Wind turbine blade root load prediction method, device, equipment and medium

Abstract

The application relates to the technical field of wind turbine units, and particularly provides a wind turbine unit blade root load prediction method, device, equipment and medium, and aims to solve the technical problem that when the load of a blade root is too large, the operation parameters of the wind turbine unit cannot be adjusted in time, the blade root is overloaded, and the blade is damaged. To this end, the application comprises the following steps: acquiring blade root load data and wind turbine unit parameters; inputting the blade root load data and the wind turbine unit parameters into a preset blade root load prediction model to obtain predicted blade root loads within a preset time; and adjusting and processing the operation parameters of the wind turbine unit according to the predicted blade root loads.

Description

Technical Field

[0001] This application relates to the field of wind turbine technology, specifically providing a method, device, equipment, and medium for predicting blade root loads of wind turbines. Background Technology

[0002] Currently, as wind turbines are developing towards larger sizes, the aspect ratio and structural toughness of wind turbine blades have increased dramatically. This can cause stall flutter in ultra-long flexible blades, leading to blade damage.

[0003] In existing technologies, sensors installed at the blade root collect load data, and then the operating parameters of the wind turbine are adjusted based on the load data and the wind turbine's operating data, which can ensure the normal operation of the wind turbine to a certain extent. However, when the load at the blade root is too large, the inability to adjust the wind turbine's operating parameters in time leads to overload operation at the blade root, resulting in blade damage.

[0004] Accordingly, there is a need in the field for a new method for predicting blade root loads of wind turbines to address the above-mentioned problems. Summary of the Invention

[0005] In order to overcome the above-mentioned defects, this application is made to provide a solution or at least a partial solution to the technical problem in the prior art that when the load on the blade root is too large, the operating parameters of the wind turbine cannot be adjusted in time, resulting in overload operation of the blade root and thus blade damage.

[0006] In a first aspect, this application provides a method for predicting blade root loads of a wind turbine, comprising:

[0007] Acquire blade root load data and wind turbine parameters;

[0008] The blade root load data and the wind turbine parameters are input into a preset blade root load prediction model to obtain the predicted blade root load within a preset time period; wherein, the predicted blade root load includes the average blade root load, the ultimate blade root load, the number of cycles, and the predicted equivalent fatigue load.

[0009] Based on the predicted blade root load, the operating parameters of the wind turbine are adjusted.

[0010] The process of obtaining the preset leaf root load prediction model includes:

[0011] Acquire training data; wherein the training data includes blade root load and wind turbine data;

[0012] When training the i-th decision tree, a data group is selected from the training data, and a preset number of feature variables are selected from the data group;

[0013] Based on the preset number of feature variables, the split point of the i-th decision tree is divided to obtain a node, and it is determined whether the node meets the stopping splitting condition.

[0014] If the stopping partitioning condition is satisfied, then the i-th decision tree is obtained;

[0015] Training stops when i = I; where I is the number of decision trees.

[0016] The obtained decision trees are combined to obtain the preset leaf root load prediction model.

[0017] In one technical solution of the above-mentioned method for predicting blade root load of a wind turbine, obtaining the blade root load includes:

[0018] Acquire the swing strain signal and flapping strain signal of the leaf root;

[0019] Based on the swing strain signal and the flapping strain signal, the blade swing torque and the blade flapping torque are obtained;

[0020] Based on the blade oscillation moment and the blade flapping moment, the maximum load, average load, and number of load cycles are obtained;

[0021] The equivalent fatigue load is obtained based on the maximum load value and the number of load cycles.

[0022] The equivalent fatigue load, the average load, and the maximum load value are combined to obtain the blade root load.

[0023] In one technical solution of the above-mentioned method for predicting blade root loads of wind turbines, obtaining the wind turbine data includes:

[0024] Obtain the initial parameters of the wind turbine;

[0025] The initial parameters of the wind turbine are filtered to obtain the wind turbine data.

[0026] In one technical solution of the above-mentioned method for predicting blade root loads of a wind turbine, the step of filtering the initial parameters of the wind turbine to obtain the wind turbine data includes:

[0027] The correlation analysis of multiple parameters in the initial parameters of the wind turbine with the blade root load is performed to obtain a set of correlation coefficients;

[0028] The multiple correlation coefficients in the set of correlation coefficients are compared with a preset correlation coefficient threshold.

[0029] Obtain a subset of correlation coefficients greater than the corresponding preset correlation coefficient threshold;

[0030] Match the subset of correlation coefficients with the initial parameters of the wind turbine to obtain wind turbine data.

[0031] In a technical solution of the above method for predicting the root load of a wind turbine, after combining the obtained multiple decision trees to obtain the preset root load prediction model, the method further includes:

[0032] Obtain the true root load;

[0033] According to the true root load and the predicted root load, obtain the accurate value of the preset root load prediction model;

[0034] If it is determined that the accurate value is lower than the preset accurate threshold, screen the training data to obtain a screened data set;

[0035] According to the screened data set, obtain a root load prediction update model.

