A method and system for predicting tilt deformation of offshore wind towers

Through the combination of mechanism model pre-training and Bayesian convolutional neural network, the physical prediction model and sensor data are used to solve the accuracy of wind tower tilt and deformation prediction in offshore wind farms, and the prediction accuracy and safety are improved.

CN118708980BActive Publication Date: 2024-11-05NANTONG BLUE ISLAND OFFSHORE CO LTD
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202411206949.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-30
Publication Date
2024-11-05
Estimated Expiration
2044-08-30

AI Technical Summary

Technical Problem

The prior art is difficult to accurately predict the inclination and deformation of wind power towers in offshore wind farms, especially in special cases, the insufficient sample number leads to inaccurate models, affecting the performance and safety of wind turbines.

Method used

The mechanism model is used to generate coarse training data for pre-training. After reaching a stable state, it is carefully trained based on a small number of high-precision real samples. Combined with Bayesian convolutional neural network, it uses physical prediction models and sensor data to establish a tilt prediction model to improve prediction. Accuracy.

Benefits of technology

More accurate prediction of wind power tower inclination and deformation is achieved, the safety and maintenance efficiency of wind power tower are improved, and the consumption of manpower and financial resources is reduced.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118708980B_ABST
    Figure CN118708980B_ABST
Patent Text Reader

Abstract

The present invention discloses a method and system for predicting the inclination and deformation of an offshore wind tower, and belongs to the technical field of offshore wind tower monitoring. The present invention collects and analyzes inclination sensor data in real time, determines whether the size and change trend of the tower base inclination in the standby state exceeds a set threshold value to monitor the inclination of the wind tower, and obtains information on factors affecting the inclination and deformation of the offshore wind tower; establishes a physical prediction model, and realizes a preliminary prediction of the inclination of the wind tower through the physical prediction model, and determines a pre-training set and a pre-verification set based on the preliminary prediction results of the inclination of the wind tower; uses a Bayesian convolutional neural network to train based on the pre-training set and the pre-verification set to obtain a tilt prediction model, and predicts the inclination of the offshore wind tower through the tilt prediction model. The present invention can construct a more accurate tilt prediction model based on a small amount of high-precision real samples, and improves the accuracy of the prediction to a certain extent.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of offshore wind tower monitoring, and in particular to a method and system for predicting the inclination and deformation of an offshore wind tower. Background Art

[0002] Wind power is a renewable energy source with low development cost and can be developed on a large scale. With the rapid development of my country's economy and science and technology, the number and scale of wind turbines in my country are growing rapidly. Wind farms are generally set up in valleys or on the sea far away from human habitation. Compared with onshore wind power, offshore wind power has to withstand not only its own gravity, wind thrust, impeller torque and complex and variable loads, but also the harsh marine environment (such as salt spray corrosion, wave load, sea ice collision, typhoon and other factors). These factors often lead to the tilt of wind turbine towers, which in turn affects the performance and life of wind turbines.

[0003] Wind turbine towers are basically unattended, and real-time monitoring is not possible. When safety problems occur, management personnel cannot be notified immediately for maintenance. In addition, the equipment operation status of wind turbine towers is maintained by arranging regular manual inspections, which consumes a lot of manpower and financial resources. However, severe tilt or deformation of wind turbine towers will increase the risk of tower collapse.

[0004] In the prior art, there are also methods that use deep learning to predict the tilt of wind towers. However, due to the complex conditions at sea, there may be some special circumstances that affect the tilt of wind towers. The frequency of these special circumstances is low, resulting in a small number of samples obtained. This makes it impossible to use deep learning for prediction, or causes the trained model to be inaccurate, resulting in a low prediction degree of the wind tower tilt.

[0005] Therefore, it is necessary to monitor and predict the tilt and deformation state of wind turbine towers with higher accuracy. Summary of the invention

[0006] The purpose of the present invention is to provide a method and system for predicting the tilt deformation of offshore wind towers, which uses a mechanism model to generate coarse training data for pre-training. When the model reaches a relatively stable state, it is fine-trained based on a small amount of high-precision real samples to obtain a more accurate tilt prediction model.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] A method for predicting the tilt deformation of an offshore wind tower, the steps comprising:

[0009] Determine the influencing factors of offshore wind tower tilt and deformation and obtain the prediction data of the influencing factors of offshore wind tower tilt and deformation;

[0010] The tilt sensor data is collected and analyzed in real time to determine whether the tower base tilt angle and change trend in the standby state exceed the set threshold to monitor the wind tower tilt and obtain the influencing factor data of the offshore wind tower tilt deformation; the technology for analyzing the tilt sensor data belongs to the existing technology and will not be described in detail here;

[0011] Establish a physical prediction model, use the physical prediction model to achieve preliminary prediction of wind tower inclination based on the prediction data of offshore wind tower inclination deformation factors, and determine the pre-training set and pre-validation set based on the preliminary prediction results of wind tower inclination;

[0012] Based on the pre-training set and the pre-validation set, a Bayesian convolutional neural network is used to train a tilt prediction model, and the tilt prediction of offshore wind towers is achieved through the tilt prediction model.

