WO2019165753A1 - 风力发电机组的载荷预测方法和装置 - Google Patents

风力发电机组的载荷预测方法和装置 Download PDF

Info

Publication number
WO2019165753A1
WO2019165753A1 PCT/CN2018/098003 CN2018098003W WO2019165753A1 WO 2019165753 A1 WO2019165753 A1 WO 2019165753A1 CN 2018098003 W CN2018098003 W CN 2018098003W WO 2019165753 A1 WO2019165753 A1 WO 2019165753A1
Authority
WO
WIPO (PCT)
Prior art keywords
load
prediction model
data
load prediction
wind turbine
Prior art date
Application number
PCT/CN2018/098003
Other languages
English (en)
French (fr)
Inventor
李岩
余梦婷
刘佳赐
李景旸
黄丽丽
Original Assignee
北京金风科创风电设备有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 北京金风科创风电设备有限公司 filed Critical 北京金风科创风电设备有限公司
Publication of WO2019165753A1 publication Critical patent/WO2019165753A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation

Definitions

  • the present application relates generally to the field of wind power generation technology, and more particularly to a method and apparatus for predicting load of a wind turbine.
  • the present application provides a load prediction method and apparatus for a wind power generator capable of realizing a rapid assessment of the load of a wind turbine set specific wind turbine.
  • the present application provides a load forecasting method for a wind power generator, comprising: acquiring standard related data corresponding to all working conditions; and calculating, according to standard related data, a load corresponding to each working condition to obtain corresponding load data; Using the standard correlation data as the input of the load prediction model, the load data as the output of the load prediction model, training and establishing the load prediction model; obtaining the actual relevant data of the wind turbine, and predicting the wind turbine in the corresponding working condition through the load prediction model The load underneath.
  • the present invention relates to a load forecasting apparatus for a wind power generator, comprising: a standard data acquisition module that acquires standard related data corresponding to all working conditions; and a load data obtaining module that calculates and calculates respectively based on the standard related data
  • the load corresponding to the working condition obtains corresponding load data
  • the load prediction model establishing module uses the standard correlation data as an input of the load prediction model, and the load data is used as an output of the load prediction model to train and establish the load The prediction model;
  • the load prediction module acquires actual relevant data of the wind turbine, and predicts the load of the wind turbine under the corresponding working condition by using the load prediction model.
  • the present application provides a computer readable storage medium storing a computer program that, when executed by a processor, implements the load forecasting method of the wind turbine described above.
  • the present application provides a computing device including: a processor; a memory storing a computer program, and implementing the load forecasting method of the wind turbine set when the computer program is executed by the processor.
  • FIG. 1 illustrates a flowchart of a load prediction method of a wind power generator set according to an exemplary embodiment of the present application
  • FIG. 2 illustrates a flow chart of steps of training a load prediction model in accordance with an exemplary embodiment of the present application
  • FIG. 3 illustrates a configuration diagram of a load predicting device of a wind power generator set according to an exemplary embodiment of the present application.
  • FIG. 1 illustrates a flow chart of a load prediction method of a wind power generator set according to an exemplary embodiment of the present application.
  • step S10 standard related data corresponding to all operating conditions is acquired.
  • the above standard related data may refer to wind turbine related data corresponding to all working conditions defined in the IEC wind turbine design specification.
  • the standard related data corresponding to all the working conditions herein may refer to the wind turbine related data within the parameter range corresponding to each working condition defined in the IEC wind turbine design specification. That is to say, the parameter range corresponding to each working condition is defined in the IEC fan design specification, and the standard related data acquired in step S10 is the relevant data of the wind turbine set in the parameter range corresponding to each working condition.
  • the above operating conditions may include extreme operating conditions (eg, conditions affecting the ultimate load of the wind turbine) and/or fatigue conditions (eg, conditions affecting the fatigue load of the wind turbine), and may also include extreme operating conditions. And/or various sub-conditions under fatigue conditions (eg, electronic conditions, faulty sub-conditions, air rotor conditions, and maintenance sub-conditions, etc.).
  • extreme operating conditions eg, conditions affecting the ultimate load of the wind turbine
  • fatigue conditions eg, conditions affecting the fatigue load of the wind turbine
  • various sub-conditions under fatigue conditions eg, electronic conditions, faulty sub-conditions, air rotor conditions, and maintenance sub-conditions, etc.
  • the standard related data may include wind resource data, in addition to which configuration data of the wind turbine and status data of the wind turbine may be included.
  • the wind resource data may include at least one of the following parameters: wind speed, wind speed deviation, turbulence intensity, wind shear, inflow angle, air density, yaw angle.
  • the configuration data of the wind turbine may include at least one of the following parameters: tower height of the wind turbine, rated power, and impeller diameter.
  • the wind turbine generator configuration data is used as an input of the load prediction model when the load forecasting model is used for load evaluation of the wind power generator, that is, only the above configuration data may exist in the wind farm.
  • the wind turbines with the differences in the parameters involved in the load share the same load prediction model for load evaluation, which expands the use range of the load prediction model, and does not need to establish load prediction separately for wind turbines with different tower heights, rated powers or impeller diameters.
  • the model effectively reduces the workload of building a load forecasting model.
  • the status data of the wind turbine may be a corresponding state parameter value and/or impeller azimuth when the wind turbine is in a predetermined state.
  • the predetermined state may include a grid fault, a yaw fault, and/or a pitch fault.
  • the corresponding state parameter may refer to a time point when the wind turbine generates a grid fault, and a yaw fault occurs. The yaw angle corresponding to the time and/or the pitch angle corresponding to the pitch failure.
  • standard related data corresponding to extreme operating conditions may include wind speed, turbulence intensity, air density, wind shear, yaw angle, rated power, tower height, and impeller diameter.
  • standard related data corresponding to fault sub-conditions may include wind shear, air density, turbulence intensity, rated power, tower height, impeller diameter, time point at which the fault occurred, and impeller azimuth.
  • step S20 the load corresponding to all working conditions is calculated based on the acquired standard correlation data to obtain corresponding load data.
  • the load corresponding to all working conditions can be calculated based on the acquired standard correlation data by means of time domain simulation.
  • the present application is not limited thereto, and the load corresponding to the working condition may be obtained by other means.
  • the load output channel of the simulation calculation may be defined to determine the calculation result as the load of the predetermined component of the wind power generator under the corresponding working condition.
  • the corresponding working conditions herein may refer to the working conditions in which the user desires to perform load evaluation in all working conditions. That is to say, the load obtained by the simulation calculation may be the load of the predetermined component of the wind power generator under the corresponding working conditions.
  • the predetermined component may be a key component in the wind turbine, for example, the predetermined component may include any one of the following: a blade root of the wind turbine, a blade section, a hub center, a yaw bearing, a tower bottom, Tower section.
  • the defined simulated calculated load output channel may be a critical load (e.g., blade root load, etc.) that affects the strength of the predetermined component of the wind turbine.
  • the load data at this time may include at least one of the following: the mean of the load, the standard deviation of the load, the maximum value of the load, the minimum value of the load, and the 1HZ equivalent load.
  • the load may include bending moments and forces around three coordinate axes in a Cartesian coordinate system, for example, a bending moment (Mx) along the X-axis direction, a bending moment (My) along the Y-axis direction, and an X-axis direction.
  • a bending moment (Mx) along the X-axis direction a bending moment along the X-axis direction
  • a bending moment (My) along the Y-axis direction
  • an X-axis direction Synthetic bending moment (Mxy) in the Y-axis direction
  • bending moment (Mz) in the Z-axis direction force (Fx) in the X-axis direction
  • force (Fy) in the Y-axis direction force (Fy) in the Y-axis direction
  • X-axis direction X-axis direction
  • Y-axis direction Y-axis direction
  • the working condition is a fatigue sub-condition
  • the 1HZ equivalent load of the predetermined component of the wind-generator set under the fatigue sub-condition can be obtained through simulation calculation.
  • the working condition is the limit sub-condition
  • the maximum value and load of the load of the predetermined component of the wind-generator set under the limit sub-condition can be obtained through simulation calculation. The minimum value.
  • step S30 the acquired standard correlation data is used as an input of the load prediction model, and the obtained load data is used as an output of the load prediction model, and the load prediction model is trained and established.
  • the load prediction model may include a load prediction model corresponding to each of the operating conditions.
  • each seed condition corresponds to a load prediction model.
  • the load prediction model can be optimized by mathematical methods such as optimization algorithm or least squares method to improve the output precision of the load prediction model.
  • the load prediction model may be a regression equation based on statistics and training of a plurality of multivariate inputs and multi-channel output data.
  • the regression equation can be a second-order polynomial model, a polynomial hybrid expansion model, or a Kriging model.
  • the present application is not limited thereto, and other regression equations (for example, linear regression equations) may also be employed to represent the load prediction model.
  • the following describes the process of training the load prediction model: first determine the input parameters of the load prediction model (for example, the standard correlation data corresponding to all working conditions), and calculate it in the time domain simulation mode before training the load prediction model.
  • the output load corresponding to the input parameter of the determined load prediction model is then trained based on the input parameters of the determined load prediction model and the output load calculated by the time domain simulation mode. It should be understood that the present application is not limited thereto, and an output load corresponding to an input parameter of the determined load prediction model may be calculated by other means as an output of the load prediction model.
  • the obtained output load may also be filtered to use the filtered output load as the output of the load prediction model to predict the load.
  • the model is trained. It should be understood that the output load can be filtered in various ways, which is not limited in this application.
  • the process of training the second-order polynomial model may be: taking the input parameters of the determined load prediction model and the output load obtained by the time domain simulation method as the second-order polynomial
  • the input and output of the model training, the training process may include the construction order matrix, the combined argument matrix, the solution coefficient matrix, the preservation coefficient matrix and the order matrix, and the output and input parameters of the second-order polynomial model after training. The relevance of the user meets the needs of the user.
  • the process of training the polynomial hybrid expansion model may be as follows: the input parameters of the determined load prediction model and the output load obtained by the time domain simulation method are mixed as high-order polynomials.
  • the input and output of the model training are dynamically expanded.
