Offshore wind power extreme environment parameter calculation method based on artificial intelligence

Abstract

The invention discloses an offshore wind power extreme environment parameter calculation method based on artificial intelligence, and the method comprises the following steps: S1, obtaining meteorological data of an offshore wind power design region, and selecting an artificial intelligence time sequence prediction model for training; s2, inputting a historical time sequence into the time sequence prediction model to obtain wind and wave time sequence data of the offshore wind power in the future operation period; fitting wind speed, wave height and wave period joint distribution by using a maximum likelihood method; and S3, based on the joint probability distribution of the wind speed, the wave height and the wave period, obtaining an environment contour line by using an inverse first-order reliability method and approximate normal distribution transformation, and further obtaining offshore wind power extreme environment data in a selected regression period. According to the method, extreme weather frequency change caused by sea surface temperature change and the requirement that an offshore wind power large fan needs to be designed more strictly are effectively considered, and more accurate and strict extreme environment design parameters are provided for offshore wind power design.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of wind power, and particularly relates to a method for calculating extreme environmental parameters of offshore wind power based on artificial intelligence. BACKGROUND

[0002] In recent years, offshore wind power has developed rapidly and shows a trend of high-power and deep-sea offshore wind turbines. However, with the increase of rated power, the rotor diameter, cabin structure size and tower height are gradually increased. Therefore, for large structures, the dynamic response under extreme environment is a problem that must be considered in the design process.

[0003] The extreme environmental parameters often consider the 10-minute extreme wind speed in the design return period, which is often obtained by fitting the joint distribution probability of past wind and wave time series data, calculating the wind and wave height and wave period curve corresponding to the exceedance probability. However, due to the change of sea surface temperature with global warming, the frequency of extreme weather changes, and the meteorological data is also quite different from the past time series, and the design of offshore wind power large structure should also be more stringent. These factors may lead to inaccurate and unsafe extreme environment calculated. SUMMARY

[0004] In order to solve the above problems, the application provides a method for calculating extreme environmental parameters of offshore wind power based on artificial intelligence, which can predict wind and wave time series data during the operation of offshore wind power, and can calculate more stringent extreme environmental design parameters with 10-year, 50-year and 100-year return periods, and provide more accurate and stringent safety offshore wind power design extreme environmental parameters.

[0005] In order to achieve the above purpose, the technical scheme provided by the application is a method for calculating extreme environmental parameters of offshore wind power based on artificial intelligence, comprising the following steps: S1, obtaining meteorological data of offshore wind power design area, preprocessing the meteorological data to obtain wind and wave time series data set, performing feature extraction and labeling data set, training the suitable time series prediction artificial intelligence model using the labeled data set, and performing hyperparameter optimization.

[0006] S2, then inputting the historical data into the artificial intelligence model to obtain the wind and wave time series data during the future operation of offshore wind power, and on this basis, using the maximum likelihood estimation method to fit the joint distribution of wind speed, wave height and wave period.

[0007] S3, on the basis of the joint distribution of wind speed, wave height and wave period, the environmental contour line is obtained by Rosenblatt and Iform transformation, and the wind speed corresponding to the selected return period and the corresponding wave height and wave period contour line are calculated, and the extreme environmental parameters of offshore wind power are extracted.

[0008] Preferably, the preprocessing of meteorological data in the S1 offshore wind power design area includes data smoothing, denoising or normalization operations.

[0009] Preferably, the suitable time series model in S1 is selected from any one of a recurrent neural network (RNN), a long short-term memory network (LSTM), a convolutional neural network (CNN) and a neural network framework based on an attention mechanism.

[0010] Preferably, the maximum likelihood estimation method in S2 calculates the probability distribution parameters by knowing the probability distribution of wind speed, wave height and wave period, specifically: ; ; in, are the probability distribution function parameters, is a sample with n values ​​from the distribution, is the probability distribution function, P is the probability, When the probability P is maximum value.

[0011] Preferably, the joint probability distribution of the three parameters of wind speed, wave height and wave period in S2 is calculated as follows: ; in, is the three-parameter joint probability distribution of wind speed, wave height and wave period, is the marginal distribution of wind speed, is the conditional probability distribution of wave height under wind speed, is the conditional probability distribution of wave period under wind speed and wave height.