[0013] According to the above technical solution, the offshore wind tower tilt deformation influencing factors include the angle between the wind tower tilt direction and the average wind direction, wind speed, and wave data. Wind direction is also one of the important influencing factors affecting the tilt of offshore wind towers. Different wind directions have different degrees of influence on the tilt of wind towers. Therefore, the angle between the wind tower tilt direction and the average wind direction is used as the offshore wind tower tilt deformation influencing factor.

[0014] According to the above technical solution, the physical prediction model :

[0015] ;

[0016] In the formula, y represents the inclination, It represents the angle between the inclination direction of the wind tower and the average wind direction, P is the coupling value of wind load and wave load, a, b, c are unknown coefficients, and t represents the time when the influencing factors of wind tower inclination deformation act.

[0017] According to the above technical solution, the coupling calculation steps of wind load and wave load include:

[0018] Determine the relationship between wind speed and significant wave height: , It means pulsating wind; It represents the effective wave height of ocean waves, and regards wind load as static wind (average wind) and dynamic wind (pulsating wind). The relationship between wind load and wave load is taken into account in the mechanism model, which can make the mechanism model more realistic.

[0019] Based on the relationship between wind speed and significant wave height The total wave hydrodynamic load on the wind tower is calculated using the Morison equation. The total wave hydrodynamic load can also be called the sea wave load. According to the Morison equation, the wave hydrodynamic load per unit height is It is expressed as:

[0020] ;

[0021] in, represents the density of seawater, represents the drag coefficient, represents the inertia coefficient, represents the horizontal velocity of seawater, is the horizontal acceleration of seawater, Indicates the diameter of the supporting structure pipe;

[0022] According to the effective wave height The total wave hydrodynamic load is determined by the wave hydrodynamic load per unit height; the wave load is the load caused by the interaction between the sea waves and the offshore wind tower support structure;

[0023] The wind load is the sum of the wind load on the tower, the wind load on the nacelle and the wind load on the blade as a whole;

[0024] Based on the coupling of the wind load and the wave load using weights, P is obtained.

[0025] Due to the complex offshore conditions, the proportion of wind load and wave load on the inclination of wind towers is different under different sea wind and wave conditions. Therefore, the proportion of wind load and wave load on the inclination of wind towers under similar scenarios can be calculated based on historical data to determine the weights of wind load and wave load.

[0026] ;

[0027] in, Indicates wind load The weight ratio of Indicates wave load The weight ratio of .

[0028] According to the above technical scheme, based on the prediction data of the influencing factors of the offshore wind tower inclination and deformation and the time when the influencing factors of the offshore wind tower inclination and deformation act on the wind tower, a physical prediction model is used to make a preliminary prediction of the inclination of the wind tower, and a pre-training set is constructed according to the prediction data of the influencing factors of the offshore wind tower inclination and deformation and the corresponding preliminary predicted values; the pre-training set is divided into a pre-training set and a pre-verification set.

[0029] According to the above technical solution, the tilt prediction model construction steps are:

[0030] Obtain the historical offshore wind tower tilt deformation influencing factor information and its corresponding tilt data to construct a precise validation set;

[0031] A Bayesian convolutional neural network is used to train a rough tilt prediction model based on a pre-training set; and the rough tilt prediction model is corrected and adjusted based on a pre-validation set;

[0032] The trained coarse tilt prediction model is corrected and adjusted based on the precise validation set to obtain the tilt prediction model, in which the determination coefficient, root mean square error and Kling-Gupta efficiency coefficient are used as discrimination criteria to determine whether the tilt prediction model or the coarse tilt prediction model meets the forecast accuracy. If the forecast accuracy is not met, the tilt prediction model or the coarse tilt prediction model is continuously trained and optimized until the forecast accuracy is met.

[0033] According to the above technical solution, when generating each offshore wind tower tilt prediction value, some neurons of the trained tilt prediction model are randomly discarded according to probability, and then an offshore wind tower tilt prediction value is obtained based on the remaining neurons.