  • the training process may include constructing the order matrix, defining the basis function matrix, combining the independent variable matrix, solving the coefficient matrix, the preservation coefficient matrix, and the order matrix in order of execution, and the high order after the final training.
  • the correlation between the output of the polynomial mixed expansion model and the input parameters meets the user's needs.
  • the process of training the Kriging model may be: training the input parameters of the determined load prediction model and the output load obtained by the time domain simulation method as the Kriging model.
  • the training process may include constructing the order matrix, constructing the trend function, constructing the deviation function, saving the trend function and the deviation function coefficient matrix and the order matrix, and outputting and inputting the Kriging model after the final training.
  • the relevance of the parameters meets the needs of the user.
  • FIG. 2 illustrates a flow chart of steps of training a load prediction model in accordance with an exemplary embodiment of the present application.
  • step S201 parameters applicable to the load prediction model (i.e., the load prediction model to be trained) included in the standard correlation data are determined.
  • the parameter applicable to the load prediction model may be standard related data in a working condition corresponding to the load prediction model to be trained.
  • step S202 sample points for input to the load prediction model are selected from the determined parameters by a pseudo-random sample point generation algorithm.
  • the pseudo-random sample point generation algorithm can select the first predetermined number of sample points from the wind speed and the second predetermined number of sample points from the air density as the load prediction model. input of.
  • sample points are selected from each parameter by means of uniform sampling, resulting in a larger amount of data of the selected sample points, increasing the calculation amount of the subsequent load prediction model, which is passed in the exemplary embodiment of the present application.
  • the above pseudo-random sample point generation algorithm to select sample points can effectively reduce the calculation amount of the simulation sample.
  • the application is not limited thereto, and a sample point for inputting to the load prediction model may be selected by using a random number sequence, a uniform sampling, a custom selection rule, and the like.
  • step S203 the selected sample point is used as an input of the load prediction model, and the output load obtained by the simulation calculation (for example, time domain simulation mode) according to the selected sample point is output as a load prediction model, and the load prediction model is trained.
  • the output load obtained from the selected sample points may also be otherwise determined.
  • the load prediction method of the wind power generator may further include a step of verifying the load prediction model.
  • the load prediction model can be verified based on input data used to train the load prediction model.
  • the standard correlation data for training the load prediction model may be input to the load prediction model, and the output of the load and load prediction model of the wind turbine generated by comparing the standard correlation data simulation based on the training load prediction model may be compared.
  • the load prediction model has passed verification. It should be appreciated that various ways can be utilized to calculate the load of the wind turbine based on the standard related data used to train the load prediction model.
  • the load prediction model can be verified based on data obtained by simulation calculation.
  • the standard related data corresponding to the corresponding working condition corresponding to the load prediction model can be arbitrarily selected, and the arbitrarily selected standard related data is input to the load prediction model, and the wind power generation corresponding to the arbitrarily selected standard related data is obtained through simulation calculation.
  • the load of the unit is determined by comparing the output of the load and load prediction model obtained through simulation calculation to determine whether the load prediction model has passed verification.
  • the load prediction method of the wind power generator of the above-described exemplary embodiment of the present application it is possible to construct a load prediction model of the wind power generation unit, and to implement a load prediction model for a working condition of the wind power generation unit, which may be different according to working conditions.
  • the load of the corresponding wind turbine is determined, which increases the flexibility of the load prediction model.
  • the establishment and maintenance of a huge load database is also avoided.
  • step S40 actual correlation data of the wind turbine is acquired, and the load of the wind turbine under the corresponding operating conditions is predicted by the load prediction model. That is to say, after establishing the load prediction model corresponding to all the working conditions, the load corresponding to the wind turbine can be obtained by inputting the actual relevant data of the wind turbine into the constructed load prediction model, and the constructed load prediction model is constructed. It can be used for rapid assessment of specific wind turbine load at wind farm site and for wind farm optimization process load constraints.
  • the actual correlation data of the acquired wind turbine may be decomposed to obtain data corresponding to each working condition, and the data obtained by the decomposition is input to the load prediction.
  • the model is used to obtain loads under various operating conditions.
  • the corresponding working condition can be one working condition or multiple working conditions in all working conditions.
  • the actual relevant data of the wind turbine should be consistent with the input data used to train the load prediction model corresponding to the corresponding operating conditions. That is to say, the type of actual relevant data acquired should be consistent with the type of standard related data of the load prediction model corresponding to the training corresponding to the corresponding working condition.
  • the actual relevant data of the acquired wind turbine may be screened, and then the actual correlation data of the selected wind turbine is input into the load prediction model.
  • the step of screening the actual correlation data of the acquired wind turbine may include: screening the actual relevant data of the acquired wind turbine with the predetermined parameters as constraints. That is, unreasonable data in the actual related data of the wind turbine is filtered based on predetermined parameters.
  • the predetermined parameter may be at least one of a wind speed, a generator speed, a pitch angle, an electromagnetic torque, and a shaft power.
  • the step of screening the actual correlation data of the acquired wind power generator may include: determining a parameter range corresponding to the predetermined parameter, and filtering out parameters corresponding to the predetermined parameter in the actual relevant data of the wind power generator set.
  • the actual relevant data within the range is used to input the actual relevant data of the selected wind turbine into the load prediction model to determine the load of the wind turbine under the corresponding working conditions.
  • the wind turbine after predicting the load of the wind turbine under the respective operating conditions, the wind turbine may be subjected to rapid condition assessment based on the predicted load.
  • Step S40 may further include: post-processing the predicted wind turbine load under the corresponding working condition for the condition evaluation.
  • Post-processing can refer to the processing of the load that is suitable for the evaluation of the corresponding condition.
  • the above condition assessment may include a fatigue condition assessment or a limit condition assessment.
  • the condition assessment may include an assessment of various sub-conditions under extreme conditions and/or fatigue conditions.
  • the treatment may refer to the calculation of the obtained 1HZ equivalent load by linear superposition, and the equivalent load of N years can be obtained, thereby evaluating the fatigue life of the wind turbine under the corresponding working conditions.
  • the predicted wind turbine is correspondingly
  • the post-loading of the load under working conditions can be used to calculate the maximum value of the load under all extreme conditions and the minimum value of the load to evaluate the extreme conditions of the wind turbine.
  • FIG. 3 illustrates a configuration diagram of a load predicting device of a wind power generator set according to an exemplary embodiment of the present application.
  • the load prediction apparatus of the wind power generator includes a standard data acquisition module 10, a load data acquisition module 20, a load prediction model establishment module 30, and a load prediction module 40.
  • the standard data acquisition module 10 acquires standard related data corresponding to all working conditions.
  • the above standard related data may refer to wind turbine related data corresponding to all working conditions defined in the IEC wind turbine design specification.
  • the standard related data corresponding to all working conditions may refer to the wind turbine related data within the parameters corresponding to each working condition defined in the IEC wind turbine design specification. That is to say, the parameter range corresponding to each working condition is defined in the IEC fan design specification, and the standard related data acquired by the standard data obtaining module 10 is the relevant data of the wind turbine set in the parameter range corresponding to each working condition.
  • the above operating conditions may include extreme operating conditions (eg, operating conditions that affect wind turbine extreme loads) and/or fatigue operating conditions (eg, operating conditions that affect wind turbine fatigue loading).
  • the above operating conditions may be various sub-conditions under extreme conditions and/or fatigue conditions (eg, electronic operating conditions, fault sub-conditions, air rotor conditions, and maintenance sub-conditions, etc.).
  • the load predicting device of the wind turbine shown in FIG. 3 can quickly evaluate the load of the wind turbine under various sub-limit conditions or various sub-fatigue conditions.
  • the standard related data may include wind resource data, in addition to which may include configuration data of the wind turbine and state data of the wind turbine.
  • the wind resource data may include at least one of the following parameters: wind speed, wind speed deviation, turbulence intensity, wind shear, inflow angle, air density, yaw angle.
  • the configuration data of the wind turbine may include at least one of the following parameters: tower height of the wind turbine, rated power, and impeller diameter.
  • the status data of the wind turbine may be a corresponding state parameter value and/or impeller azimuth when the wind turbine is in a predetermined state.
  • the predetermined state may include a grid fault, a yaw fault, and/or a pitch fault. Accordingly, when the wind turbine is in a predetermined state, the corresponding state parameter may refer to a time point when the wind turbine generates a grid fault, and a yaw fault occurs.
  • the load data acquisition module 20 simulates and calculates loads corresponding to all working conditions based on the acquired standard correlation data to obtain corresponding load data.
  • the load data acquisition module 20 may calculate a load corresponding to all operating conditions based on the acquired standard correlation data by means of time domain simulation.
  • the present application is not limited thereto, and the load corresponding to the working condition may be obtained by other means.
  • the load data acquisition module 20 may define a load output channel of the simulation calculation when performing the simulation calculation to determine the calculation result as the load of the predetermined component of the wind power generator under the corresponding working condition.
  • the corresponding operating conditions may refer to conditions in which all of the operating conditions are desired by the user for load evaluation. That is to say, the load obtained by the simulation calculation may be the load of the predetermined component of the wind power generator under the corresponding working conditions.
  • the predetermined component may be a key component in a wind turbine, for example, the predetermined component may include any one of the following: a blade root of a wind turbine, a blade section, a hub center, a yaw bearing, Tower bottom, tower section.
  • the defined simulated calculated load output channel should be a critical load that affects the strength of the predetermined component of the wind turbine.
  • the load data at this time may include at least one of the following: the mean of the load, the standard deviation of the load, the maximum value of the load, the minimum value of the load, and the 1HZ equivalent load.
