A wind turbine status early warning method based on digital twin

Abstract

The present invention discloses a method for early warning of wind turbine status based on digital twins. The method comprises: an entity of key components of a wind turbine; constructing an initial digital twin model based on real-time state quantity data of the key component entities; obtaining statistical features and correlation characteristics of the initial digital twin model; performing simulation prediction on the monitored state quantity, and updating the digital twin model based on the simulation prediction data; obtaining statistical features and correlation characteristics of the updated digital twin model; comparing the features of the updated digital twin model with the features of the initial digital twin model, judging the operating status of key components of the wind turbine and issuing an early warning. The present invention solves the technical problem that the current related technology relies on state monitoring data and a large number of fault cases to construct a pattern recognition model when issuing an early warning of the operating status of the wind turbine, ignoring the impact of the hidden latent defects of the wind turbine on the judgment model and the problem that the recognition model cannot consider the future operating trend of the wind turbine, resulting in the inability to accurately judge and early warning the operating status of the wind turbine.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power generation, and in particular to a wind turbine status early warning method based on digital twins. Background Art

[0002] Wind turbines are a vital component of wind power generation systems, and their operating status directly impacts the efficiency and quality of wind power generation. However, due to the complex operating conditions of wind turbines subjected to alternating loads, key components such as the gearbox, discharge motor, and blades are subject to varying degrees of impact and even damage. Therefore, real-time assessment of wind turbine operating status and early warning of abnormal operating conditions are crucial for preventing component damage and reducing operating and maintenance costs. In recent years, digital twin technology has rapidly developed, providing a key means for accurate early warning of wind turbine status. Digital twin technology incorporates data such as physical models, sensor updates, and operational history, integrating multidisciplinary, multi-physics, and multi-scale simulation processes to map physical entities in a virtual space, reflecting the entire lifecycle of the physical entity. The data simulation process within the virtual space of digital twin technology provides a technical means for uncovering patterns in wind turbine equipment operating trends.

[0003] At present, digital twin technology has also been applied to wind power generation systems and has made some progress. The patent "Wang Congjun, et al., Intelligent monitoring and early warning system for deep foundation pits at construction sites based on digital twin technology, CN 113404029 A" integrates multiple monitoring data from the construction site to construct a digital twin, but the early warning method is still based on the threshold comparison method; the patent "Fang Fang, et al., A digital twin system for wind power generation, CN 113236491 A" establishes a digital twin model for the entire wind turbine, including interactive control, human-computer interaction and other functions, focusing on the control analysis of the entire wind turbine, but does not involve status early warning; the patent "Wang Wei, et al., Offshore wind power digital twin test pile test system and establishment method, CN113297769 A" constructs a digital twin system including data monitoring and visualization for the offshore wind turbine pile construction process, but does not involve status judgment; the patent "Wang Liguo, et al., DFIG wind farm synchronous oscillation suppression method based on digital twin simulation, CN 113193589 A》The generator was virtualized using digital twin technology, and a subsynchronous oscillation suppression method was proposed based on this, but status warning was not involved.

[0004] In summary, existing research has not yet effectively combined digital twin technology with wind turbine status warning, that is, an effective wind turbine status warning method based on digital twin technology has not yet been formed. Summary of the Invention

[0005] An embodiment of the present invention provides a wind turbine status warning method based on digital twins, which at least solves the technical problem that the current related technology relies on status monitoring data and a large number of fault cases to build a pattern recognition model when warning the wind turbine operating status, ignores the impact of the wind turbine's hidden latent defects on the judgment model, and the recognition model cannot consider the future operating trend of the wind turbine, resulting in the inability to accurately judge and warn the wind turbine operating status.

[0006] The technical solution of the present invention is: key component entities of the wind turbine; constructing an initial digital twin model based on the real-time state quantity data of the key part entities; obtaining the statistical characteristics and correlation characteristics of the initial digital twin model; performing simulation prediction on the monitored state quantities, and updating the digital twin model based on the simulation prediction data; obtaining the statistical characteristics and correlation characteristics of the updated digital twin model; comparing the characteristics of the updated digital twin model with the characteristics of the initial digital twin model, judging the operating status of the key components of the wind turbine and issuing an early warning.