[0034] According to the above technical solution, the angle between the wind tower inclination direction and the average wind direction, the wind speed, and the wave data in the offshore wind tower inclination deformation influencing factor prediction data are obtained by predicting the offshore wind and waves.

[0035] An offshore wind tower tilt deformation prediction system, comprising:

[0036] An influencing factor acquisition module is used to determine the influencing factors of the offshore wind tower tilt deformation and obtain the prediction data of the influencing factors of the offshore wind tower tilt deformation;

[0037] The sensor data acquisition module collects and analyzes the inclination sensor data in real time to determine whether the tower base inclination and its change trend exceed the set threshold in the standby state to monitor the inclination of the wind tower and obtain information on factors affecting the inclination and deformation of the offshore wind tower;

[0038] The mechanism model building module builds a physical prediction model, and realizes the preliminary prediction of wind tower inclination through the physical prediction model based on the prediction data of offshore wind tower inclination deformation influencing factors, and determines the pre-training set and pre-verification set based on the preliminary prediction results of wind tower inclination;

[0039] The tilt prediction module uses a Bayesian convolutional neural network to train a tilt prediction model based on a pre-training set and a pre-validation set, and predicts the tilt of offshore wind towers through the tilt prediction model.

[0040] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: the present invention uses a mechanism model to generate coarser training data for pre-training, and when the model reaches a relatively stable state, it performs fine training based on a small amount of high-precision real samples to obtain a more accurate tilt prediction model. For some special cases with less observation data, deep learning methods can also be used to predict the tilt of wind towers, which also improves the accuracy of wind tower tilt prediction to a certain extent. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0042] Figure 1 It is a flow chart of the steps of a method for predicting the tilt deformation of an offshore wind tower according to the present invention. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments 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 creative work are within the scope of protection of the present invention.

[0044] Embodiment 1: A method for predicting the tilt deformation of an offshore wind tower, comprising the following steps:

[0045] S1. Determine the influencing factors of offshore wind tower inclination deformation and obtain the prediction data of the influencing factors of offshore wind tower inclination deformation; the influencing factors of offshore wind tower inclination deformation include the angle between the inclination direction of the wind tower and the average wind direction, the wind speed, and the wave data; the angle between the inclination direction of the wind tower and the average wind direction, the wind speed, and the wave data in the prediction data of the inclination deformation influencing factors of offshore wind tower are obtained by predicting the offshore wind and waves.

[0046] S2. Collect and analyze the inclination sensor data in real time to determine whether the inclination angle and change trend of the tower base in the standby state exceed the set threshold to monitor the inclination of the wind tower and obtain information on factors affecting the inclination and deformation of the offshore wind tower.

[0047] S3. Establish a physical prediction model, use the physical prediction model to achieve preliminary prediction of wind tower inclination, determine the pre-training set and pre-verification set based on the preliminary prediction results of wind tower inclination, physical prediction model:

[0048] ;

[0049] In the formula, y represents the inclination, It represents the angle between the inclination direction of the wind tower and the average wind direction, P is the coupling value of wind load and wave load, a, b, c are unknown coefficients, and t represents the time when the influencing factors of wind tower inclination deformation act.

[0050] The coupling calculation steps of wind load and wave load include:

[0051] Determine the relationship between wind speed and significant wave height: , It means pulsating wind; It represents the effective wave height of ocean waves, and regards wind load as static wind (average wind) and dynamic wind (pulsating wind). The relationship between wind load and wave load is taken into account in the mechanism model, which can make the mechanism model more realistic.

[0052] Based on the relationship between wind speed and significant wave height The total wave hydrodynamic load on the wind tower is calculated using the Morison equation. The total wave hydrodynamic load can also be called the sea wave load. According to the Morison equation, the wave hydrodynamic load per unit height is It is expressed as: ;

[0053] in, represents the density of seawater, represents the drag coefficient, represents the inertia coefficient, represents the horizontal velocity of seawater, is the horizontal acceleration of seawater, Indicates the diameter of the supporting structure pipe;

[0054] According to the effective wave height The total wave hydrodynamic load is determined by the wave hydrodynamic load per unit height, and the wind load is calculated; the wind load is the sum of the wind load on the tower, the wind load on the nacelle and the wind load on the entire blade;

[0055] According to historical data, the influence of wind load and wave load on wind tower inclination in similar scenarios is 0.8 and 0.2 respectively; .