  • the load data acquiring module 20 may calculate, by simulation, that the predetermined component of the wind-generator set is 1HZ equivalent load under fatigue sub-conditions. If the working condition is the limit sub-condition, after the standard data obtaining module 10 obtains the standard-related data corresponding to the limit sub-case, the load data acquiring module 20 may calculate the predetermined component of the wind-generator set by the simulation. The maximum value of the load and the minimum value of the load.
  • the load prediction model establishing module 30 uses the acquired standard correlation data as an input of the load prediction model, and the obtained load data is used as an output of the load prediction model, and the load prediction model is trained and established.
  • the load prediction model may include a load prediction model corresponding to each of the operating conditions. For example, each seed condition corresponds to a load prediction model.
  • the load prediction model establishing module 30 may perform optimization training on the load prediction model by using a mathematical method such as an optimization algorithm or a least squares method to improve the output precision of the load prediction model.
  • the load prediction model may be a regression equation based on statistics and training of a plurality of multivariate inputs and multi-channel output data.
  • the regression equation can be a second-order polynomial model, a polynomial hybrid expansion model, or a Kriging model.
  • the present application is not limited thereto, and other regression equations (for example, linear regression equations) may also be employed to represent the load prediction model.
  • the load prediction model establishing module 30 first determines the input parameters of the load prediction model (for example, standard correlation data corresponding to all working conditions), and the load prediction model. Before the training, the output load corresponding to the input parameters of the determined load prediction model can be calculated by time domain simulation, and then the load parameters are predicted based on the input parameters of the determined load prediction model and the output load calculated by the time domain simulation method. The model is trained. It should be understood that the present application is not limited thereto, and an output load corresponding to an input parameter of the determined load prediction model may be calculated by other means as an output of the load prediction model.
  • the load prediction model establishing module 30 may also filter the obtained output load to use the filtered output load as the load prediction model.
  • the output is trained on the load prediction model. It should be understood that the output load can be filtered in various ways, which is not limited in this application.
  • the process of training the second-order polynomial model by the load prediction model building module 30 may be: the load prediction model establishing module 30 will determine the input parameters of the load prediction model and pass the time domain simulation.
  • the output load obtained by the method is used as the input and output of the second-order polynomial model training.
  • the training process may include constructing the order matrix, combining the independent variable matrix, solving the coefficient matrix, saving the coefficient matrix and the order matrix in the order of execution, after the final training.
  • the correlation between the output of the second-order polynomial model and the input parameters meets the user's needs.
  • the load prediction model establishing module 30 may train the polynomial hybrid expansion model by: the load prediction model establishing module 30 will determine the input parameters of the load prediction model and the passage time.
  • the output load obtained by the domain simulation method is used as the input and output of the high-order polynomial hybrid expansion model training.
  • the training process may include constructing the order matrix, defining the basis function matrix, combining the independent variable matrix, solving the coefficient matrix, and saving the coefficients in the order of execution. Matrix and order matrix, the correlation between the output and input parameters of the high-order polynomial expansion model after the final training meets the user's needs.
  • the load prediction model establishing module 30 may train the Kriging model by: the load prediction model establishing module 30 will determine the input parameters of the load prediction model and pass the time domain simulation.
  • the output load obtained by the method is used as the input and output of the Kriging model training.
  • the training process may include constructing the order matrix, constructing the trend function, constructing the deviation function, preserving the trend function, and the deviation function coefficient matrix and the order matrix in order of execution. The correlation between the output of the final training Kriging model and the input parameters meets the user's needs.
  • the process of the load prediction model building module 30 training the load prediction model is described below.
  • the load prediction model establishing module 30 determines parameters included in the standard correlation data that are applicable to the load prediction model. Specifically, the parameter applicable to the load prediction model may be standard related data in a working condition corresponding to the load prediction model to be trained.
  • the load prediction model establishing module 30 selects sample points for input to the load prediction model from the determined parameters by using a pseudo-random sample point generation algorithm, and obtains an output load according to the selected sample points by time domain simulation, and the load prediction model establishing module 30 The selected sample point is used as the input of the load prediction model, and the output load obtained by the simulation calculation according to the selected sample point is output as the load prediction model, and the load prediction model is trained.
  • the load forecasting device of the wind power generator may further include: a model verification module (not shown).
  • the model validation module may validate the load prediction model based on input data used to train the load prediction model.
  • the model verification module may input standard correlation data for training the load prediction model into the load prediction model, and compare the load and load prediction models of the wind turbine generated by the standard correlation data simulation based on the training load prediction model. The results are output to determine if the load prediction model has passed verification. It should be appreciated that various ways can be utilized to calculate the load of the wind turbine based on the standard related data used to train the load prediction model. In addition, the model verification module can also verify the load prediction model based on the data obtained by simulation calculation.
  • the model verification module can arbitrarily select the standard related data corresponding to the corresponding working condition corresponding to the load prediction model, input the arbitrarily selected standard related data into the load prediction model, and obtain the corresponding data corresponding to the arbitrarily selected standard by simulation calculation.
  • the load of the wind turbine is determined by comparing the output of the load and the load prediction model obtained by the simulation calculation to determine whether the load prediction model passes verification.
  • the load prediction module 40 obtains actual correlation data of the wind turbine, and predicts the load of the wind turbine under the corresponding operating conditions by the load prediction model. Specifically, after the load prediction model establishing module 30 establishes load prediction models respectively corresponding to all working conditions, the load prediction module 40 may obtain the wind power generating set by inputting actual relevant data of the wind power generating set to the constructed load forecasting model. Corresponding load, the constructed load prediction model can be used for the rapid assessment of the specific wind turbine load of the wind farm site and the load constraint of the wind farm optimization process.
  • the load prediction module 40 may decompose the actual correlation data of the acquired wind power generator set to obtain data corresponding to each working condition, and input the data obtained by the decomposition into the load forecasting.
  • the model is used to obtain loads under various operating conditions.
  • the corresponding working condition can be one working condition or multiple working conditions in all working conditions.
  • the actual relevant data of the wind turbine should be consistent with the input data used to train the load prediction model corresponding to the corresponding operating conditions.
  • the load forecasting device of the wind turbine may further include: a data screening module (not shown) that inputs the actual relevant data of the acquired wind turbine to the load forecasting Before the model, the data screening module may first filter the actual relevant data of the acquired wind turbine, and then the load prediction module 40 inputs the actual relevant data of the selected wind turbine into the load prediction model to determine that the wind turbine is in The load under the corresponding working conditions.
  • a data screening module (not shown) that inputs the actual relevant data of the acquired wind turbine to the load forecasting Before the model, the data screening module may first filter the actual relevant data of the acquired wind turbine, and then the load prediction module 40 inputs the actual relevant data of the selected wind turbine into the load prediction model to determine that the wind turbine is in The load under the corresponding working conditions.
  • the data screening module filters the actual relevant data of the acquired wind turbine with the predetermined parameters as constraints, and the load prediction module 40 uses the load prediction model based on the actual correlation data of the selected wind turbine. Determine the load of the wind turbine under the corresponding operating conditions.
  • the predetermined parameter may be at least one of wind speed, generator speed, pitch angle, electromagnetic torque, and shaft power.
  • the data screening module may determine a parameter range corresponding to the predetermined parameter, and filter out actual relevant data in the actual correlation data of the wind turbine set that is within the parameter range corresponding to the predetermined parameter, so as to correlate the actual wind turbine generator after the screening.
  • the data is input to a load prediction model to determine the load of the wind turbine under the corresponding operating conditions.
  • the load forecasting device of the wind power generator may further include: a condition evaluation module (not shown) for post-processing the predicted wind turbine load under the corresponding working condition to Used for condition assessment.
  • Post-processing can refer to the processing of the load that is suitable for the evaluation of the corresponding condition.
  • the condition assessment may include a fatigue condition assessment or a limit condition assessment.
  • the condition assessment may include an assessment of various sub-conditions under extreme conditions and/or fatigue conditions.
  • condition evaluation module predicts the wind turbine under the corresponding working conditions.
  • the post-processing of the load can be calculated by linearly superimposing the obtained 1HZ equivalent load, and the equivalent load of N years can be obtained, thereby evaluating the fatigue life of the wind turbine under the corresponding working conditions.
  • condition evaluation module predicts the wind.
  • Post-processing of the load of the generator set under the corresponding working conditions may refer to the calculation of the maximum value of the load under all extreme conditions and the minimum value of the load to evaluate the extreme operating conditions of the wind turbine.
  • a computing device is also provided in accordance with an exemplary embodiment of the present application.
  • the computing device includes a processor and a memory.
  • the memory is used to store computer programs.
  • the computer program is executed by a processor such that the processor executes a computer program of the wind power generation load prediction method as described above.
  • a computer readable storage medium storing a computer program is also provided in accordance with an exemplary embodiment of the present application.
  • the computer readable storage medium stores a computer program that, when executed by a processor, causes the processor to perform the load prediction method of the wind turbine described above.
  • the computer readable recording medium is any data storage device that can store data read by a computer system. Examples of the computer readable recording medium include read only memory, random access memory, read-only optical disk, magnetic tape, floppy disk, optical data storage device, and carrier wave (such as data transmission via the Internet via a wired or wireless transmission path).