[0036] In a technical solution of the above method for predicting the root load of a wind turbine, after inputting the root load data and the wind turbine parameters into a preset root load prediction model to obtain the predicted root load within a preset time, the method further includes:

[0037] In the i-th iteration, according to the root ultimate load and the number of cycles, obtain the calculated value of the equivalent fatigue load;

[0038] According to the predicted equivalent fatigue load and the calculated value of the equivalent fatigue load, obtain the equivalent fatigue load error value;

[0039] Judge whether the equivalent fatigue load error value is greater than the preset error value;

[0040] Determine that the equivalent fatigue load error value is greater than the preset error value, and retrain the (i - 1)-th root load prediction model to obtain the i-th root load prediction model; where when i = 1, the (i - 1)-th root load prediction model is the preset root load prediction model;

[0041] Where, 0 < i ≤ I, and I is the number when the equivalent fatigue load error value is less than the preset error value to obtain the I-th preset root load prediction model.

[0042] In a technical solution of the above method for predicting the root load of a wind turbine, the method further includes:

[0043] Determine that the adjustment process of the operating parameters of the wind turbine is completed;

[0044] Monitor leaf root load data;

[0045] If the blade root load data is determined to be greater than the preset blade root load, the wind turbine unit will be shut down.

[0046] Secondly, this application provides a wind turbine blade root load prediction device, comprising:

[0047] The acquisition module is used to acquire blade root load data and wind turbine parameters;

[0048] The processing module is used to input the blade root load data and the wind turbine parameters into a preset blade root load prediction model to obtain the predicted blade root load within a preset time period; wherein, the predicted blade root load includes the average blade root load, the ultimate blade root load, the number of cycles, and the predicted equivalent fatigue load.

[0049] The analysis module is used to adjust the operating parameters of the wind turbine based on the predicted blade root load.

[0050] The process of obtaining the preset leaf root load prediction model includes:

[0051] Acquire training data; wherein the training data includes blade root load and wind turbine data;

[0052] When training the i-th decision tree, a data group is selected from the training data, and a preset number of feature variables are selected from the data group;

[0053] Based on the preset number of feature variables, the split point of the i-th decision tree is divided to obtain a node, and it is determined whether the node meets the stopping splitting condition.

[0054] If the stopping partitioning condition is satisfied, then the i-th decision tree is obtained;

[0055] Training stops when i = I; where I is the number of decision trees.

[0056] The obtained decision trees are combined to obtain the preset leaf root load prediction model.

[0057] Thirdly, this application provides a leaf root load prediction device, including a processor and a storage device, the storage device being adapted to store a plurality of program codes, the program codes being adapted to be loaded and run by the processor to perform the method as described in any one of the first aspects.

[0058] Fourthly, this application provides a computer-readable storage medium storing a plurality of program codes adapted to be loaded and run by a processor to perform the method as described in any one of the first aspects.

[0059] This application provides a method, apparatus, device, and medium for predicting blade root loads of wind turbine generators. The method specifically includes: acquiring blade root load data and wind turbine generator parameters; inputting the blade root load data and wind turbine generator parameters into a preset blade root load prediction model to obtain predicted blade root loads within a preset time period; wherein the predicted blade root loads include average blade root load, ultimate blade root load, number of cycles, and predicted equivalent fatigue load; and adjusting the operating parameters of the wind turbine generator based on the predicted blade root loads; wherein acquiring the preset blade root load prediction model includes: acquiring training data; wherein the training data includes blade root loads and... Wind turbine data; when training the i-th decision tree, a data group is selected from the training data, and a preset number of feature variables are selected from the data group; according to the preset number of feature variables, the split points of the i-th decision tree are divided to obtain nodes, and it is determined whether the nodes meet the stopping splitting condition; if it is determined that the stopping splitting condition is met, the i-th decision tree is obtained; when it is determined that i=I, training is stopped; where I is the number of decision trees; the obtained multiple decision trees are combined to obtain the preset blade root load prediction model, so that the blades operate within the normal load range, thereby ensuring the normal operation of the wind turbine. Attached Figure Description

[0060] The disclosure of this application will become more readily understood with reference to the accompanying drawings. It will be readily understood by those skilled in the art that these drawings are for illustrative purposes only and are not intended to limit the scope of protection of this application. Furthermore, similar numbers in the drawings are used to denote similar components, wherein:

[0061] Figure 1 A flowchart illustrating an embodiment of a wind turbine blade root load prediction method provided in this application.

[0062] Figure 2 A flowchart illustrating a second embodiment of a wind turbine blade root load prediction method provided in this application.

[0063] Figure 3 A flowchart illustrating a third embodiment of a wind turbine blade root load prediction method provided in this application;

[0064] Figure 4 A flowchart illustrating a method for predicting blade root loads of a wind turbine generator according to an embodiment of this application;

[0065] Figure 5 A flowchart illustrating a fifth embodiment of a wind turbine blade root load prediction method provided in this application.

[0066] Figure 6This is a flowchart illustrating a method for predicting blade root loads of a wind turbine generator, as provided in this application.

[0067] Figure 7 A flowchart illustrating Embodiment Seven of a method for predicting blade root loads of a wind turbine provided in this application;

[0068] Figure 8 A flowchart illustrating an eighth embodiment of a wind turbine blade root load prediction method provided in this application.