[0056] S4. Based on the pre-training set and the pre-verification set, a Bayesian convolutional neural network is used to train a tilt prediction model, and the tilt prediction of the offshore wind tower is realized through the tilt prediction model, specifically including:

[0057] Obtain the historical offshore wind tower tilt deformation influencing factor information and its corresponding tilt data to construct a precise validation set;

[0058] A Bayesian convolutional neural network is used to train a coarse tilt prediction model based on a pre-training set, and the pre-validation set is input into the coarse tilt prediction model. Then, the coefficient of determination, root mean square error, and Kling-Gupta efficiency coefficient are calculated. The coarse tilt prediction model is corrected and adjusted according to the coefficient of determination, root mean square error, and Kling-Gupta efficiency coefficient until the prediction accuracy is met.

[0059] The refined validation set is input into a coarse tilt prediction model that meets the forecast accuracy, and then the determination coefficient, root mean square error and Kling-Gupta efficiency coefficient are calculated. The coarse tilt prediction model is corrected and adjusted according to the determination coefficient, root mean square error and Kling-Gupta efficiency coefficient until the forecast accuracy is met to obtain the tilt prediction model.

[0060] When generating each offshore wind tower tilt prediction value, some neurons of the trained tilt prediction model are randomly discarded according to probability, and then an offshore wind tower tilt prediction value is obtained based on the remaining neurons.

[0061] Embodiment 2: A system for predicting the tilt and deformation of an offshore wind tower, comprising:

[0062] An influencing factor acquisition module is used to determine the influencing factors of the offshore wind tower tilt deformation and obtain the prediction data of the influencing factors of the offshore wind tower tilt deformation;

[0063] The sensor data acquisition module collects and analyzes the inclination sensor data in real time to determine whether the tower base inclination and its change trend exceed the set threshold in the standby state to monitor the inclination of the wind tower and obtain information on factors affecting the inclination and deformation of the offshore wind tower;

[0064] The mechanism model building module builds a physical prediction model, and based on the prediction data of offshore wind tower tilt deformation influencing factors, the physical prediction model is used to achieve a preliminary prediction of wind tower tilt, obtain the prediction data of offshore wind tower tilt deformation influencing factors, and determine the pre-training set and pre-verification set based on the preliminary prediction results of wind tower tilt;

[0065] The tilt prediction module uses a Bayesian convolutional neural network to train a tilt prediction model based on a pre-training set and a pre-validation set, and predicts the tilt of offshore wind towers through the tilt prediction model.

[0066] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0067] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for predicting the tilt deformation of an offshore wind tower, characterized in that: The prediction methods include: Determine the influencing factors of offshore wind tower tilt and deformation and obtain the prediction data of the influencing factors of offshore wind tower tilt and deformation; Collect and analyze the tilt sensor data in real time to determine whether the tower base tilt angle and change trend exceed the set threshold in the standby state to monitor the wind tower tilt and obtain the influencing factor data of the offshore wind tower tilt deformation; A physical prediction model is established, and a preliminary prediction of wind tower inclination is achieved through the physical prediction model based on the prediction data of the influencing factors of offshore wind tower inclination deformation, and a pre-training set and a pre-verification set are determined based on the preliminary prediction results of wind tower inclination; the physical prediction model: ; In the formula, y represents the inclination, It represents the angle between the tilt direction of the wind tower and the average wind direction, P is the coupling value of wind load and wave load, a, b, c are unknown coefficients, and t represents the time when the influencing factor of wind tower tilt deformation acts; The coupling calculation steps of the wind load and the wave load include: Determine the relationship between wind speed and significant wave height : , It means pulsating wind; Based on the relationship between wind speed and significant wave height The wave hydrodynamic load per unit height is calculated using the Morison equation to obtain the total wave hydrodynamic load, and the wind load is calculated; the total wave hydrodynamic load may also be referred to as the sea wave load; wherein, according to the Morison equation, the wave hydrodynamic load per unit height is It is expressed as: ; In the formula, represents the density of seawater, represents the drag coefficient, represents the inertia coefficient, represents the horizontal velocity of seawater, is the horizontal acceleration of seawater, Indicates the diameter of the supporting structure pipe; The wind load is the sum of the wind load on the tower, the wind load on the nacelle and the wind load on the blade as a whole; The proportion of the influence of wind load and wave load on the inclination of the wind tower in similar scenes is used as the corresponding weight of wind load and wave load, and then the wind load and the wave load are coupled by using the weight to obtain P; ; in, Indicates wind load The weight ratio of Indicates wave load The weight ratio of Based on the pre-training set and the pre-validation set, a Bayesian convolutional neural network is used to train a tilt prediction model, and the tilt prediction of offshore wind towers is achieved through the tilt prediction model.