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Hardware Design (AREA)
  • Evolutionary Computation (AREA)
  • Geometry (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Supply And Distribution Of Alternating Current (AREA)
  • Wind Motors (AREA)

Abstract

提供一种风力发电机组的载荷预测方法和装置，所述载荷预测方法包括：获取与所有工况对应的标准相关数据；基于所述标准相关数据仿真计算分别与所有工况对应的载荷，得到相应的载荷数据；将所述标准相关数据作为载荷预测模型的输入，所述载荷数据作为所述载荷预测模型的输出，训练并建立所述载荷预测模型；获取风力发电机组的实际相关数据，通过所述载荷预测模型预测风力发电机组在相应工况下的载荷。

Description

风力发电机组的载荷预测方法和装置 技术领域
本申请总体说来涉及风力发电技术领域，更具体地讲，涉及一种风力发电机组的载荷预测方法和装置。
背景技术
随着对风电场定制化需求的日益增加，当采用时域仿真方式进行风力发电机组的载荷评估时，往往需要数小时甚至数天的仿真周期，有可能导致赶不上项目投标和后期风力发电机组优化的时效需求。当采用数据库插值方式时，则需要提前进行大量的仿真计算、数据处理和数据库建立和维护的时间，耗时较长。
发明内容
本申请提供一种风力发电机组的载荷预测方法和装置，能够实现对风电场机位点特定风力发电机组的载荷的快速评估。
一方面，本申请提供一种风力发电机组的载荷预测方法，包括：获取与所有工况对应的标准相关数据；基于标准相关数据仿真计算分别与所有工况对应的载荷，得到相应的载荷数据；将标准相关数据作为载荷预测模型的输入，载荷数据作为载荷预测模型的输出，训练并建立所述载荷预测模型；获取风力发电机组的实际相关数据，通过载荷预测模型预测风力发电机组在相应工况下的载荷。
另一方面，本申请一种风力发电机组的载荷预测装置，包括：标准数据获取模块，获取与所有工况对应的标准相关数据；载荷数据获取模块，基于所述标准相关数据仿真计算分别与所有工况对应的载荷，得到相应的载荷数据；载荷预测模型建立模块，将所述标准相关数据作为载荷预测模型的输入，所述载荷数据作为所述载荷预测模型的输出，训练并建立所述载荷预测模型；载荷预测模块，获取风力发电机组的实际相关数据，通过所述载荷预测模型预测风力发电机组在相应工况下的载荷。
另一方面，本申请提供一种存储有计算机程序的计算机可读存储介质，当计算机程序在被处理器执行时实现上述的风力发电机组的载荷预测方法。
另一方面，本申请提供一种计算装置，计算装置包括：处理器；存储器，存储有计算机程序，当计算机程序被处理器执行时，实现上述的风力发电机组的载荷预测方法。
附图说明
图1示出根据本申请示例性实施例的风力发电机组的载荷预测方法的流程图；
图2示出根据本申请示例性实施例的对载荷预测模型进行训练的步骤的流程图；
图3示出根据本申请示例性实施例的风力发电机组的载荷预测装置的结构图。
具体实施方式
现在，将参照附图更充分地描述不同的示例实施例，一些示例性实施例在附图中示出。
图1示出根据本申请示例性实施例的风力发电机组的载荷预测方法的流程图。
参照图1，在步骤S10中，获取与所有工况对应的标准相关数据。
上述标准相关数据可指与IEC风机设计规范中定义的所有工况对应的风力发电机组的相关数据。具体地，这里与所有工况对应的标准相关数据可指处于IEC风机设计规范中定义的与每种工况对应的参数范围内的风力发电机组的相关数据。也就是说，IEC风机设计规范中定义了每种工况对应的参数范围，步骤S10中获取的标准相关数据为分别处于每种工况对应的参数范围内的风力发电机组的相关数据。
作为示例，上述工况可包括极限工况(例如，影响风力发电机组极限载荷的工况)和/或疲劳工况(例如，影响风力发电机组疲劳载荷的工况)，还可包括极限工况和/或疲劳工况下的各种子工况(例如，发电子工况、故障子工况、空转子工况和维护子工况等)。
作为示例，标准相关数据可包括风资源数据，除此之外还可包括风力发 电机组的配置数据和风力发电机组的状态数据。
风资源数据可包括以下参数中的至少一个：风速、风速偏差、湍流强度、风剪切、入流角、空气密度、偏航角。风力发电机组的配置数据可包括以下参数中的至少一个：风力发电机组的塔架高度、额定功率、叶轮直径。在本申请示例性实施例中，在利用载荷预测模型对风力发电机组进行载荷评估时将风力发电机组的配置数据作为载荷预测模型的输入，也就是说，可将风电场中仅存在上述配置数据中所涉及的参数的差异的风力发电机组共用同一载荷预测模型进行载荷评估，扩大了载荷预测模型的使用范围，不必针对不同塔架高度、额定功率或叶轮直径的风力发电机组来分别建立载荷预测模型，有效减少了构建载荷预测模型的工作量。
此外，风力发电机组的状态数据可为风力发电机组处于预定状态时对应的状态参数值和/或叶轮方位角。所述预定状态可包括电网故障、偏航故障和/或变桨故障，相应地，风力发电机组处于预定状态时对应的状态参数可指风力发电机组发生电网故障时的时间点、发生偏航故障时所对应的偏航角和/或发生变桨故障时所对应的桨距角。
作为示例，与极限工况对应的标准相关数据可包括风速、湍流强度、空气密度、风剪切、偏航角、额定功率、塔架高度和叶轮直径。例如，与故障子工况对应的标准相关数据可包括风剪切、空气密度、湍流强度、额定功率、塔架高度、叶轮直径、发生故障的时间点、叶轮方位角。
在步骤S20中，基于获取的标准相关数据仿真计算分别与所有工况对应的载荷，得到相应的载荷数据。
作为示例，可通过时域仿真的方式来分别基于获取的标准相关数据计算得到与所有工况对应的载荷。但本申请不限于此，还可通过其他方式来获得与工况对应的载荷。
作为示例，步骤S20中在进行仿真计算时，可定义仿真计算的载荷输出通道，以确定计算结果为风力发电机组的预定部件在相应工况下的载荷。应理解，这里相应工况可指所有工况中用户希望进行载荷评估的工况。也就是说，通过仿真计算获得的载荷可为风力发电机组的预定部件在相应工况下的载荷。所述预定部件可为风力发电机组中的关键部件，例如，所述预定部件可包括以下项中的任意一项：风力发电机组的叶根、叶片截面、轮毂中心、偏航轴承、塔底、塔架截面。在该示例中，定义的仿真计算的载荷输出通道 可以为影响风力发电机组的预定部件强度的关键载荷(例如，叶根载荷等)。例如，此时所述载荷数据可包括以下项中的至少一项：载荷的均值、载荷的标准偏差、载荷的极大值、载荷的极小值、1HZ等效载荷。
作为示例，载荷可包括在笛卡尔坐标系下围绕三个坐标轴的弯矩和力，例如，沿X轴方向的弯矩(Mx)、沿Y轴方向的弯矩(My)、X轴方向和Y轴方向的合成弯矩(Mxy)、沿Z轴方向的弯矩(Mz)、沿X轴方向的力(Fx)、沿Y轴方向的力(Fy)、X轴方向和Y轴方向的合成力(Fxy)、沿Z轴方向的力(Fz)。
作为示例，如工况为疲劳子工况，则在获取与疲劳子工况对应的标准相关数据之后，可通过仿真计算得到风力发电机组的预定部件在该疲劳子工况下的1HZ等效载荷。如工况为极限子工况，则在获取与极限子工况对应的标准相关数据之后，可通过仿真计算得到风力发电机组的预定部件在该极限子工况下的载荷的极大值和载荷的极小值。
在步骤S30中，将获取的标准相关数据作为载荷预测模型的输入，得到的载荷数据作为载荷预测模型的输出，训练并建立载荷预测模型。
载荷预测模型可包括与所有工况分别对应的载荷预测模型。例如，每种子工况分别对应一个载荷预测模型。可采用优化算法或最小二乘法等数学方法对载荷预测模型进行优化训练，以提高载荷预测模型的输出精度。在本申请示例性实施例中，载荷预测模型可为基于大量多变量输入和多通道输出数据的统计和训练得到的回归方程。该回归方程可为二阶多项式模型、多项式混动展开模型或克里金模型。然而，本申请不限于此，还可采用其他回归方程(例如，线性回归方程)来表示载荷预测模型。
下面介绍对载荷预测模型进行训练的过程：首先确定载荷预测模型的输入参数(例如，与所有工况对应的标准相关数据)，在对载荷预测模型进行训练之前，可通过时域仿真方式计算得到与确定的载荷预测模型的输入参数对应的输出载荷，然后基于确定的载荷预测模型的输入参数和通过时域仿真方式计算得到的输出载荷对载荷预测模型进行训练。应理解，本申请不限于此，还可通过其他方式来计算与确定的载荷预测模型的输入参数对应的输出载荷以作为载荷预测模型的输出。
在通过时域仿真方式计算得到与确定的载荷预测模型的输入参数对应的输出载荷之后，还可对得到的输出载荷进行筛选，以将筛选后的输出载荷作 为载荷预测模型的输出以对载荷预测模型进行训练。应理解，可利用各种方式来对输出载荷进行筛选，本申请对此不做限定。
作为示例，以载荷预测模型为二阶多项式模型为例，对二阶多项式模型进行训练的过程可为：将确定的载荷预测模型的输入参数和通过时域仿真方式获得的输出载荷作为二阶多项式模型训练的输入和输出，训练过程按执行顺序先后可包括构建阶数矩阵、组合自变量矩阵、求解系数矩阵、保存系数矩阵和阶数矩阵，最终训练后的二阶多项式模型的输出与输入参数的相关性满足用户需求。