[0069] Figure 9 A schematic diagram of the structure of a wind turbine blade root load prediction device according to an embodiment of this application;

[0070] Figure 10 This is a schematic diagram of a leaf root load prediction device according to an embodiment of this application.

[0071] List of reference numerals :

[0072] 11: Acquisition module; 12: Processing module; 13: Analysis module; 21: Processor; 22: Memory. Detailed Implementation

[0073] Some embodiments of this application are described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of this application and are not intended to limit the scope of protection of this application.

[0074] In the description of this application, "module" and "processor" can include hardware, software, or a combination of both. A module can include hardware circuitry, various suitable sensors, communication ports, memory, and can also include software components, such as program code, or a combination of software and hardware. A processor can be a central processing unit, microprocessor, image processor, digital signal processor, or any other suitable processor. The processor has data and / or signal processing capabilities. The processor can be implemented in software, in hardware, or a combination of both. Non-transitory computer-readable storage media includes any suitable medium capable of storing program code, such as magnetic disks, hard disks, optical disks, flash memory, read-only memory, random access memory, etc. The term "A and / or B" means all possible combinations of A and B, such as only A, only B, or A and B. The terms "at least one A or B" or "at least one of A and B" have a similar meaning to "A and / or B" and can include only A, only B, or A and B. The singular terms "a" or "this" can also include plural forms.

[0075] In existing technologies, the load at the root of wind turbine blades is monitored in real time to obtain real-time data. Operators then send control commands to the wind turbine based on this data to ensure its normal operation. However, when the load at the blade root is too high, the inability to adjust the wind turbine's operating parameters in a timely manner leads to overloading of the blade root, resulting in blade damage.

[0076] Based on this, in order to solve the above-mentioned technical problems, this application provides a method for predicting the blade root load of a wind turbine, so as to adjust the operating parameters of the wind turbine according to the prediction results of the blade root, thereby ensuring that the blade operates within the normal load range.

[0077] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0078] Figure 1 This is a flowchart illustrating an embodiment of a wind turbine blade root load prediction method provided in this application. Figure 1 As shown, specifically, the method includes:

[0079] Step S101: Obtain blade root load data and wind turbine parameters.

[0080] In this embodiment, an electrical signal from the leaf root is acquired by a sensor at the leaf root, and then the electrical signal is converted into leaf root load data.

[0081] In this embodiment, the wind turbine parameters include environmental parameters, blade parameters, operating parameters, and load parameters. Environmental parameters include wind speed, wind direction, temperature, and humidity; blade parameters include blade size and shape; operating parameters include blade rotational speed, pitch angle, and rotor tilt angle; and load parameters include bending moment, shear force, and torque at the blade root.

[0082] Step S102: Input the blade root load data and wind turbine parameters into the preset blade root load prediction model to obtain the predicted blade root load within a preset time period.

[0083] In this embodiment, the predicted blade root load includes the average blade root load, the ultimate blade root load, the number of cycles, and the predicted equivalent fatigue load.

[0084] In this embodiment, the current blade root load data and the current wind turbine parameters are input into the preset blade root load prediction model to obtain the predicted blade root model within a preset time period. The preset blade root load prediction model can be obtained by training the data.

[0085] Step S103: Adjust the operating parameters of the wind turbine based on the predicted blade root load.

[0086] In this embodiment, by analyzing the predicted blade root load, the blade speed adjustment value, the pitch angle adjustment value, and the rotor tilt angle adjustment value are obtained, and the blade speed, pitch angle, and rotor tilt angle are adjusted according to the blade speed adjustment value, the pitch angle, and the rotor tilt angle.

[0087] In this embodiment, blade root load data and wind turbine parameters are acquired; the blade root load data and wind turbine parameters are input into a preset blade root load prediction model to obtain the predicted blade root load within a preset time period; based on the predicted blade root load, the operating parameters of the wind turbine are adjusted. Compared with the prior art, which cannot adjust the operating parameters of the wind turbine in a timely manner, leading to blade root overload and subsequent blade damage, this application ensures that the blades will not operate under overload by inputting the acquired blade root load data and wind turbine parameters into a preset blade root load prediction model to obtain the predicted blade root load within a preset time period, and adjusts the operating parameters of the wind turbine based on the predicted blade root load, thereby ensuring the normal operation of the wind turbine.

[0088] Figure 2 This is a flowchart illustrating a second embodiment of a wind turbine blade root load prediction method provided in this application. Figure 2 As shown, specifically, obtaining the preset leaf root load prediction model in step S102 includes:

[0089] Step S201: Obtain training data; the training data includes blade root load and wind turbine data.

[0090] Step S202: When training the i-th decision tree, select a data group from the training data and select a preset number of feature variables from the data group.

[0091] In this embodiment, all training data are treated as a data group, and a preset number of feature variables are randomly selected from the data group.

[0092] Step S203: Based on a preset number of feature variables, divide the split point of the i-th decision tree to obtain nodes, and determine whether the nodes meet the stop splitting condition.

[0093] In this embodiment, for each feature variable, a split point is randomly selected for partitioning, generating two nodes, and it is determined whether the two nodes meet the stopping partitioning condition.