[0007] Furthermore, the key component entity of the wind turbine includes at least one of a gearbox, a generator and a blade.

[0008] Furthermore, the construction of the initial digital twin model based on the real-time state quantity data of the key part entity includes: the state quantity reflecting the operating status of the key part gearbox includes but is not limited to at least two of the gearbox oil temperature, gearbox bearing temperature, ambient temperature, and vibration amplitude; the state quantity reflecting the operating status of the generator includes but is not limited to at least two of the generator speed, generator front shaft temperature, generator rear shaft temperature, generator temperature, A phase current, B phase current, C phase current, AB phase voltage, BC phase voltage, CA phase voltage, frequency, active power, reactive power, and power factor; the state quantity reflecting the operating status of the blade includes but is not limited to at least two of wind speed, lightning, icing, stress, tip pressure, blade angle, torque, and blade average speed; the steps of constructing the initial digital twin model are: obtaining the size information, material information, structure information, manufacturer information, own attribute information, and operating environment information of the key part entity to construct a static 3D model that can intuitively reflect the key part entity; digitally mapping the real-time state data of the key part entity obtained in real time to the static 3D model to form an initial digital twin model of the key part entity.

[0009] Furthermore, obtaining the statistical characteristics and correlation characteristics of the initial digital twin model includes: obtaining the statistical characteristics of the initial digital twin model is to perform statistical analysis on the monitoring data of each type of state quantity in the digital twin model, obtain the distribution model that the state quantity conforms to, and obtain the key parameters of the distribution model; the distribution model includes but is not limited to at least one of the normal distribution, Weibull distribution, and chi-square distribution; obtaining the correlation characteristics of the initial digital twin model is to perform correlation analysis on the time series of all state quantities reflecting the key part entities, and obtain the correlation coefficient between each state quantity.

[0010] Furthermore, simulation prediction is performed on the monitored state quantity, and the digital twin model is updated based on the simulation prediction data, including: the simulation prediction of the monitored state quantity is to predict the time series of all state quantities of the key part entity; the prediction method includes but is not limited to at least one of ARIMA prediction, BP neural network prediction, and LSTM prediction; the updating of the digital twin model based on the simulation prediction data is to add the predicted data to the initial digital twin model of the key part entity.

[0011] Furthermore, obtaining the statistical characteristics and correlation characteristics of the updated digital twin model includes: obtaining the statistical characteristics of the updated digital twin model is to perform statistical analysis on the monitoring data of each type of state quantity in the updated digital twin model, obtain the distribution model that the state quantity conforms to, and obtain the key parameters of the distribution model; the distribution model includes but is not limited to at least one of the normal distribution, Weibull distribution, and chi-square distribution; obtaining the correlation characteristics of the updated digital twin model is to perform correlation analysis on the time series of all state quantities reflecting the key part entities, and obtain the correlation coefficient between each state quantity.

[0012] Furthermore, the characteristics of the updated digital twin model are compared with the initial digital twin model to judge the operating status of the key components of the wind turbine and issue an early warning, including: comparing the distribution model characteristics of each state quantity in the updated digital twin model with the distribution model characteristics of each state quantity in the initial digital twin model, and if the distribution model characteristic change range exceeds 30%, the state quantity is judged to be abnormal; comparing the correlation characteristics between each state quantity in the updated digital twin model with the correlation characteristics between each state quantity in the initial digital twin model, and if the correlation characteristic change range exceeds 30%, the state quantity is judged to be abnormal; the specific steps of judging the operating status of the key components of the wind turbine and issuing an early warning are: if the correlation characteristics and distribution model characteristics of the state quantity are both normal, the operating status of the key component is judged to be normal; if one of the correlation characteristics and distribution model characteristics of the state quantity is abnormal, the operating status of the key component entity is judged to be attention, and an early warning is issued for the abnormal state quantity; if the correlation characteristics and distribution model characteristics of the state quantity are abnormal at the same time, the operating status of the key component entity is judged to be abnormal, and an early warning is issued for the abnormal state quantity.