以载荷预测模型为多项式混动展开模型为例，对多项式混动展开模型进行训练的过程可为：将确定的载荷预测模型的输入参数和通过时域仿真方式获得的输出载荷作为高阶多项式混动展开模型训练的输入和输出，训练过程按执行顺序先后可包括构建阶数矩阵、定义基函数矩阵、组合自变量矩阵、求解系数矩阵、保存系数矩阵和阶数矩阵，最终训练后的高阶多项式混动展开模型的输出与输入参数的相关性满足用户需求。
以载荷预测模型为克里金模型为例，对克里金模型进行训练的过程可为：将确定的载荷预测模型的输入参数和通过时域仿真方式获得的输出载荷作为克里金模型训练的输入和输出，训练过程按执行顺序先后可包括构建阶数矩阵、构建趋势函数、构建偏差函数、保存趋势函数和偏差函数系数矩阵和阶数矩阵，最终训练后的克里金模型的输出与输入参数的相关性满足用户需求。
下面参照图2来介绍对载荷预测模型进行训练的步骤。
图2示出根据本申请示例性实施例的对载荷预测模型进行训练的步骤的流程图。
参照图2，在步骤S201中，确定标准相关数据中包含的适用于载荷预测模型(即，待训练的载荷预测模型)的参数。具体地，适用于载荷预测模型的参数可为与待训练的载荷预测模型对应工况下的标准相关数据。
在步骤S202中，通过伪随机样本点生成算法从确定的参数中选取用于输入到载荷预测模型的样本点。
假设标准相关数据中包含风速和空气密度两种参数，可通过伪随机样本点生成算法从风速中选取第一预定数量的样本点、从空气密度中选取第二预定数量的样本点作为载荷预测模型的输入。
在现有技术中是采用均匀抽样的方式从各参数中选择样本点，导致被选 择的样本点的数据量较大，增加了后续载荷预测模型的计算量，在本申请示例性实施例中通过上述伪随机样本点生成算法来选取样本点可有效降低仿真样本的计算量。但本申请不限于此，还可采用采用随机数序列、均匀抽样、自定义选取规则等方式来选取用于输入到载荷预测模型的样本点。
在步骤S203中，将选取的样本点作为载荷预测模型的输入，将根据选取的样本点通过仿真计算(例如，时域仿真方式)获得的输出载荷作为载荷预测模型输出，对载荷预测模型进行训练。应理解，也可通过其他方式来根据选取的样本点获得的输出载荷。
作为示例，根据本申请示例性实施例的风力发电机组的载荷预测方法还可包括对载荷预测模型进行验证的步骤。例如，可基于用于训练载荷预测模型的输入数据对载荷预测模型进行验证。具体地，可将用于训练载荷预测模型的标准相关数据输入到载荷预测模型，通过比较基于用于训练载荷预测模型的标准相关数据仿真计算得到的风力发电机组的载荷与载荷预测模型的输出结果来确定载荷预测模型是否通过验证。应理解，可利用各种方式来基于用于训练载荷预测模型的标准相关数据计算风力发电机组的载荷。或者，还可基于通过仿真计算方式获得的数据对载荷预测模型进行验证。具体地，可任意选择与载荷预测模型所对应的相应工况相应的标准相关数据，将任意选择的标准相关数据输入到载荷预测模型，通过仿真计算获得与任意选择的标准相关数据对应的风力发电机组的载荷，通过比较通过仿真计算获得的载荷与载荷预测模型的输出结果来确定载荷预测模型是否通过验证。
根据本申请示例性实施例的上述风力发电机组的载荷预测方法，能够构建风力发电机组的载荷预测模型，实现针对风力发电机组的一种工况建立对应的载荷预测模型，可根据工况的不同确定出相应的风力发电机组的载荷，提高了载荷预测模型的灵活性。同时，也避免了庞大载荷数据库的建立与维护。
返回图1，在步骤S40中，获取风力发电机组的实际相关数据，通过所述载荷预测模型预测风力发电机组在相应工况下的载荷。也就是说，在建立分别与所有工况对应的载荷预测模型之后，可通过将风力发电机组的实际相关数据输入到构建的载荷预测模型来获得该风力发电机组对应的载荷，构建的载荷预测模型可用于风电场机位点特定风力发电机组载荷的快速评估和风电场优化过程载荷的约束。
作为示例，在步骤S40中在获取风力发电机组的实际相关数据之后，可对获取的风力发电机组的实际相关数据进行分解获得与各工况分别对应的数据，将分解获得的数据输入到载荷预测模型中以获得在各工况下的载荷。相应工况可为所有工况中的一种工况或多种工况。风力发电机组的实际相关数据应与用于训练与相应工况对应的载荷预测模型的输入数据一致。也就是说，获取的实际相关数据的种类应与训练与相应工况对应的载荷预测模型的标准相关数据的种类一致。
在将获取的风力发电机组的相关数据输入到载荷预测模型之前，可先对获取的风力发电机组的实际相关数据进行筛选，然后再将筛选后的风力发电机组的实际相关数据输入到载荷预测模型，以确定风力发电机组在相应工况下的载荷。作为示例，对获取的风力发电机组的实际相关数据进行筛选的步骤可包括：以预定参数为约束条件，对获取的风力发电机组的实际相关数据进行筛选。即，基于预定参数来过滤掉风力发电机组的实际相关数据中不合理的数据。所述预定参数可为风速、发电机转速、桨距角、电磁扭矩、轴功率中的至少一个。在此情况下，对获取的风力发电机组的实际相关数据进行筛选的步骤可包括：确定预定参数所对应的参数范围，筛选出风力发电机组的实际相关数据中处于所述预定参数所对应的参数范围内的实际相关数据，以将筛选后的风力发电机组的实际相关数据输入到载荷预测模型来确定风力发电机组在相应工况下的载荷。
在本申请示例性实施例中，在预测得到风力发电机组在相应工况下的载荷之后，可基于预测的载荷对风力发电机组进行快速工况评估。
步骤S40还可包括：对预测的风力发电机组在相应工况下的载荷进行后处理，以用于工况评估。后处理可指对载荷进行适用于相应工况评估的处理。作为示例，上述工况评估可包括疲劳工况评估或极限工况评估。工况评估可包括极限工况和/或疲劳工况下的各种子工况的评估。
以工况评估为疲劳工况评估为例，在通过载荷预测模型预测得到风力发电机组在疲劳工况下的1HZ等效载荷之后，对预测得到的风力发电机组在相应工况下的载荷进行后处理可指对获得的1HZ等效载荷通过线性叠加的方式进行计算，可获得N年的等效载荷，从而对风力发电机组在相应工况下的疲劳寿命进行评估。
以工况评估为极限工况评估为例，在通过载荷预测模型预测得到风力发 电机组在极限工况下的载荷的极大值和载荷的极小值之后，对预测得到的风力发电机组在相应工况下的载荷进行后处理可指统计出所有极限工况下的载荷的极大值和载荷的极小值，以对风力发电机组的极限工况进行评估。
图3示出根据本申请示例性实施例的风力发电机组的载荷预测装置的结构图。
如图3所示，根据本申请示例性实施例的风力发电机组的载荷预测装置包括：标准数据获取模块10、载荷数据获取模块20、载荷预测模型建立模块30和载荷预测模块40。
标准数据获取模块10获取与所有工况对应的标准相关数据。
上述标准相关数据可指与IEC风机设计规范中定义的所有工况对应的风力发电机组的相关数据。与所有工况对应的标准相关数据可指处于IEC风机设计规范中定义的与每种工况对应的参数范围内的风力发电机组的相关数据。也就是说，IEC风机设计规范中定义了每种工况对应的参数范围，标准数据获取模块10获取的标准相关数据为分别处于每种工况对应的参数范围内的风力发电机组的相关数据。
作为示例，上述工况可包括极限工况(例如，影响风力发电机组极限载荷的工况)和/或疲劳工况(例如，影响风力发电机组疲劳载荷的工况)。优选地，上述工况可为极限工况和/或疲劳工况下的各种子工况(例如，发电子工况、故障子工况、空转子工况和维护子工况等)。
在本申请示例性实施例中，图3所示的风力发电机组的载荷预测装置可对风力发电机组在各种子极限工况或各种子疲劳工况下的载荷进行快速评估。
作为示例，标准相关数据可包括风资源数据，除此之外可还包括风力发电机组的配置数据和风力发电机组的状态数据。
风资源数据可包括以下参数中的至少一个：风速、风速偏差、湍流强度、风剪切、入流角、空气密度、偏航角。风力发电机组的配置数据可包括以下参数中的至少一个：风力发电机组的塔架高度、额定功率、叶轮直径。风力发电机组的状态数据可为风力发电机组处于预定状态时对应的状态参数值和/或叶轮方位角。所述预定状态可包括电网故障、偏航故障和/或变桨故障，相应地，风力发电机组处于预定状态时对应的状态参数可指风力发电机组发生电网故障时的时间点、发生偏航故障时所对应的偏航角和/或发生变桨故障时所对应的桨距角。
载荷数据获取模块20基于获取的标准相关数据仿真计算分别与所有工况对应的载荷，得到相应的载荷数据。
作为示例，载荷数据获取模块20可通过时域仿真的方式来分别基于获取的标准相关数据计算得到与所有工况对应的载荷。但本申请不限于此，还可通过其他方式来获得与工况对应的载荷。
载荷数据获取模块20在进行仿真计算时，可定义仿真计算的载荷输出通道，以确定计算结果为风力发电机组的预定部件在相应工况下的载荷。应理解，相应工况可指所有工况中用户希望进行载荷评估的工况。也就是说，通过仿真计算获得的载荷可为风力发电机组的预定部件在相应工况下的载荷。作为示例，所述预定部件可为风力发电机组中的关键部件，例如，所述预定部件可包括以下项中的任意一项：风力发电机组的叶根、叶片截面、轮毂中心、偏航轴承、塔底、塔架截面。在该示例中，定义的仿真计算的载荷输出通道应为影响风力发电机组的预定部件强度的关键载荷。例如，此时所述载荷数据可包括以下项中的至少一项：载荷的均值、载荷的标准偏差、载荷的极大值、载荷的极小值、1HZ等效载荷。