[0094] In this embodiment, the m-th split point is divided according to Formula 1 and Formula 2:

[0095] (1)

[0096] (2)

[0097] Obtain subset Q m l (θ) and Q m r (θ) subset. Where θ=(j,t) m ) represents the partitioning combination, j represents the partitioning feature, corresponding to different feature variables in the array, and t represents the partitioning combination. m The feature threshold is randomly assigned; x is the input data; y is the output data; Q is the output data. m This is the dataset on node m.

[0098] In this embodiment, the stopping condition is that the entropy value of the subset is minimized.

[0099] In this embodiment, according to formula 3:

[0100] (3)

[0101] The entropy of the subset is obtained, where pi is the proportion of the i-th type of feature.

[0102] Step S204: If the stopping partitioning condition is met, the i-th decision tree is obtained.

[0103] In this embodiment, a decision tree is obtained after determining that the stopping partitioning condition is met.

[0104] Step S205: When i=I is determined, stop training.

[0105] In this embodiment, I represents the number of decision trees, which is a preset value.

[0106] In this embodiment, for example, the number of decision trees is 150.

[0107] In this embodiment, steps S201 to S204 are executed cyclically to train multiple decision trees.

[0108] Step S206: Combine the obtained decision trees to obtain the preset leaf root load prediction model.

[0109] In this embodiment, multiple decision trees are combined to obtain an extreme random forest, which is a preset leaf root load prediction model.

[0110] In this embodiment, training data is acquired; when training the i-th decision tree, a data group is selected from the training data, and a preset number of feature variables are selected from the data group; based on the preset number of feature variables, multiple nodes of the i-th decision tree are partitioned to obtain nodes, and it is determined whether the nodes meet the stopping partitioning condition; if it is determined that the stopping partitioning condition is met, the i-th decision tree is obtained; when it is determined that i=1, training is stopped; the multiple decision trees obtained are combined to obtain a preset leaf root load prediction model, and then the preset leaf root load prediction model is obtained.

[0111] Figure 3 This is a flowchart illustrating a third embodiment of a wind turbine blade root load prediction method provided in this application. Based on the above embodiments, as... Figure 3 As shown, specifically, obtaining the blade root load in step S201 includes:

[0112] Step S301: Obtain the swing strain signal and flapping strain signal of the leaf root.

[0113] In this embodiment, the swing strain signal and the wave strain signal are collected by sensors.

[0114] In this embodiment, under wind conditions where the ratio of wind load to gravity load is small, the gravity method is used to allow the wind turbine to idle or the blades to be fixed in a horizontal position under different blade pitch angles, and the sensor is placed under a typical gravitational torque, thereby achieving the calibration of the sensor position.

[0115] Step S302: Based on the swing strain signal and the flapping strain signal, obtain the blade swing torque and the blade flapping torque.

[0116] In this embodiment, according to formula 4:

[0117] (4)

[0118] Obtain the blade oscillation torque M be and blade flapping torque M bf Among them, D 11 D represents the calibration coefficient between the oscillation strain signal and the oscillation torque. 12 D represents the calibration coefficient between the swing strain signal and the swing moment. 21 D represents the calibration coefficient between the oscillation strain signal and the oscillation torque. 22 S represents the calibration coefficient between the swing strain signal and the swing torque. e Indicates the pendulum strain signal; S f Indicates a waving response signal; O e This indicates the offset in the direction of the oscillation, representing the output offset of the strain gauge in the oscillation direction when no external torque is applied. fThis indicates the offset in the swing direction, representing the output offset of the strain gauge in the swing direction when no external torque is applied.

[0119] Step S303: Based on the blade oscillation torque and blade flapping torque, obtain the maximum load, average load, and number of load cycles.

[0120] In this embodiment, the rainflow counting method is used to obtain the maximum load, average load, and load cycle number based on the blade oscillation torque and blade flapping torque.

[0121] Step S304: Obtain the equivalent fatigue load based on the maximum load value and the number of load cycles.

[0122] In this embodiment, according to formula 5:

[0123] (5)

[0124] The equivalent fatigue load L is obtained eq Among them, L i m n represents the m-th power of the maximum value of the i-th load level; i N represents the number of load cycles for the i-th stage; eq The number of load cycles is represented by ; m is the slope of the material stress versus life curve.

[0125] Step S305: Combine the equivalent fatigue load, the average load, and the maximum load to obtain the blade root load.

[0126] In this embodiment, the equivalent fatigue load, average load, and maximum load are combined into a single data point to obtain the blade root load.

[0127] In this embodiment, for example, the blade root load is {equivalent fatigue load; average load; maximum load}.

[0128] In this embodiment, the flapping strain signal and the oscillation strain signal of the blade root are acquired; based on the flapping strain signal and the oscillation strain signal, the blade flapping torque and the blade oscillation torque are obtained; based on the blade flapping torque and the blade oscillation torque, the maximum load, the average load, and the number of load cycles are obtained; based on the maximum load and the number of load cycles, the equivalent fatigue load is obtained; the equivalent fatigue load, the average load, and the maximum load are combined to obtain the blade root load, and thus the accurate value of the blade root load can be obtained.