2. A method for predicting the inclination deformation of an offshore wind tower according to claim 1, characterized in that: The influencing factors of the offshore wind tower tilt deformation include the angle between the wind tower tilt direction and the average wind direction, the wind speed, and the wave data.

3. The method for predicting the inclination deformation of an offshore wind tower according to claim 1, characterized in that: The angle between the wind tower tilt direction and the average wind direction, the wind speed, and the sea wave data in the offshore wind tower tilt deformation influencing factor prediction data are obtained by predicting offshore wind and waves.

4. The method for predicting the tilt deformation of an offshore wind tower according to claim 1, characterized in that: Based on the prediction data of offshore wind tower inclination and deformation influencing factors and the time when the offshore wind tower inclination and deformation influencing factors act on the wind tower, a physical prediction model is used to make a preliminary prediction of the wind tower inclination, and a pre-training set is constructed according to the prediction data of offshore wind tower inclination and deformation influencing factors and their corresponding preliminary predicted values; the pre-training set is divided into a pre-training set and a pre-validation set.

5. The offshore wind tower tilt deformation prediction method according to claim 1, characterized in that: The tilt prediction model construction steps are: Obtain the historical offshore wind tower tilt deformation influencing factor information and its corresponding tilt data to construct a precise validation set; A Bayesian convolutional neural network is used to train a rough tilt prediction model based on a pre-training set; and the rough tilt prediction model is corrected and adjusted based on a pre-validation set; The trained rough tilt prediction model is corrected and adjusted based on the refined validation set to obtain the tilt prediction model.

6. The method for predicting the tilt deformation of an offshore wind tower according to claim 1, characterized in that: When generating each offshore wind tower tilt prediction value, some neurons of the trained tilt prediction model are randomly discarded according to probability, and then an offshore wind tower tilt prediction value is obtained based on the remaining neurons.

7. An offshore wind tower tilt deformation prediction system, characterized in that: include: An influencing factor acquisition module is used to determine the influencing factors of the offshore wind tower tilt deformation and obtain the prediction data of the influencing factors of the offshore wind tower tilt deformation; The sensor data acquisition module collects and analyzes the inclination sensor data in real time to determine whether the tower base inclination angle and its change trend exceed the set threshold in the standby state to monitor the inclination of the wind tower and obtain the influencing factor data of the offshore wind tower inclination and deformation; The mechanism model building module builds a physical prediction model, realizes the preliminary prediction of wind tower inclination through the physical prediction model based on the prediction data of the influencing factors of offshore wind tower inclination deformation, and determines the pre-training set and pre-verification set based on the preliminary prediction results of wind tower inclination; the physical prediction model: ; In the formula, y represents the inclination, It represents the angle between the tilt direction of the wind tower and the average wind direction, P is the coupling value of wind load and wave load, a, b, c are unknown coefficients, and t represents the time when the influencing factor of wind tower tilt deformation acts; The mechanism model building module performs coupling calculation steps of wind load and wave load including: Determine the relationship between wind speed and significant wave height : , It means pulsating wind; Based on the relationship between wind speed and significant wave height The wave hydrodynamic load per unit height is calculated using the Morison equation to obtain the total wave hydrodynamic load, and the wind load is calculated; the total wave hydrodynamic load may also be referred to as the sea wave load; wherein, according to the Morison equation, the wave hydrodynamic load per unit height is It is expressed as: ; In the formula, represents the density of seawater, represents the drag coefficient, represents the inertia coefficient, represents the horizontal velocity of seawater, is the horizontal acceleration of seawater, Indicates the diameter of the supporting structure pipe; The wind load is the sum of the wind load on the tower, the wind load on the nacelle and the wind load on the entire blade; The proportion of the influence of wind load and wave load on the inclination of the wind tower in similar scenes is used as the corresponding weight of wind load and wave load, and then the wind load and the wave load are coupled by using the weight to obtain P; ; in, Indicates wind load The weight ratio of Indicates wave load The weight ratio of The tilt prediction module uses a Bayesian convolutional neural network to train a tilt prediction model based on a pre-training set and a pre-validation set, and predicts the tilt of offshore wind towers through the tilt prediction model.

Citation Information

Patent Citations

  • Wind power tower overturning intelligent diagnosis method based on inclination angle sensor information

    CN112560916A

  • Digital twin hydraulic engineering operation safety monitoring system and operation method

    CN115688227A

  • Telegraph pole tower inclination monitoring method based on rotating target detection

    CN116503798A

  • Tower footing monitoring method and device, computer equipment, storage medium and product

    CN116716927A

  • Methods and systems for detection in an industrial internet of things data collection environment with large data sets

    US20180284737A1