作为示例，如工况为疲劳子工况，则在标准数据获取模块10获取与疲劳子工况对应的标准相关数据之后，载荷数据获取模块20可通过仿真计算得到风力发电机组的预定部件在该疲劳子工况下的1HZ等效载荷。如工况为极限子工况，则在标准数据获取模块10获取与极限子工况对应的标准相关数据之后，载荷数据获取模块20可通过仿真计算得到风力发电机组的预定部件在该极限子工况下的载荷的极大值和载荷的极小值。
载荷预测模型建立模块30将获取的标准相关数据作为载荷预测模型的输入，得到的载荷数据作为载荷预测模型的输出，训练并建立载荷预测模型。载荷预测模型可包括与所有工况分别对应的载荷预测模型。例如，每种子工况分别对应一个载荷预测模型。载荷预测模型建立模块30可采用优化算法或最小二乘法等数学方法对载荷预测模型进行优化训练，以提高载荷预测模型的输出精度。在本申请示例性实施例中，载荷预测模型可为基于大量多变量输入和多通道输出数据的统计和训练得到的回归方程。例如，该回归方程可为二阶多项式模型、多项式混动展开模型或克里金模型。然而，本申请不限于此，还可采用其他回归方程(例如，线性回归方程)来表示载荷预测模型。
下面介绍载荷预测模型建立模块30对载荷预测模型进行训练的过程：载 荷预测模型建立模块30首先确定载荷预测模型的输入参数(例如，与所有工况对应的标准相关数据)，在对载荷预测模型进行训练之前，可通过时域仿真方式计算得到与确定的载荷预测模型的输入参数对应的输出载荷，然后基于确定的载荷预测模型的输入参数和通过时域仿真方式计算得到的输出载荷对载荷预测模型进行训练。应理解，本申请不限于此，还可通过其他方式来计算与确定的载荷预测模型的输入参数对应的输出载荷以作为载荷预测模型的输出。
在通过时域仿真方式计算得到与确定的载荷预测模型的输入参数对应的输出载荷之后，载荷预测模型建立模块30还可对得到的输出载荷进行筛选，以将筛选后的输出载荷作为载荷预测模型的输出以对载荷预测模型进行训练。应理解，可利用各种方式来对输出载荷进行筛选，本申请对此不做限定。
以载荷预测模型为二阶多项式模型为例，载荷预测模型建立模块30对二阶多项式模型进行训练的过程可为：载荷预测模型建立模块30将确定的载荷预测模型的输入参数和通过时域仿真方式获得的输出载荷作为二阶多项式模型训练的输入和输出，训练过程按执行顺序先后可包括构建阶数矩阵、组合自变量矩阵、求解系数矩阵、保存系数矩阵和阶数矩阵，最终训练后的二阶多项式模型的输出与输入参数的相关性满足用户需求。
以载荷预测模型为多项式混动展开模型为例，载荷预测模型建立模块30对多项式混动展开模型进行训练的过程可为：载荷预测模型建立模块30将确定的载荷预测模型的输入参数和通过时域仿真方式获得的输出载荷作为高阶多项式混动展开模型训练的输入和输出，训练过程按执行顺序先后可包括构建阶数矩阵、定义基函数矩阵、组合自变量矩阵、求解系数矩阵、保存系数矩阵和阶数矩阵，最终训练后的高阶多项式混动展开模型的输出与输入参数的相关性满足用户需求。
以载荷预测模型为克里金模型为例，载荷预测模型建立模块30对克里金模型进行训练的过程可为：载荷预测模型建立模块30将确定的载荷预测模型的输入参数和通过时域仿真方式获得的输出载荷作为克里金模型训练的输入和输出，训练过程按执行顺序先后可包括构建阶数矩阵、构建趋势函数、构建偏差函数、保存趋势函数和偏差函数系数矩阵和阶数矩阵，最终训练后的克里金模型的输出与输入参数的相关性满足用户需求。
下面介绍载荷预测模型建立模块30训练载荷预测模型的过程。
载荷预测模型建立模块30确定标准相关数据中包含的适用于载荷预测模型的参数。具体地，适用于载荷预测模型的参数可为与待训练的载荷预测模型对应工况下的标准相关数据。载荷预测模型建立模块30通过伪随机样本点生成算法从确定的参数中选取用于输入到载荷预测模型的样本点，并根据选取的样本点通过时域仿真方式获得输出载荷，载荷预测模型建立模块30将选取的样本点作为载荷预测模型的输入，将根据选取的样本点通过仿真计算获得的输出载荷作为载荷预测模型输出，对载荷预测模型进行训练。
在本申请示例性实施例中，风力发电机组的载荷预测装置还可包括：模型验证模块(图中未示出)。
作为示例，模型验证模块可基于用于训练载荷预测模型的输入数据对载荷预测模型进行验证。例如，模型验证模块可将用于训练载荷预测模型的标准相关数据输入到载荷预测模型，通过比较基于用于训练载荷预测模型的标准相关数据仿真计算得到的风力发电机组的载荷与载荷预测模型的输出结果来确定载荷预测模型是否通过验证。应理解，可利用各种方式来基于用于训练载荷预测模型的标准相关数据计算风力发电机组的载荷。此外，模型验证模块还可基于通过仿真计算方式获得的数据对载荷预测模型进行验证。例如，模型验证模块可任意选择与载荷预测模型所对应的相应工况相应的标准相关数据，将任意选择的标准相关数据输入到载荷预测模型，通过仿真计算获得与任意选择的标准相关数据对应的风力发电机组的载荷，通过比较通过仿真计算获得的载荷与载荷预测模型的输出结果来确定载荷预测模型是否通过验证。
载荷预测模块40获取风力发电机组的实际相关数据，通过所述载荷预测模型预测风力发电机组在相应工况下的载荷。具体地，在载荷预测模型建立模块30建立分别与所有工况对应的载荷预测模型之后，载荷预测模块40可通过将风力发电机组的实际相关数据输入到构建的载荷预测模型来获得该风力发电机组对应的载荷，构建的载荷预测模型可用于风电场机位点特定风力发电机组载荷的快速评估和风电场优化过程载荷的约束。
应理解，载荷预测模块40在获取风力发电机组的实际相关数据之后，可对获取的风力发电机组的实际相关数据进行分解获得与各工况分别对应的数据，将分解获得的数据输入到载荷预测模型中以获得在各工况下的载荷。这里，相应工况可为所有工况中的一种工况或多种工况。风力发电机组的实际 相关数据应与用于训练与相应工况对应的载荷预测模型的输入数据一致。
在本申请示例性实施例中，风力发电机组的载荷预测装置还可包括：数据筛选模块(图中未示出)，载荷预测模块40在将获取的风力发电机组的实际相关数据输入到载荷预测模型之前，数据筛选模块可先对获取的风力发电机组的实际相关数据进行筛选，然后载荷预测模块40再将筛选后的风力发电机组的实际相关数据输入到载荷预测模型，以确定风力发电机组在相应工况下的载荷。
作为示例，数据筛选模块以预定参数为约束条件，对获取的风力发电机组的实际相关数据进行筛选，载荷预测模块40基于筛选后的风力发电机组的实际相关数据，通过利用所述载荷预测模型，确定风力发电机组在相应工况下的载荷。预定参数可为风速、发电机转速、桨距角、电磁扭矩、轴功率中的至少一个。数据筛选模块可确定预定参数所对应的参数范围，筛选出风力发电机组的实际相关数据中处于所述预定参数所对应的参数范围内的实际相关数据，以将筛选后的风力发电机组的实际相关数据输入到载荷预测模型来确定风力发电机组在相应工况下的载荷。
在本申请示例性实施例中，风力发电机组的载荷预测装置还可包括：工况评估模块(图中未示出)，对预测的风力发电机组在相应工况下的载荷进行后处理，以用于工况评估。后处理可指对载荷进行适用于相应工况评估的处理。作为示例，工况评估可包括疲劳工况评估或极限工况评估。工况评估可包括极限工况和/或疲劳工况下的各种子工况的评估。
以工况评估为疲劳工况评估为例，在通过载荷预测模型预测得到风力发电机组在疲劳工况下的1HZ等效载荷之后，工况评估模块对预测得到的风力发电机组在相应工况下的载荷进行后处理可指对获得的1HZ等效载荷通过线性叠加的方式进行计算，可获得N年的等效载荷，从而对风力发电机组在相应工况下的疲劳寿命进行评估。
以工况评估为极限工况评估为例，在通过载荷预测模型预测得到风力发电机组在极限工况下的载荷的极大值和载荷的极小值之后，工况评估模块对预测得到的风力发电机组在相应工况下的载荷进行后处理可指统计出所有极限工况下的载荷的极大值和载荷的极小值，以对风力发电机组的极限工况进行评估。
根据本申请的示例性实施例还提供一种计算装置。该计算装置包括处理 器和存储器。存储器用于存储计算机程序。所述计算机程序被处理器执行使得处理器执行如上所述的风力发电机组的载荷预测方法的计算机程序。
根据本申请的示例性实施例还提供一种存储有计算机程序的计算机可读存储介质。该计算机可读存储介质存储有当被处理器执行时使得处理器执行上述风力发电机组的载荷预测方法的计算机程序。该计算机可读记录介质是可存储由计算机系统读出的数据的任意数据存储装置。计算机可读记录介质的示例包括：只读存储器、随机存取存储器、只读光盘、磁带、软盘、光数据存储装置和载波(诸如经有线或无线传输路径通过互联网的数据传输)。
采用本申请示例性实施例的风力发电机组的载荷预测方法和装置，能够对风电场机位点特定风力发电机组的载荷进行快速评估。
尽管已经参照其示例性实施例具体显示和描述了本申请，但是本领域的技术人员应该理解，在不脱离权利要求所限定的本申请的精神和范围的情况下，可以对其进行形式和细节上的各种改变。