[0129] Figure 4 This is a flowchart illustrating Embodiment 4 of a wind turbine blade root load prediction method provided in this application. Figure 4 As shown, specifically, obtaining the wind turbine data in step S201 includes:

[0130] Step S401: Obtain the initial parameters of the wind turbine.

[0131] Step S402: Filter the initial parameters of the wind turbine to obtain the wind turbine data.

[0132] In this embodiment, the minimum and maximum values ​​in the initial parameters of the wind turbine are deleted to obtain the wind turbine data.

[0133] In this embodiment, the initial parameters of the wind turbine are obtained; the initial parameters of the wind turbine are then filtered to obtain the wind turbine data.

[0134] Figure 5 This is a flowchart illustrating Embodiment 5 of a wind turbine blade root load prediction method provided in this application. Figure 5 As shown, a specific implementation of step S402 includes:

[0135] Step S501: Perform correlation analysis between multiple parameters in the initial parameters of the wind turbine and the blade root load to obtain a set of correlation coefficients.

[0136] In this embodiment, according to formula 6:

[0137] (6)

[0138] Obtain the correlation coefficient ρ X,Y Where X is a parameter; Y is the blade root load; ρ X ρ is the standard deviation of the parameter. Y is the standard deviation of the leaf root load; cov(X,Y) is the covariance between X and Y.

[0139] In this embodiment, multiple parameters are used to obtain multiple correlation coefficients according to Formula 6, and these multiple correlation coefficients are combined into a set of correlation coefficients.

[0140] Step S502: Compare the multiple correlation coefficients in the correlation coefficient set with the preset correlation coefficient threshold.

[0141] In this embodiment, multiple correlation coefficients from the correlation coefficient set are extracted one by one and compared with a preset correlation coefficient threshold.

[0142] In this embodiment, for example, the correlation coefficient threshold is 0.6.

[0143] Step S503: Obtain the subset of correlation coefficients that are greater than the preset correlation coefficient threshold.

[0144] In this embodiment, correlation coefficients greater than a preset correlation coefficient threshold are extracted from the correlation coefficient set to obtain a subset of correlation coefficients.

[0145] In this embodiment, when the correlation coefficient is greater than the preset correlation coefficient threshold, the current parameter is strongly correlated with the blade root load; when the correlation coefficient is greater than 0.2 and less than 0.6, the current parameter is moderately correlated with the blade root load; when the correlation coefficient is less than 0.2, the current parameter is weakly correlated with or unrelated to the blade root load.

[0146] In this embodiment, a preset correlation coefficient threshold can be selected within a range greater than 0 and less than 1, as needed.

[0147] Step S504: Match the subset of correlation coefficients with the initial parameters of the wind turbine to obtain the wind turbine data.

[0148] In this embodiment, the coefficients in the relevant coefficient subset are matched with the initial parameters of the wind turbine to obtain the wind turbine data.

[0149] In this embodiment, the wind turbine data includes multiple wind turbine parameters.

[0150] In this embodiment, multiple parameters in the initial parameters of the wind turbine are correlated with the blade root load to obtain a set of correlation coefficients; multiple correlation coefficients in the set of correlation coefficients are compared with a preset correlation coefficient threshold; a subset of correlation coefficients greater than the preset correlation coefficient threshold is obtained; the subset of correlation coefficients is matched with the initial parameters of the wind turbine to obtain wind turbine data, and then wind turbine data with high correlation to blade root load is obtained.

[0151] Figure 6 This is a flowchart illustrating Embodiment Six of a wind turbine blade root load prediction method provided in this application. Figure 6 As shown, specifically, after step S206, the method further includes:

[0152] Step S601: Obtain the actual leaf root load.

[0153] In this embodiment, the actual leaf root load is the data in the test dataset.

[0154] Step S602: Based on the actual leaf root load and the predicted leaf root load, obtain the accurate value of the preset leaf root load prediction model.

[0155] In this embodiment, according to formula 7:

[0156] (7)

[0157] The mean absolute error ε is obtained. Where n represents the sample size; y i To predict the leaf root load; representing the actual leaf root load.

[0158] In this embodiment, according to formula 8:

[0159] (8)

[0160] The root mean square error ρ is obtained.

[0161] In this embodiment, according to formula 9:

[0162] (9)

[0163] Obtain the coefficient of determination R 2 Where represents the mean of the actual leaf root load.

[0164] In this embodiment, since the actual leaf root load is the leaf root load within a preset time period, the actual leaf root load has multiple leaf root load values.

[0165] In this embodiment, the mean absolute error, root mean square error, or coefficient of determination can be used as the accurate value of the preset leaf root load prediction model.

[0166] Step S603: If the accuracy value is determined to be lower than the preset accuracy threshold, the training data is filtered to obtain the filtered dataset.

[0167] In this embodiment, if the accuracy value is determined to be lower than the preset accuracy threshold, the dataset is refitted using the frequency domain plot and time plot of the data obtained by the sensor, unreasonable segments in the dataset are filtered out, and the filtered dataset is reconstructed.

[0168] Step S604: Based on the filtered dataset, obtain the leaf root load prediction update model.

