diagnosis and prediction of performance degradation of the tower's global connection structure details, and inferring the health status of the entire wind farm through limited measurement point data. In addition, this digital twin diagnostic monitoring system has real-time monitoring and adjustment functions for active control and semi-active control systems of this type of hybrid tower. Based on twin data and computational analysis, the system can provide source data to the active control or semi-active control system to assist in structural vibration control.

[0005] The present invention provides a technical solution to solve the technical problem: a truss-tube hybrid wind power tower and wind farm-level diagnostic and testing system based on digital twin technology, a sensor module, a customized monitoring system, a data processing algorithm module, a digital twin model, and a diagnosis and prediction module; the sensor module is connected to the data acquisition and monitoring system; The sensor module is used to obtain the operation data of the monitoring part of the wind power tower; The customized monitoring system is used to collect and manage the data acquired by the sensor module and control the operation of each sensor in the sensor module; The data processing algorithm module generates real-time load data of the wind power tower top using sensor data and load reconstruction technology; The digital twin model updates model parameters according to the input real-time load data; The diagnosis and prediction module combines the updated model parameters and real-time load data to diagnose the health status and predict the performance degradation of the wind power tower.

[0006] Preferably, the digital twin model is constructed using modeling tools based on the design parameters and engineering requirements of the physical wind power tower and includes all the physical properties of the wind power tower.

[0007] Preferably, the diagnosis and prediction module includes a pre-trained performance degradation model, which is established by integrating the latest model parameters, response data and real-time load data obtained in the digital twin model, extracting time series data and frequency domain feature data as well as physical model feature data; and using a machine learning algorithm to train the extracted features.

[0008] Preferably, the data processing algorithm module utilizes a Kalman filter algorithm and a formula inverse method to achieve load reconstruction of the top of the wind tower.

[0009] The present invention also provides a truss-tube hybrid wind farm diagnostic system based on digital twin technology, which includes several truss-tube hybrid wind tower diagnostic systems, each wind tower diagnostic system corresponds to a wind tower in the wind farm; through virtual sensing and association models, the load history and performance degradation status of each wind tower in the wind farm are predicted, thereby realizing the health status diagnosis and performance degradation prediction of all wind towers in the wind farm.

[0010] The present invention further provides a truss-tube hybrid wind power tower diagnosis and testing method based on digital twin technology, using the wind power tower diagnosis and testing system, comprising the following steps: (1) Collect monitoring data in real time and transmit the data to the data processing algorithm module; based on the monitoring data, reconstruct the load and calculate the real-time load data of the top of the wind turbine tower; (2) Based on the digital twin model, dynamically update, adjust and optimize model parameters according to the input real-time load data; (3) Combining the updated model with real-time load data, the diagnosis and prediction module is used to diagnose the health status of wind turbine towers and predict their performance degradation.

[0011] Compared with the prior art, the present invention has the following beneficial effects: (1) Comprehensive and customized monitoring of truss-tube hybrid wind turbine towers covering key parts; (2) Using digital twin technology and load reconstruction methods, the structural model is dynamically updated to improve the accuracy of diagnosis; (3) Effectively prevent and address potential safety hazards of wind turbine tower structures and extend the service life of wind turbine towers through performance degradation diagnosis and prediction models; (4) Through wind farm-level correlation analysis, the health status of all wind turbine towers in the wind farm can be inferred and performance optimized, improving the overall efficiency and reliability of wind farm operation and maintenance; (5) Monitoring and diagnosis of key components of wind turbines, such as gearboxes and drive trains, to ensure efficient operation and safety of the entire wind power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a framework diagram of a truss-tube hybrid wind turbine tower and a wind farm-level diagnostic system based on digital twin technology in an embodiment of the present invention; Figure 2 This is a flow chart of load reconstruction in an embodiment of the present invention; Figure 3 A flowchart of updating a digital twin model in an embodiment of the present invention; Figure 4 A schematic diagram of a monitoring and diagnosis framework for key components of a unit according to the present invention; Figure 5A schematic diagram of a process for diagnosing and predicting a wind power tower according to an embodiment of the present invention; Figure 6 It is a schematic diagram of a logical framework for wind farm-level diagnosis and prediction implementation in an embodiment of the present invention; Figure 7 The present invention implements a twin monitoring and analysis framework schematic diagram for the entire wind power tower. DETAILED DESCRIPTION

[0013] In order to facilitate the understanding of the present invention, the present invention is described in more detail below in conjunction with the accompanying drawings and specific embodiments. Preferred embodiments of the present invention are provided in the accompanying drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described in this specification. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present invention more thorough and comprehensive.