Claims (18)

  1. 一种风力发电机组的载荷预测方法，其特征在于，所述载荷预测方法包括：
    获取与所有工况对应的标准相关数据；
    基于所述标准相关数据仿真计算分别与所有工况对应的载荷，得到相应的载荷数据；
    将所述标准相关数据作为载荷预测模型的输入，所述载荷数据作为所述载荷预测模型的输出，训练并建立所述载荷预测模型；
    获取风力发电机组的实际相关数据，通过所述载荷预测模型预测风力发电机组在相应工况下的载荷。
  2. 如权利要求1所述的载荷预测方法，其特征在于，所述标准相关数据包括风资源数据，
    其中，所述风资源数据包括以下参数中的至少一个：风速、风速偏差、湍流强度、风剪切、入流角、空气密度、偏航角。
  3. 如权利要求2所述的载荷预测方法，其特征在于，所述标准相关数据还包括风力发电机组的配置数据和风力发电机组的状态数据，
    其中，所述配置数据包括以下参数中的至少一个：风力发电机组的塔架高度、额定功率、叶轮直径，
    所述状态数据为风力发电机组处于预定状态时对应的状态参数值和/或叶轮方位角。
  4. 如权利要求1-3中任一项所述的载荷预测方法，其特征在于，基于所述标准相关数据仿真计算分别与所有工况对应的载荷，得到相应的载荷数据，包括：
    定义仿真计算的载荷输出通道，以确定计算结果为风力发电机组的预定部件在相应工况下的载荷，
    其中，所述预定部件包括以下项中的任意一项：风力发电机组的叶根、叶片截面、轮毂中心、偏航轴承、塔底、塔架截面。
  5. 如权利要求4所述的载荷预测方法，其特征在于，基于所述标准相关数据仿真计算分别与所有工况对应的载荷，得到相应的载荷数据，包括：
    所述载荷数据包括以下项中的至少一项：载荷的均值、载荷的标准偏差、 载荷的极大值、载荷的极小值、1HZ等效载荷。
  6. 如权利要求1所述的载荷预测方法，其特征在于，所述载荷预测模型通过以下方式进行验证：将用于训练所述载荷预测模型的标准相关数据输入到所述载荷预测模型，通过比较基于所述标准相关数据仿真计算得到的风力发电机组的载荷与所述载荷预测模型的输出结果来确定所述载荷预测模型是否通过验证，
    或者，任意选择与所述载荷预测模型对应的标准相关数据，将任意选择的标准相关数据输入到所述载荷预测模型，通过仿真计算获得与任意选择的标准相关数据对应的风力发电机组的载荷，通过比较通过仿真计算获得的载荷与所述载荷预测模型的输出结果来确定所述载荷预测模型是否通过验证。
  7. 如权利要求1所述的载荷预测方法，其特征在于，获取风力发电机组的实际相关数据，通过所述载荷预测模型预测风力发电机组在相应工况下的载荷，还包括：
    对预测的风力发电机组在相应工况下的载荷进行后处理，以用于工况评估。
  8. 如权利要求7所述的载荷预测方法，其特征在于，所述工况评估包括：疲劳工况评估或极限工况评估。
  9. 一种风力发电机组的载荷预测装置，其特征在于，所述载荷预测装置包括：
    标准数据获取模块，获取与所有工况对应的标准相关数据；
    载荷数据获取模块，基于所述标准相关数据仿真计算分别与所有工况对应的载荷，得到相应的载荷数据；
    载荷预测模型建立模块，将所述标准相关数据作为载荷预测模型的输入，所述载荷数据作为所述载荷预测模型的输出，训练并建立所述载荷预测模型；
    载荷预测模块，获取风力发电机组的实际相关数据，通过所述载荷预测模型预测风力发电机组在相应工况下的载荷。
  10. 如权利要求9所述的载荷预测装置，其特征在于，所述标准相关数据包括风资源数据，
    其中，所述风资源数据包括以下参数中的至少一个：风速、风速偏差、湍流强度、风剪切、入流角、空气密度、偏航角。
  11. 如权利要求10所述的载荷预测装置，其特征在于，所述标准相关数 据还包括风力发电机组的配置数据和风力发电机组的状态数据，
    其中，所述配置数据包括以下参数中的至少一个：风力发电机组的塔架高度、额定功率、叶轮直径，
    所述状态数据为风力发电机组处于预定状态时对应的状态参数值和/或叶轮方位角。
  12. 如权利要求9-11中任一项所述的载荷预测装置，其特征在于，载荷数据获取模块定义仿真计算的载荷输出通道，以确定计算结果为风力发电机组的预定部件在当前的工况下的载荷，
    其中，所述预定部件包括以下项中的任意一项：风力发电机组的叶根、叶片截面、轮毂中心、偏航轴承、塔底、塔架截面。
  13. 如权利要求12所述的载荷预测装置，其特征在于，所述载荷数据包括以下项中的至少一项：载荷的均值、载荷的标准偏差、载荷的极大值、载荷的极小值、1HZ等效载荷。
  14. 如权利要求9所述的载荷预测装置，其特征在于，所述载荷预测模型通过以下方式进行验证：将用于训练所述载荷预测模型的风力发电机组的标准相关数据输入到所述载荷预测模型，通过比较基于所述标准相关数据仿真计算得到的风力发电机组的载荷与所述载荷预测模型的输出结果来确定所述载荷预测模型是否通过验证，
    或者，任意选择与所述载荷预测模型对应的风力发电机组的标准相关数据，将任意选择的标准相关数据输入到所述载荷预测模型，通过仿真计算获得与任意选择的标准相关数据对应的风力发电机组的载荷，通过比较通过仿真计算获得的载荷与所述载荷预测模型的输出结果来确定所述载荷预测模型是否通过验证。
  15. 如权利要求9所述的载荷预测装置，其特征在于，所述载荷预测装置还包括：工况评估模块，对预测的风力发电机组在相应工况下的载荷进行后处理，以用于工况评估。
  16. 如权利要求15所述的载荷预测装置，其特征在于，所述工况评估包括：疲劳工况评估或极限工况评估。
  17. 一种存储有计算机程序的计算机可读存储介质，当所述计算机程序在被处理器执行时实现如权利要求1-8中的任意一项所述的风力发电机组的载荷预测方法。
  18. 一种计算装置，其特征在于，所述计算装置包括：
    处理器；
    存储器，存储有计算机程序，当所述计算机程序被处理器执行时，实现如权利要求1-8中的任意一项所述的风力发电机组的载荷预测方法。
PCT/CN2018/098003 2018-02-28 2018-08-01 风力发电机组的载荷预测方法和装置 WO2019165753A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810168969.8 2018-02-28
CN201810168969.8A CN110210044A (zh) 2018-02-28 2018-02-28 风力发电机组的载荷预测方法和装置

Publications (1)

Publication Number Publication Date
WO2019165753A1 true WO2019165753A1 (zh) 2019-09-06

Family

ID=67779033

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/098003 WO2019165753A1 (zh) 2018-02-28 2018-08-01 风力发电机组的载荷预测方法和装置

Country Status (2)