[0169] In this embodiment, the leaf root load prediction update model is obtained by using a filtered dataset and training according to the method of embodiment two. Then, the accuracy value of the leaf root load prediction update model is obtained again. If the accuracy value is determined to be higher than the preset accuracy threshold, the prediction result is obtained.

[0170] In this embodiment, the actual leaf root load is obtained; based on the actual leaf root load and the predicted leaf root load, the accuracy value of the preset leaf root load prediction model is obtained; if the accuracy value is determined to be lower than the preset accuracy threshold, the training data is filtered to obtain a filtered dataset; based on the filtered dataset, the leaf root load prediction update model is obtained, thereby obtaining a more accurate leaf root load prediction model.

[0171] Figure 7 This is a flowchart illustrating Embodiment Seven of a wind turbine blade root load prediction method provided in this application. Figure 7 As shown, specifically, after step S102, the method further includes:

[0172] Step S701: In the i-th iteration, obtain the calculated value of the equivalent fatigue load according to the blade root ultimate load and the number of cycles.

[0173] In this embodiment, substitute the blade root ultimate load and the number of cycles into Formula 5 to obtain the calculated value of the equivalent fatigue load.

[0174] Step S702: Obtain the equivalent fatigue load error value according to the predicted equivalent fatigue load and the calculated value of the equivalent fatigue load.

[0175] In this embodiment, subtract the predicted equivalent fatigue load from the calculated value of the equivalent fatigue load to obtain a difference, and this difference is the equivalent fatigue load error value.

[0176] In this embodiment, according to the equivalent fatigue load error value, the accuracy of the predicted equivalent fatigue load can be judged. That is, the smaller the equivalent fatigue load error value, the higher the accuracy of the predicted equivalent fatigue load; the larger the equivalent fatigue load error value, the lower the accuracy of the predicted equivalent fatigue load.

[0177] Step S703: Judge whether the equivalent fatigue load error value is greater than the preset error value.

[0178] Step S704: Determine that the equivalent fatigue load error value is greater than the preset error value, and retrain the (i - 1)-th blade root load prediction model to obtain the i-th preset blade root load prediction model.

[0179] In this embodiment, when i = 1, the (i - 1)-th blade root load prediction model is the preset blade root load prediction model.

[0180] In this embodiment, if it is determined that the equivalent fatigue load error value is greater than the preset error value, then retrain the (i - 1)-th blade root load prediction model according to Steps S201 to S206.

[0181] In this embodiment, 0 < i ≤ I, where I is the number when the equivalent fatigue load error value is less than the preset error value, and the I-th preset blade root load prediction model is obtained.

[0182] In this embodiment, in the i-th iteration, obtain the calculated value of the equivalent fatigue load according to the blade root ultimate load and the number of cycles; obtain the equivalent fatigue load error value according to the predicted equivalent fatigue load and the calculated value of the equivalent fatigue load; judge whether the equivalent fatigue load error value is greater than the preset error value; determine that the equivalent fatigue load error value is greater than the preset error value, and retrain the (i - 1)-th blade root load prediction model to obtain the i-th blade root load prediction model, so as to obtain a blade root load prediction model with high prediction accuracy.

[0183] Figure 8 It is a schematic flowchart of Embodiment 8 of a method for predicting the blade root load of a wind turbine provided by an embodiment of the present application. As Figure 8 As shown, specifically, after step S103, the method further includes:

[0184] Step S801: Confirm that the adjustment of the operating parameters of the wind turbine unit has been completed.

[0185] In this embodiment, after determining that the operating parameters of the wind turbine have been adjusted, completion information is obtained, and based on the completion information, it is determined that the adjustment of the operating parameters of the wind turbine has been completed.

[0186] Step S802: Monitor leaf root load data.

[0187] In this embodiment, after the operating parameters of the wind turbine are adjusted, the blade root load data needs to be monitored in order to determine the change in blade root load after the adjustment.

[0188] Step S803: If the blade root load data is determined to be greater than the preset blade root load, the wind turbine unit shall be shut down.

[0189] In this embodiment, after adjusting the operating parameters of the wind turbine, the monitored blade root load data is greater than the preset blade root load. In order to avoid damage to the blades due to excessive blade root load, the wind turbine needs to be shut down.

[0190] In this embodiment, the operating parameters of the wind turbine are adjusted; the blade root load data is monitored; if the blade root load data is found to be greater than the preset blade root load, the wind turbine is shut down to ensure the stable operation of the wind turbine.

[0191] It should be noted that although the steps in the above embodiments are described in a specific order, those skilled in the art will understand that in order to achieve the effect of this application, different steps do not necessarily have to be executed in such an order. They can be executed simultaneously (in parallel) or in other orders, and these variations are all within the scope of protection of this application.

[0192] Furthermore, this application also provides a blade root load prediction device for wind turbines.