[0013] This wind turbine status warning method constructs a digital twin model for the key parts of the wind turbine, and simulates and predicts the digital twin model based on the characteristics of the state quantity time series. It judges and warns the operating status of the key parts of the wind turbine from two aspects: the distribution characteristics of the state quantities and the correlation characteristics between the state quantities. It solves the technical problem that the current related technologies rely on status monitoring data and a large number of fault cases to build pattern recognition models when warning the operating status of wind turbines, ignore the impact of hidden latent defects of wind turbines on the judgment model, and the recognition model cannot consider the future operating trends of wind turbines, resulting in the inability to accurately judge and warn the operating status of wind turbines. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The accompanying drawings are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification, are used to explain the principles of the present disclosure. Obviously, the drawings described below are only some embodiments of the present disclosure, and those skilled in the art can derive other drawings based on these drawings without inventive effort. In the drawings:

[0015] Figure 1 This is a flowchart of a method for building a digital twin model of power equipment in an embodiment of the present invention.

[0016] Figure 2 This is a flowchart of a wind turbine gearbox status warning based on digital twins in an embodiment of the present invention.

[0017] Figure 3It is a comparison of the statistical characteristics of the initial digital twin model of the gearbox in the embodiment of the present invention and the statistical characteristics of the updated digital twin model.

[0018] Figure 4 It is a graph of original monitoring data and predicted data of oil temperature, amplitude, and wind speed in the initial digital twin model and the updated digital twin model of the gearbox in an embodiment of the present invention. DETAILED DESCRIPTION

[0019] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0020] like Figure 1 As shown, the embodiment of the present invention proposes a flow chart of a wind turbine status early warning method based on digital twins, including the following steps:

[0021] Step S102: Entity of key components of the wind turbine.

[0022] Step S104: construct an initial digital twin model based on the real-time state data of key parts and entities.

[0023] Step S106: Obtain statistical characteristics and correlation characteristics of the initial digital twin model.

[0024] Step S108: perform simulation prediction on the monitored state quantity and update the digital twin model based on the simulation prediction data.

[0025] Step S110: Obtain statistical features and correlation features of the updated digital twin model.

[0026] Step S112 , comparing the features of the updated digital twin model with the features of the initial digital twin model, determining the operating status of key components of the wind turbine and issuing an early warning.

[0027] Through the above steps, a digital twin model of the key parts of the wind turbine was constructed, and the digital twin model was simulated and predicted based on the characteristics of the state quantity time series. The judgment and early warning of the operating status of the key parts of the wind turbine were realized from two aspects: the distribution characteristics of the state quantities and the correlation characteristics between the state quantities.

[0028] The key components of the wind turbine include at least one of a gearbox, a generator and a blade. This embodiment uses the gearbox of the wind turbine as an example to illustrate the status warning process. The process is as follows: Figure 2 shown.

[0029] The initial digital twin model is constructed based on the real-time state quantity data of the key parts. In this embodiment, the state quantities selected to reflect the operation status of the gearbox are gearbox oil temperature, vibration amplitude and wind speed. The corresponding monitoring data are shown in Table 1.

[0030] Table 1

[0031] .

[0032] The steps for constructing the initial digital twin model are: obtaining the gearbox's size information, material information, structure information, manufacturer information, self-attribute information, and operating environment information to construct a static 3D model that can intuitively reflect the entities of key parts; digitally mapping the real-time acquired time series of oil temperature, amplitude, and wind speed to the static 3D model to form the initial digital twin model of the gearbox.