Country Link
CN (1) CN110210044A (zh)
WO (1) WO2019165753A1 (zh)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111475951A (zh) * 2020-04-09 2020-07-31 国网山东省电力公司电力科学研究院 一种热电机组工况分析方法
CN111505406A (zh) * 2020-03-16 2020-08-07 剑科云智(深圳)科技有限公司 一种配电柜、线材的监测方法
CN111767641A (zh) * 2020-05-29 2020-10-13 北京金风科创风电设备有限公司 风力发电机组极限载荷数据的处理方法和系统
CN112611584A (zh) * 2020-05-18 2021-04-06 北京金风慧能技术有限公司 风力发电机组的疲劳失效检测方法、装置、设备及介质
CN113361058A (zh) * 2020-03-05 2021-09-07 北京金风科创风电设备有限公司 确定风电场的代表风参数的方法和设备
CN113609622A (zh) * 2021-08-20 2021-11-05 浙江大学 风力发电机组的塔架载荷建模方法和装置
CN113780356A (zh) * 2021-08-12 2021-12-10 北京金水永利科技有限公司 基于集成学习模型的水质预测方法及系统
CN115983054A (zh) * 2023-03-21 2023-04-18 中车山东风电有限公司 一种风力发电机组的能力评估及载荷处理方法及终端机
CN116933663A (zh) * 2023-09-15 2023-10-24 中国航空工业集团公司金城南京机电液压工程研究中心 一种机载涡轮转子气动特性高保真模型建模方法及装置

Families Citing this family (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110594106B (zh) * 2019-10-15 2020-11-13 浙江运达风电股份有限公司 一种风电机组载荷在线预测方法、装置、设备、介质
CN113468711B (zh) * 2020-03-30 2024-02-06 金风科技股份有限公司 缩减风电机组的载荷工况的方法和设备
CN112109919B (zh) * 2020-04-30 2024-04-19 中国飞机强度研究所 一种强度试验加载点布局方法
CN112597601B (zh) * 2020-05-11 2022-09-16 河北新天科创新能源技术有限公司 一种用于风机不同轮毂高度塔筒极限载荷快速评估方法
CN111720271B (zh) * 2020-06-30 2022-01-25 国电联合动力技术有限公司 一种风电机组载荷在线预测的智能方法及风电机组
CN111997831B (zh) * 2020-09-01 2021-11-19 新疆金风科技股份有限公司 风电机组的载荷控制方法和装置
CN112131719B (zh) * 2020-09-07 2022-11-25 上海电气风电集团股份有限公司 风机载荷结果的获取方法及装置、计算机可读存储介质
CN112302886B (zh) * 2020-10-10 2022-05-31 上海电气风电集团股份有限公司 风电机组变桨系统载荷的自动测量方法、系统及计算机可读存储介质
CN112668124B (zh) * 2021-01-04 2023-04-25 上海电气风电集团股份有限公司 风力发电机组极限设计载荷的确定方法、装置及计算机可读存储介质
CN113094997B (zh) * 2021-04-19 2022-04-01 华北电力大学 一种风电机组运行模拟方法、装置、设备及存储介质
CN113669201B (zh) * 2021-09-15 2022-09-06 中国华能集团清洁能源技术研究院有限公司 一种风电机组极端条件下的极限载荷控制方法
CN117989050A (zh) * 2022-10-28 2024-05-07 金风科技股份有限公司 风力发电机组控制方法、装置及控制器
CN116146421A (zh) * 2023-03-08 2023-05-23 大唐凉山新能源有限公司 一种基于风机状态感知的智能控制方法及系统

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101476541A (zh) * 2008-12-26 2009-07-08 华锐风电科技有限公司 用于风力发电机组的独立变桨控制系统及控制方法
CN102588211A (zh) * 2012-02-29 2012-07-18 沈阳华人风电科技有限公司 一种风力发电机组全工况模型预测控制方法及系统
CN102708266A (zh) * 2012-06-12 2012-10-03 中国科学院工程热物理研究所 一种水平轴风力机叶片的极限载荷预测计算方法
US20140039843A1 (en) * 2012-07-31 2014-02-06 Universiti Brunei Darussalam Wind farm layout in consideration of three-dimensional wake

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101476541A (zh) * 2008-12-26 2009-07-08 华锐风电科技有限公司 用于风力发电机组的独立变桨控制系统及控制方法
CN102588211A (zh) * 2012-02-29 2012-07-18 沈阳华人风电科技有限公司 一种风力发电机组全工况模型预测控制方法及系统
CN102708266A (zh) * 2012-06-12 2012-10-03 中国科学院工程热物理研究所 一种水平轴风力机叶片的极限载荷预测计算方法
US20140039843A1 (en) * 2012-07-31 2014-02-06 Universiti Brunei Darussalam Wind farm layout in consideration of three-dimensional wake

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
HAN, HUALI: "Study on Modeling, Load Simulation and Load Optimization of Wind Turbine", SCIENCE -ENGINEERING (B) , CHINA DOCTORAL DISSERTATIONS FULL-TEXT DATABASE, 15 January 2016 (2016-01-15), pages C042 - 16, ISSN: 1674-022X *

Cited By (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113361058A (zh) * 2020-03-05 2021-09-07 北京金风科创风电设备有限公司 确定风电场的代表风参数的方法和设备
CN113361058B (zh) * 2020-03-05 2024-01-23 北京金风科创风电设备有限公司 确定风电场的代表风参数的方法和设备
CN111505406A (zh) * 2020-03-16 2020-08-07 剑科云智(深圳)科技有限公司 一种配电柜、线材的监测方法
CN111475951A (zh) * 2020-04-09 2020-07-31 国网山东省电力公司电力科学研究院 一种热电机组工况分析方法
CN112611584B (zh) * 2020-05-18 2023-06-02 北京金风慧能技术有限公司 风力发电机组的疲劳失效检测方法、装置、设备及介质
CN112611584A (zh) * 2020-05-18 2021-04-06 北京金风慧能技术有限公司 风力发电机组的疲劳失效检测方法、装置、设备及介质
CN111767641A (zh) * 2020-05-29 2020-10-13 北京金风科创风电设备有限公司 风力发电机组极限载荷数据的处理方法和系统
CN111767641B (zh) * 2020-05-29 2024-04-19 北京金风科创风电设备有限公司 风力发电机组极限载荷数据的处理方法和系统
CN113780356A (zh) * 2021-08-12 2021-12-10 北京金水永利科技有限公司 基于集成学习模型的水质预测方法及系统
CN113780356B (zh) * 2021-08-12 2023-08-08 北京金水永利科技有限公司 基于集成学习模型的水质预测方法及系统
CN113609622A (zh) * 2021-08-20 2021-11-05 浙江大学 风力发电机组的塔架载荷建模方法和装置
CN113609622B (zh) * 2021-08-20 2023-12-26 浙江大学 风力发电机组的塔架载荷建模方法和装置
CN115983054A (zh) * 2023-03-21 2023-04-18 中车山东风电有限公司 一种风力发电机组的能力评估及载荷处理方法及终端机
CN115983054B (zh) * 2023-03-21 2023-08-11 中车山东风电有限公司 一种风力发电机组的能力评估及载荷处理方法及终端机
CN116933663A (zh) * 2023-09-15 2023-10-24 中国航空工业集团公司金城南京机电液压工程研究中心 一种机载涡轮转子气动特性高保真模型建模方法及装置
CN116933663B (zh) * 2023-09-15 2023-12-08 中国航空工业集团公司金城南京机电液压工程研究中心 一种机载涡轮转子气动特性高保真模型建模方法及装置

Also Published As

Publication number Publication date
CN110210044A (zh) 2019-09-06

Similar Documents

Publication Publication Date Title
WO2019165753A1 (zh) 风力发电机组的载荷预测方法和装置
Preece et al. Probabilistic small-disturbance stability assessment of uncertain power systems using efficient estimation methods
CA2867253A1 (en) System and method for analyzing oscillatory stability in electrical power transmission systems
Müller et al. Damage assessment of floating offshore wind turbines using response surface modeling
WO2019070590A3 (en) Method and system for automating resource selection and building plan layout generation
Jasa et al. Effectively using multifidelity optimization for wind turbine design
Robertson et al. Assessment of wind parameter sensitivity on ultimate and fatigue wind turbine loads
CN106547695B (zh) 一种规模软件的测试系统及方法
CN112052604B (zh) 风机等效疲劳载荷的预测方法、系统、设备及可读介质
CN112287484B (zh) 一种基于矢量代理模型的复杂工程系统可靠性设计方法
Capaldo et al. Damping analysis of Floating Offshore Wind Turbine (FOWT): a new control strategy reducing the platform vibrations
Verdonck et al. Uncertainty quantification of structural blade parameters for the aeroelastic damping of wind turbines: a code-to-code comparison
Johnson et al. Balancing fatigue damage and turbine performance through innovative pitch control algorithm
Robertson et al. Assessment of wind parameter sensitivity on extreme and fatigue wind turbine loads
Tallman et al. An assessment of machine learning techniques for predicting turbine airfoil component temperatures, using FEA simulations for training data
Borodulin Validation of wind turbine generator stability models for wind generation interconnection studies
Verdonck et al. An open-source framework for the uncertainty quantification of aeroelastic wind turbine simulation tools
Cruse et al. Confidence interval simulation for systems of random variables
Natarajan et al. Determination of wind farm life consumption in complex terrain using ten-minute SCADA measurements
CN113361058A (zh) 确定风电场的代表风参数的方法和设备
CN113468711B (zh) 缩减风电机组的载荷工况的方法和设备
Aguilera et al. Experiences in power system multi-domain modeling and simulation with modelica & FMI: The case of gas power turbines and power systems
Küstner et al. Design for noise reduction–The architecture of an engineering assistance system for the development of noise-reduced rotating systems
CN117010259B (zh) 风电机组门洞结构的优化方法
Aguilera et al. Experiences in power systems multi-domain modeling and simulation with Modelica & FMI

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18907757

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18907757

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 20/01/2021)

122 Ep: pct application non-entry in european phase

Ref document number: 18907757

Country of ref document: EP

Kind code of ref document: A1