[0014] Embodiment 1 This embodiment provides a diagnosis and prediction system for a truss-tube hybrid wind turbine tower based on digital twin technology, such as Figure 1-5 As shown, the system includes: sensor module, customized monitoring system, data processing algorithm module, digital twin model, diagnosis and prediction module.

[0015] The sensor module includes various sensors, such as visual sensors, audio sensors, axial force sensors, acceleration sensors, stress sensors, displacement sensors, wind speed sensors, wind direction sensors, etc. These sensors are arranged at the connection nodes between the truss and the cylinder and other key parts (such as the gearbox and drive chain of the wind turbine) to collect monitoring data of the monitoring parts.

[0016] The sensor module is connected to the customized monitoring system through the data collector and the frequency synchronizer, and the acquired sensor data is transmitted to the customized monitoring system of the host computer using wireless and wired transmission technologies. The wireless transmission protocols include Zigbee, MQTT, Wifi and Bluetooth transmission.

[0017] The customized monitoring system manages the collected sensor data by storing and deleting, and controls and adjusts the work of each sensor according to different monitoring purposes. The customized monitoring system can receive data from different sensor modules in real time, and classify, store, delete, back up and restore them. The system also has the function of managing historical data, which can query, analyze and export data of different time periods according to needs to ensure the integrity and traceability of the data. The system can perform preliminary processing and analysis on the received sensor data, such as data filtering, noise reduction, and outlier detection, to ensure the accuracy and effectiveness of the data. Advanced analysis algorithms can also be integrated to calculate the structural health indicators of the tower in real time, such as the changing trends of stress, vibration and displacement. The customized monitoring system provides real-time monitoring functions, and users can view the real-time status and data of the sensors through a graphical interface. The system also has an early warning mechanism. When it detects that key parameters exceed the set safety threshold, it will automatically trigger an alarm to remind maintenance personnel to intervene.

[0018] The data processing algorithm module has a built-in Kalman filter algorithm, including the original Kalman filter algorithm, the extended Kalman filter algorithm, and variants of the Kalman filter algorithm, as well as a load reconstruction algorithm.

[0019] The digital twin model is a detailed model created based on the design parameters and engineering requirements of the physical wind tower using advanced modeling tools and techniques. This model includes all the key physical characteristics of the wind tower, such as the tower's structural composition, material properties, geometric dimensions, and its response to environmental factors (wind speed, temperature, and humidity). The model is dynamically updated and optimized using data collected by field sensors and real-time load data generated by load reconstruction technology. During the model optimization process, an iterative solution method is used to continuously adjust and optimize the model's parameters. Through multiple iterative optimizations, the digital twin model will be able to accurately reflect the performance and behavior of the physical wind tower.

[0020] The diagnosis and prediction module is a health diagnosis model and performance degradation prediction model for wind turbine towers built using machine learning technology. The latest updated model, response data and real-time load data obtained from the digital twin model are integrated, and the time series data, frequency domain feature data and physical model feature data are analyzed to establish a performance degradation model, including a fatigue degradation model and a tower performance degradation model. The extracted features are trained using machine learning algorithms to learn the complexity of tower performance degradation, and to establish a health diagnosis model and a performance degradation prediction model. The data processing methods involved are as follows:

[0021] 1. Data integration method The main purpose of integrating data is to unify data from different sources (such as updated models, response data, and real-time load data) into a framework for consistent analysis. Data preprocessing: First, data from different sources need to be preprocessed, such as data cleaning (removing noise or outliers), normalization (unifying dimensions), interpolation (handling missing data), etc. This is to ensure the consistency and comparability of the data. Time synchronization: Synchronize the time series data collected by the sensor with the load data to ensure that the data is on the same timeline. It is usually necessary to align the data from different sensors, either by timestamp or by using a clock synchronization algorithm. Use data fusion technology (Kalman filtering, Bayesian reasoning) to merge data from different sources.

[0022] 2. Analyze time series data, frequency domain feature data, and physical model feature data (1) Time series data analysis Trend analysis: smoothing time series data, observing long-term trends, and understanding long-term changes in the structure, such as tilt, fatigue, etc. Short-term event detection: using spike detection algorithms or sliding windows to detect abnormal events that occur in the short term, such as excessive vibration, transient loads, etc. Time series prediction: using autoregressive models (AR), autoregressive moving average models (ARMA), or long short-term memory networks (LSTM) to predict future structural states or responses.

[0023] (2) Frequency domain feature data analysis Frequency domain data analysis aims to extract frequency information from time series data to help identify the dynamic characteristics of the structure. Fourier transform (FFT) converts time domain data to frequency domain to identify the natural frequency, vibration mode and energy distribution of the structure.