[0033] The acquisition of the statistical characteristics and correlation characteristics of the initial digital twin model includes: performing statistical analysis on the monitoring data of each type of state quantity in the initial digital twin model of the gearbox, obtaining the distribution model that the state quantity conforms to, and obtaining the key parameters of the distribution model. In this embodiment, the oil temperature, amplitude, and wind speed of the gearbox all conform to the normal distribution, such as Figure 3 As shown on the left, the parameters of the oil temperature, amplitude, and wind speed distribution model are shown in Table 2. Correlation analysis is performed on the time series of the oil temperature, amplitude, and wind speed state quantities of the initial digital twin model of the gearbox to obtain the correlation coefficients between the two state quantities, as shown in Table 3.

[0034] Table 2

[0035] .

[0036] Table 3

[0037] .

[0038] The simulated prediction of the monitored state variables and updating the digital twin model based on the simulated prediction data include predicting the oil temperature, amplitude, and wind speed time series of the initial digital twin model of the gearbox using a BP neural network. The predicted data are shown in Table 4. The predicted data is added to the original monitoring data to update the digital twin model.

[0039] Table 4

[0040] .

[0041] The obtaining of the statistical characteristics and correlation characteristics of the updated digital twin model includes: performing statistical analysis on the monitoring data of each type of state quantity in the updated digital twin model of the gearbox, obtaining the distribution model that the state quantity conforms to, and obtaining the key parameters of the distribution model. In this embodiment, the distribution model of oil temperature, amplitude, and wind speed in the updated digital twin model of the gearbox is as follows: Figure 3 As shown on the right, the corresponding distribution model parameters are shown in Table 5. The correlation analysis of the oil temperature, amplitude, and wind speed state quantity time series of the updated digital twin model of the gearbox is performed to obtain the correlation coefficients between the two state quantities, as shown in Table 6.

[0042] Table 5

[0043] .

[0044] Table 6

[0045] .

[0046] Comparing the characteristics of the updated digital twin model with the initial digital twin model to determine the operating status of key wind turbine components and issue a warning includes comparing the distribution model characteristics of each state variable in the updated digital twin model of the gearbox with the distribution model characteristics of each state variable in the initial digital twin model. If the distribution model characteristic changes by more than 30%, the state variable is determined to be abnormal. In this implementation, the distribution characteristic change rates of the initial digital twin model and the updated digital twin model in Tables 2 and 5 are compared, as shown in Table 7. The results in Table 7 indicate that the gearbox temperature is abnormal.

[0047] Table 7

[0048] .

[0049] Compare the correlation characteristics between the various state quantities in the updated digital twin model with those in the initial digital twin model. If the correlation characteristic changes by more than 30%, the state quantity is considered abnormal. In this embodiment, Table 8 compares the change rates of the correlation characteristics of the initial and updated digital twin models in Tables 3 and 6. The results in Table 8 indicate that the correlation between the temperature and vibration of the gearbox is abnormal.

[0050] Table 8

[0051] .

[0052] The specific steps for determining the operating status of a gearbox and issuing an early warning are as follows: if both the correlation characteristics and the distribution model characteristics of the state quantities are normal, the operating status of the key component is determined to be normal; if either the correlation characteristics or the distribution model characteristics of the state quantities is abnormal, the operating status of the key component entity is determined to be cautionary, and an early warning is issued for the abnormal state quantity; if both the correlation characteristics and the distribution model characteristics of the state quantities are abnormal, the operating status of the key component entity is determined to be abnormal, and an early warning is issued for the abnormal state quantity. In this embodiment, the temperature distribution characteristics are abnormal, and the correlation between temperature and vibration is abnormal. Therefore, the gearbox is determined to be operating abnormally, and early warnings are issued for both the temperature and vibration state quantities.