[0193] Figure 9 This is a schematic diagram of an embodiment of a wind turbine blade root load prediction device provided in this application. Figure 9As shown, the apparatus in this embodiment mainly includes an acquisition module 11, a processing module 12, and an analysis module 13. In some embodiments, one or more of the acquisition module 11, processing module 12, and analysis module 13 can be combined into a single module. In some embodiments, the acquisition module 11 can be configured to acquire blade root load data and wind turbine parameters. The processing module 12 can be configured to input the blade root load data and wind turbine parameters into a preset blade root load prediction model to obtain the predicted blade root load within a preset time period; wherein, the predicted blade root load includes the average blade root load, the ultimate blade root load, the number of cycles, and the predicted equivalent fatigue load. The analysis module 13 can be configured to adjust the operating parameters of the wind turbine based on the predicted blade root load. The process of obtaining a preset blade root load prediction model includes: acquiring training data, which includes blade root load and wind turbine data; when training the i-th decision tree, selecting a data set from the training data and selecting a preset number of feature variables from the data set; dividing the split points of the i-th decision tree according to the preset number of feature variables to obtain nodes, and determining whether the nodes meet the stopping splitting condition; if the stopping splitting condition is met, the i-th decision tree is obtained; when i=I, training is stopped; where I is the number of decision trees; and combining the obtained decision trees to obtain the preset blade root load prediction model.

[0194] The aforementioned wind turbine blade root load prediction device is used for execution Figure 1 The wind turbine blade root load prediction method embodiments shown are similar in technical principle, the technical problems solved, and the technical effects produced. Those skilled in the art can clearly understand that, for the sake of convenience and brevity, the specific working process and related descriptions of the wind turbine blade root load prediction device can be found in the embodiments of the wind turbine blade root load prediction method, which will not be repeated here.

[0195] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments of this application can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable storage medium can include any entity or device capable of carrying the computer program code, a medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory, a random access memory, an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc. It should be noted that the content included in the computer-readable storage medium can be appropriately added to or subtracted according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable storage medium does not include electrical carrier signals and telecommunication signals.

[0196] Furthermore, this application also provides a leaf root load prediction device.

[0197] Figure 10 This is a schematic diagram of a first embodiment of a leaf root load prediction device provided in this application. Figure 10 As shown, the leaf root load prediction device includes at least one processor 21 and a memory 22, the memory 22 being configurable to store data executed as described above. Figures 1 to 8 The program for the blade root load prediction method of the wind turbine in the illustrated embodiment is such that the processor 21 can be configured to execute a program stored in the memory 22. This program includes, but is not limited to, a program for executing a blade root load prediction method for a wind turbine according to the above-described method embodiment. For ease of explanation, only the parts related to the embodiments of this application are shown; for specific technical details not disclosed, please refer to the method section of the embodiments of this application. The blade root load prediction device can be a control device comprising various electronic devices.

[0198] Furthermore, this application also provides a computer-readable storage medium. In one embodiment of the computer-readable storage medium according to this application, the computer-readable storage medium can be configured to store a program for performing the blade root load prediction method of the wind turbine in the above-described method embodiments. This program can be loaded and run by a processor to implement the blade root load prediction method of the wind turbine. For ease of explanation, only the parts related to the embodiments of this application are shown; for specific technical details not disclosed, please refer to the method section of the embodiments of this application. The computer-readable storage medium can be a storage device device including various electronic devices. Optionally, in the embodiments of this application, the computer-readable storage medium is a non-transitory computer-readable storage medium.

[0199] Furthermore, it should be understood that since the various modules are only provided to illustrate the functional units of the device described in this application, the physical devices corresponding to these modules may be the processor itself, or a part of the processor's software, hardware, or a combination of both. Therefore, the number of modules shown in the figures is merely illustrative.

[0200] Those skilled in the art will understand that the various modules in the device can be adaptively split or combined. Such splitting or combining of specific modules will not cause the technical solution to deviate from the principles of this application; therefore, the technical solutions after splitting or combining will fall within the protection scope of this application.

[0201] The technical solutions of this application have been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of this application is obviously not limited to these specific embodiments. Without departing from the principles of this application, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of this application.

Claims

1. A method for predicting blade root loads of wind turbine generators, characterized in that, Including: Obtain root load data and wind turbine parameters; Input the root load data and the wind turbine parameters into a preset root load prediction model to obtain the predicted root load within a preset time; wherein, the predicted root load includes the root average load, the root ultimate load, the number of cycles, and the predicted equivalent fatigue load; Adjust and process the operating parameters of the wind turbine according to the predicted root load; Wherein, obtaining the preset root load prediction model includes: Obtain training data; wherein, the training data includes root load and wind turbine data; When training the i-th decision tree, select a data group from the training data and select a preset number of feature variables from the data group; According to the preset number of feature variables, perform division processing on the splitting point of the i-th decision tree to obtain nodes, and determine whether the nodes meet the stop division condition; If it is determined that the stop division condition is met, obtain the i-th decision tree; When it is determined that i = I, stop training; wherein, I is the number of decision trees; Combine the obtained multiple decision trees to obtain the preset root load prediction model; Wherein, obtaining the root load includes: Obtain the flap strain signal and the pitch strain signal of the root; According to the flap strain signal and the pitch strain signal, obtain the blade flap moment and the blade pitch moment; According to the blade flap moment and the blade pitch moment, obtain the load maximum value, the average load, and the load cycle number; According to the load maximum value and the load cycle number, obtain the equivalent fatigue load; Combine the equivalent fatigue load, the average load, and the load maximum value to obtain the root load; Wherein, after inputting the root load data and the wind turbine parameters into a preset root load prediction model to obtain the predicted root load within a preset time, the method further includes: In the n-th iteration, obtain the equivalent fatigue load calculation value according to the root ultimate load and the number of cycles; According to the predicted equivalent fatigue load and the equivalent fatigue load calculation value, obtain the equivalent fatigue load error value; Judge whether the equivalent fatigue load error value is greater than a preset error value; Determine that the equivalent fatigue load error value is greater than the preset error value, and retrain the (n - 1)-th root load prediction model to obtain the n-th root load prediction model; wherein, when n = 1, the (n - 1)-th root load prediction model is the preset root load prediction model; Wherein, 0 < n ≤ N, and N is the number when the equivalent fatigue load error value is less than the preset error value to obtain the N-th preset root load prediction model.