[0024] 3. Physical model feature data analysis Finite Element Analysis (FEA): Perform static and dynamic analysis based on the updated model to predict the deformation, stress distribution, modal and other characteristics of the structure. Modal analysis: Use measured data to perform modal analysis on the physical model, extract the natural frequency, modal vibration shape and other characteristics of the updated model, and compare them with the measured values ​​to verify the accuracy of the model. Parameter sensitivity analysis: Perform sensitivity analysis on each parameter of the physical model to evaluate the impact of each parameter on the structural response, so as to find the parameters that are most sensitive to the structural behavior.

[0025] 4. Feature extraction The features extracted from time series, frequency domain data and physical models mainly include: Time domain features, maximum amplitude, mean, standard deviation, peak factor, etc. These features reflect the vibration intensity and stability of the structure in the time domain. Frequency domain features, natural frequency, frequency response function, resonant frequency, main frequency bandwidth, spectrum energy, etc. This feature reflects the dynamic behavior of the structure at different frequencies. Physical features, stiffness, damping, modal vibration shape, stress distribution, displacement field, etc. These are the key mechanical properties of the structure, which directly reflect the stiffness and deformation capacity of the structure.

[0026] In the structure of wind turbines, gearboxes and drive trains are key components of wind turbines, and their performance has a decisive impact on the overall operating efficiency and safety of wind turbine towers. In order to ensure the normal operation of these components, the monitoring data of the SCADA system (the monitoring and data acquisition system of the wind turbine tower), SHM system and customized monitoring system are used to conduct in-depth performance analysis and degradation prediction of gearboxes and drive trains. Next, a data-driven approach and machine learning algorithms are used to conduct in-depth analysis of the collected data (specifically including preprocessing, feature extraction, pattern recognition, and trend prediction of the collected data. First, the raw data is preprocessed by cleaning, noise reduction, and normalization to ensure data quality and consistency. Then, signal processing techniques (such as time-frequency analysis, Fourier transform, or wavelet transform) are used to extract key time and frequency domain feature information to reflect the health status of the gearbox and drive chain. Next, pattern recognition and classification are performed through machine learning algorithms (such as support vector machines, neural networks, or random forests) to find abnormal patterns and potential signs of degradation in the data. Finally, fault prediction and life estimation are performed based on historical data and current trends, so that maintenance measures can be taken in advance to ensure the safe and reliable operation of the system), thereby extracting key feature information that can reflect the health status of the components. Based on this feature information, efficient diagnostic and prediction models are trained, which can accurately identify the performance degradation trend of the gearbox and drive chain and predict possible future failures. At the same time, accurate degradation prediction can better arrange and optimize component maintenance plans. In addition, the real-time early warning feedback mechanism provided by the system can immediately notify the operation and maintenance team when abnormal data is detected, ensuring that the fault is handled quickly and professionally.

[0027] Example 2 Based on the digital twin technology-based truss-tube hybrid wind turbine tower diagnosis and prediction system provided in Example 1, this example provides a wind farm-level diagnosis and prediction system, such as Figure 6 , 7 shown.

[0028] Representative wind turbine towers in the wind farm are selected as specific monitoring objects, and their detailed structural response data are collected. A digital twin model of a single wind turbine tower is established, and this model is used as a benchmark. The correlation model is trained based on the data of this benchmark model, and the structural response of other wind turbine towers in the wind farm is inferred using virtual sensing technology. At the same time, limited monitoring data and unit monitoring data are used to achieve wind farm-level correlation analysis through multi-physics field simulation and spatial correlation models, and a wind farm-level digital twin scenario is established.

[0029] The system of the present invention builds a high-performance machine learning model through real-time monitoring data and load reconstruction results, and conducts in-depth performance degradation analysis on key parts of wind turbine towers, such as the connection nodes between trusses and cylinders and supporting structures. In this process, not only the changes of a single indicator are considered, but also multiple indicators are integrated for comprehensive judgment, which significantly improves the accuracy and reliability of diagnosis. The system of the present invention can realize real-time early warning based on comprehensive judgment of multiple indicators. Once any abnormal data is monitored, the early warning mechanism will be immediately activated and the early warning information will be sent to the operation and maintenance personnel in time, so as to ensure that potential structural safety hazards are dealt with quickly and effectively. The system also provides a quasi-real-time display function of the current state. The operation and maintenance personnel can intuitively grasp the real-time operating status and key performance parameters of the wind turbine tower through the visual interface of the digital twin model. The system of the present invention also has the ability to deduce and predict future evolution. Based on rich historical data and current state information, the machine learning algorithm can be used to predict the future performance degradation trend of the wind turbine tower. This function provides a scientific reference basis for the formulation of operation and maintenance planning, which helps the operation and maintenance team to plan and optimize resource allocation in advance, thereby improving the overall operation efficiency and safety of the wind farm.