[0012] Preferably, the marginal distribution of wind speed It obeys the two-parameter Weibull distribution, and its cumulative probability distribution is calculated as follows: ; in, is the shape parameter, is the scale parameter, is the average wind speed.

[0013] Preferably, the conditional probability distribution of wave height under wind speed conditions is It obeys the two-parameter Weibull distribution, and its cumulative probability distribution is calculated as follows: ; in, is the shape parameter that varies with wind speed, is a scale parameter that varies with wind speed, The average wave height.

[0014] Preferably, the probability of conditional wave cycle under wind speed and wave height It obeys the lognormal distribution, and its cumulative distribution probability is calculated as follows: ; in, is the location parameter of the lognormal distribution that varies with wind speed and wave height, is the shape parameter of the lognormal distribution that varies with wind speed and wave height.

[0015] Preferably, the Rosenblatt transformation of S3 is as follows: ; ; ; in, is the average wind speed, is the average wave height, is the wave period, for The corresponding standard normal distribution variable, for The corresponding standard normal distribution variable, for The corresponding standard normally distributed variable.

[0016] Preferably, the Iform method of S3 is specifically: ; in, To establish a normal distribution, is the regression cycle, is the observation interval.

[0017] (1) The present invention collects meteorological data of the offshore wind power design area, pre-processes and features the data to obtain wind and wave time series data of the offshore wind power design area. By comparing the training effect of the time series prediction deep learning model, a suitable deep learning model is selected and input into the historical data to obtain the wind and wave time series data during the operation of the offshore wind power. This effectively takes into account the impact of the change of ocean surface temperature with global warming on the wind and wave time series distribution of the offshore wind power operation.

[0018] (2) The present invention addresses the stricter design requirements for large offshore wind turbines and the changing frequency of extreme weather conditions. Based on the wind and wave time series data during the operation of offshore wind turbines, the joint distribution of wind speed, wave height, and wave period is fitted. The environmental contour line is then obtained through Rosenblatt and Iform transformations. The wind speed corresponding to the 10-year, 50-year, and 100-year regression periods, as well as the corresponding wave height and wave period contour lines, are calculated. Ultimately, more accurate and rigorous offshore wind turbine extreme environmental parameters are obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a flow chart of a method for calculating extreme environmental parameters of offshore wind power based on artificial intelligence in an embodiment of the present invention; Figure 2 This is an architecture diagram of an artificial intelligence-based offshore wind power extreme environment parameter calculation method in an embodiment of the present invention; Figure 3 This is a discretized distribution diagram of wind speed, wave height, and wave period in an embodiment of the present invention; where (a) wind speed; (b) wave height; (c) wave period; (d) joint distribution of the three parameters; Figure 4 The wave height and wave period corresponding to the environmental contour lines and maximum wind speed of each regression period in the embodiment of the present invention are shown; among them, (a) the regression period is 1 year; (b) the regression period is 10 years; (c) the regression period is 50 years; and (d) the regression period is 100 years. DETAILED DESCRIPTION

[0020] The technical solution of the present invention is further described below with reference to the accompanying drawings and specific embodiments.

[0021] Unless otherwise defined, technical or scientific terms used in the present invention shall have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs.

[0022] like Figure 1-4 As shown, the present invention provides an artificial intelligence-based method for calculating extreme environmental parameters of offshore wind power, comprising the following steps: S1. Obtain meteorological data for the offshore wind power design area, smooth, denoise, and normalize the meteorological data to obtain a wind and wave time series dataset. Perform feature extraction and annotation on the dataset, and divide the dataset into training, validation, and test sets in a ratio of 7:1:2. Select the Transformer model as the target model and perform hyperparameter tuning to ensure that the error between the model output and the true value is within 5%.

[0023] S2, input historical data into the Transformer model to obtain wind and wave time series data during the future operation of the offshore wind power. The maximum likelihood method is used to fit the joint distribution of the three parameters of wind speed, wave height and wave period, and the fitting accuracy is ensured, and the fitting results are as shown in Figure 3 .