2. The method according to claim 1, characterized in that, Obtaining the wind turbine data includes: Obtain the initial parameters of the wind turbine; Perform screening processing on the initial parameters of the wind turbine to obtain the wind turbine data.

3. The method according to claim 2, characterized in that, The performing screening processing on the initial parameters of the wind turbine to obtain the wind turbine data includes: Perform correlation analysis on multiple parameters in the initial parameters of the wind turbine with the root load respectively to obtain a set of correlation coefficients; The multiple correlation coefficients in the set of correlation coefficients are compared with a preset correlation coefficient threshold. Obtain the subset of correlation coefficients that are greater than the preset correlation coefficient threshold; The subset of correlation coefficients is matched with the initial parameters of the wind turbine to obtain the wind turbine data.

4. The method according to claim 1, characterized in that, After combining the obtained multiple decision trees to obtain the preset leaf root load prediction model, the method further includes: Obtain the actual leaf root load; Based on the actual leaf root load and the predicted leaf root load, the accurate value of the preset leaf root load prediction model is obtained; If the accuracy value is determined to be lower than a preset accuracy threshold, the training data is filtered to obtain a filtered dataset. Based on the filtered dataset, the leaf root load prediction update model is obtained.

5. The method according to claim 1, characterized in that, The method further includes: The adjustment of the operating parameters of the wind turbine unit has been completed. Monitor leaf root load data; If the blade root load data is determined to be greater than the preset blade root load, the wind turbine unit will be shut down.

6. A blade root load prediction device for wind turbines, characterized in that, include: The acquisition module is used to acquire blade root load data and wind turbine parameters; The processing module is used to input the blade root load data and the wind turbine parameters into a preset blade root load prediction model to obtain the predicted blade root load within a preset time period; wherein, the predicted blade root load includes the average blade root load, the ultimate blade root load, the number of cycles, and the predicted equivalent fatigue load. The analysis module is used to adjust the operating parameters of the wind turbine based on the predicted blade root load. The process of obtaining the preset leaf root load prediction model includes: Acquire training data; wherein the training data includes blade root load and wind turbine data; When training the i-th decision tree, a data group is selected from the training data, and a preset number of feature variables are selected from the data group; Based on the preset number of feature variables, the split point of the i-th decision tree is divided to obtain a node, and it is determined whether the node meets the stopping splitting condition. If the stopping partitioning condition is satisfied, then the i-th decision tree is obtained; Training stops when i = I; where I is the number of decision trees. The obtained decision trees are combined to obtain the preset leaf root load prediction model. The acquisition of the leaf root load includes: Acquire the swing strain signal and flapping strain signal of the leaf root; Based on the swing strain signal and the flapping strain signal, the blade swing torque and the blade flapping torque are obtained; Based on the blade oscillation moment and the blade flapping moment, the maximum load, average load, and number of load cycles are obtained; The equivalent fatigue load is obtained based on the maximum load value and the number of load cycles. The equivalent fatigue load, the average load, and the maximum load value are combined to obtain the blade root load; The method further includes, after inputting the blade root load data and the wind turbine parameters into a preset blade root load prediction model to obtain the predicted blade root load within a preset time period: In the nth iteration, the equivalent fatigue load is calculated based on the leaf root ultimate load and the number of cycles. An equivalent fatigue load error value is obtained based on the predicted equivalent fatigue load and the calculated value of the equivalent fatigue load; It is judged whether the equivalent fatigue load error value is greater than a preset error value; When it is determined that the equivalent fatigue load error value is greater than the preset error value, the (n - 1)-th blade root load prediction model is retrained to obtain the n-th blade root load prediction model; wherein, when n = 1, the (n - 1)-th blade root load prediction model is the preset blade root load prediction model; Where, 0 < n ≤ N, and N is the N-th preset blade root load prediction model obtained when the equivalent fatigue load error value is less than the preset error value.

7. A leaf root load prediction device, comprising a processor and a storage device, said storage device being adapted to store multiple lines of program code, characterized in that, The program code is adapted to be loaded and run by the processor to execute the method according to any one of claims 1 to 5.

8. A computer-readable storage medium storing a plurality of program codes, characterized in that, The program code is adapted to be loaded and run by a processor to execute the method according to any one of claims 1 to 5.

Images