[0030] Embodiment 3 This embodiment provides a diagnosis and prediction method for a truss-tube hybrid wind power tower based on digital twin technology, comprising the following steps: Step S1, based on the existing health monitoring system (SHM) and supervisory control data acquisition system (SCADA) of the wind turbine tower, different types of sensors are arranged at the connection nodes between the truss and the cylinder and other key parts (such as the gearbox and drive chain of the wind turbine) to collect monitoring data in real time and transmit the data to the customized monitoring system.

[0031] Step S2: The data processing algorithm module reconstructs the load based on the acquired monitoring data to generate real-time load data.

[0032] The load reconstruction of the top of the wind turbine tower is realized by using the monitoring data and surveillance data obtained in step 1, using the Kalman filter algorithm and the formula inversion method. The Kalman filter algorithm includes the original Kalman filter algorithm, the extended Kalman filter algorithm and the variant of the Kalman filter algorithm; the formula inversion takes the cantilever beam as the mechanical model benchmark.

[0033] Step S3: The established digital twin model of the hybrid wind turbine tower is dynamically updated according to real-time load data.

[0034] Step S4: The diagnosis and prediction module performs performance degradation analysis based on the updated model and real-time load data, and makes a diagnosis and performance degradation prediction on the health status of key components of the wind turbine tower.

[0035] Step S5: Effectively diagnose and predict performance degradation, and provide real-time early warning feedback based on comprehensive judgment of multiple indicators, real-time display of current status, and prediction of future evolution, so as to achieve effective monitoring and maintenance of key components of wind power towers.

[0036] Similarly, through multi-physics field simulation and spatial correlation models, wind farm-level correlation analysis can be performed to infer the health status and predict the performance of key parts of all wind turbine towers in the wind farm.

Claims

1. A wind power tower diagnostic system based on digital twin technology, characterized in that: The system includes: sensor module, customized monitoring system, data processing algorithm module, digital twin model, diagnosis and prediction module; The sensor module is used to obtain the operation data of the monitoring part of the wind power tower; The customized monitoring system is used to collect and manage the data acquired by the sensor module and control the operation of each sensor in the sensor module; The data processing algorithm module generates real-time load data of the wind power tower top using sensor data and load reconstruction technology; The digital twin model updates model parameters according to the input real-time load data; The diagnosis and prediction module combines the updated model parameters and real-time load data to diagnose the health status and predict the performance degradation of the wind power tower.

2. The wind turbine tower diagnostic system based on digital twin technology according to claim 1, characterized in that: The digital twin model is constructed using modeling tools based on the design parameters and engineering requirements of the wind turbine tower and includes all physical properties of the wind turbine tower.

3. The wind turbine tower diagnostic system based on digital twin technology according to claim 1, characterized in that: The diagnosis and prediction module includes a pre-trained performance degradation model, which is established by integrating the latest model parameters, response data and real-time load data obtained from the digital twin model, extracting time series data and frequency domain feature data as well as physical model feature data; and using a machine learning algorithm to train the extracted features.

4. The wind power tower diagnostic system based on digital twin technology according to claim 1, characterized in that: The data processing algorithm module utilizes the Kalman filter algorithm and the formula inverse method to achieve load reconstruction of the top of the wind power tower.

5. A truss-tube hybrid wind farm diagnostic system based on digital twin technology, characterized by: The method comprises a plurality of wind turbine tower diagnostic systems as described in claim 1, wherein the system comprises a plurality of the truss-tube hybrid wind turbine tower diagnostic systems, each wind turbine tower diagnostic system corresponding to a wind turbine tower in a wind farm; through virtual sensing and association models, the load history and performance degradation status of each wind turbine tower in the wind farm are predicted, thereby realizing health status diagnosis and performance degradation prediction of all wind turbine towers in the wind farm.

6. A wind power tower diagnostic method based on digital twin technology, characterized in that: Utilizing the system according to any one of claims 1 to 4, comprising the following steps: (1) Collect monitoring data in real time and transmit the data to the data processing algorithm module; based on the monitoring data, reconstruct the load and calculate the real-time load data of the top of the wind turbine tower; (2) Based on the digital twin model, the model parameters are dynamically updated, adjusted and optimized according to the input real-time load data; (3) Combining the updated model with real-time load data, the diagnosis and prediction module is used to diagnose the health status of wind turbine towers and predict their performance degradation.

Images