Claims

1. A wind turbine status early warning method based on digital twins, characterized in that: include: Entities of key components of wind turbines; Build an initial digital twin model based on the real-time status data of key entities; Obtain statistical characteristics and correlation properties of the initial digital twin model; Perform simulation predictions on monitored state quantities and update the digital twin model based on the simulation prediction data; Obtain updated statistical and correlation features of the digital twin model; Compare the characteristics of the updated digital twin model with those of the initial digital twin model to determine the operating status of key wind turbine components and issue early warnings; The key component entity of the wind turbine includes at least one of a gearbox, a generator and a blade; The construction of the initial digital twin model based on the real-time state data of key entities includes: The state quantity reflecting the operating condition of the gearbox of the key part includes but is not limited to at least two of the gearbox oil temperature, the gearbox bearing temperature, the ambient temperature, and the vibration amplitude; The state quantity reflecting the generator operation condition includes but is not limited to at least two of the following: generator speed, generator front shaft temperature, generator rear shaft temperature, generator temperature, A-phase current, B-phase current, C-phase current, AB-phase voltage, BC-phase voltage, CA-phase voltage, frequency, active power, reactive power, and power factor; The state quantity reflecting the blade operation condition includes but is not limited to at least two of wind speed, lightning, icing, stress, blade tip pressure, blade angle, torque, and blade average speed; The steps of constructing the initial digital twin model are: obtaining the size information, material information, structure information, manufacturer information, self-attribute information, and operating environment information of the key part entity to construct a static 3D model that can intuitively reflect the key part entity; digitally mapping the real-time status data of the key part entity obtained in real time to the static 3D model to form the initial digital twin model of the key part entity; Obtain statistical characteristics and correlation properties of the initial digital twin model, including: The statistical characteristics of the initial digital twin model are obtained by performing statistical analysis on the monitoring data of each type of state quantity in the digital twin model, obtaining the distribution model that the state quantity conforms to, and obtaining the key parameters of the distribution model; The distribution model includes but is not limited to at least one of normal distribution, Weibull distribution, and chi-square distribution; The correlation characteristics of the initial digital twin model are obtained by performing a correlation analysis on the time series of all state quantities reflecting the key part entities to obtain the correlation coefficients between any two state quantities.

2. The method according to claim 1, characterized in that Simulate and predict the monitored state quantities, and update the digital twin model based on the simulation prediction data, including: The simulation prediction of the monitoring state quantity is to predict the time series of all state quantities of the key part entities; The prediction method includes but is not limited to at least one of ARIMA prediction, BP neural network prediction, and LSTM prediction; The updating of the digital twin model based on simulation prediction data is to add the predicted data to the initial digital twin model of the key part entity.

3. The method according to claim 1, characterized in that Obtain updated statistical and correlation features of the digital twin model, including: The obtaining of the statistical features of the updated digital twin model is to perform statistical analysis on the monitoring data of each type of state quantity in the updated digital twin model, obtain the distribution model that the state quantity conforms to, and obtain the key parameters of the distribution model; The distribution model includes but is not limited to at least one of normal distribution, Weibull distribution, and chi-square distribution; The method for obtaining the correlation characteristics of the updated digital twin model is to perform correlation analysis on the time series of all state quantities reflecting the key part entities to obtain the correlation coefficients between any two state quantities.

4. The method according to claim 1, wherein Compare the characteristics of the updated digital twin model with the initial digital twin model to determine the operating status of key wind turbine components and issue early warnings, including: Compare the distribution model characteristics of each state quantity in the updated digital twin model with the distribution model characteristics of each state quantity in the initial digital twin model. If the distribution model characteristics change by more than 30%, the state quantity is judged to be abnormal; Compare the correlation characteristics between the various state quantities in the updated digital twin model with the correlation characteristics between the various state quantities in the initial digital twin model. If the correlation characteristics change by more than 30%, the state quantity is judged to be abnormal. The specific steps for judging the operating status of key components of the wind turbine and issuing early warnings are as follows: if the correlation characteristics and distribution model characteristics of the state quantity are both normal, the operating status of the key component is judged to be normal; if one of the correlation characteristics and distribution model characteristics of the state quantity is abnormal, the operating status of the key component entity is judged to be attention, and an early warning is issued for the abnormal state quantity; if the correlation characteristics and distribution model characteristics of the state quantity are abnormal at the same time, the operating status of the key component entity is judged to be abnormal, and an early warning is issued for the abnormal state quantity.

Images