[0024] wherein the basic principle of the maximum likelihood method is to calculate the probability distribution parameters by knowing the probability distribution of wind speed, wave height and wave period, and the calculation method is: ; ; wherein, is the parameter of the probability distribution function, is the sampling with n values in the distribution, is the probability distribution function, P is the probability, is the value of when the probability P is maximum.

[0025] The joint probability distribution of the three parameters of wind speed, wave height and wave period is calculated by: ; wherein, is the joint probability distribution of the three parameters of wind speed, wave height and wave period, is the marginal distribution of wind speed, is the conditional probability distribution of wave height under wind speed, is the conditional probability distribution of wave period under wind speed and wave height.

[0026] The marginal distribution of wind speed obeys the two-parameter Weibull distribution, and the cumulative probability distribution calculation method is: ; wherein, is the shape parameter, is the scale parameter, is the average wind speed.

[0027] The conditional probability distribution of wave height under wind speed obeys the two-parameter Weibull distribution, and the cumulative probability distribution calculation method is: ; wherein, is the shape parameter varying with wind speed, is the scale parameter varying with wind speed, is the average wave height.

[0028] wherein, the conditional probability of wave period under wind speed and wave height It obeys the lognormal distribution, and its cumulative distribution probability is calculated as follows: ; in, is the location parameter of the lognormal distribution that varies with wind speed and wave height, is the shape parameter of the lognormal distribution that varies with wind speed and wave height.

[0029] S3. Based on the joint distribution of wind speed, wave height and wave period, the environmental contour line is obtained through Rosenblatt and Iform transformation.

[0030] The purpose of Rosenblatt transformation is to transform the wind speed probability distribution, wave height conditional probability distribution and wave period conditional probability distribution into the standard normal parameter space. The process is as follows: ; ; ; in, is the average wind speed, is the average wave height, is the wave period, for The corresponding standard normal distribution variable, for The corresponding standard normal distribution variable, for The corresponding standard normally distributed variable.

[0031] The purpose of the Iform method is to establish the sphere radius corresponding to the return period contour: ; in, To establish a normal distribution, is the regression cycle, is the observation interval.

[0032] In this example, the observation interval is 1 hour, and the wind speed corresponding to the 1-year regression period, 10-year regression period, 50-year regression period and 100-year regression period are calculated respectively, and the corresponding wave height and wave period contour lines are as follows: Figure 4 As shown in Table 1, the extreme environmental parameters of offshore wind power are extracted.

[0033] Table 1 Extreme environmental parameters of offshore wind power extracted using the method of the present invention

[0034] The application adopts an offshore wind power extreme environment parameter calculation method based on artificial intelligence, obtains wind and wave time series during the operation of offshore wind power through the past meteorological data of the design area of offshore wind power, and uses a time series prediction deep learning model. On the basis of the wind and wave time series during the operation of offshore wind power, the maximum likelihood method is used to fit the joint distribution probability of the three parameters of wind speed, wave height and wave period. The wind speed, wave height and wave period contour lines under different regression periods are calculated through the Rosenblatt method and the Iform method, and the wave height and wave period contour lines corresponding to the wind speed are calculated. The frequency influence of the change of sea temperature on extreme weather during the design of offshore wind power can be considered, and the more strict extreme environment parameter design for large structures.

[0035] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application but not to limit it, although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can still be modified or replaced by equivalents, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present application.

Claims

1. A method for calculating extreme environmental parameters of offshore wind power based on artificial intelligence, characterized by: The following steps are involved: S1. Obtain meteorological data for the offshore wind power design area, preprocess the meteorological time series data to form wind and wave time series data sets, annotate the data sets through feature extraction, divide the data sets into training sets, validation sets, and test sets, use the annotated data sets to train the time series deep learning model, and perform hyperparameter tuning on the model; S2. Input historical data into the trained model to obtain the wind and wave time series during the future operation of offshore wind power; The maximum likelihood estimation method is used to fit the three-parameter joint distribution of wind speed, wave height and wave period; S3. Obtain environmental contours through Rosenblatt and Iform transformations, calculate the wind speed and wave height and wave period contours corresponding to the selected regression period, and extract extreme environmental parameters of offshore wind power.

2. The method for calculating extreme environmental parameters of offshore wind power based on artificial intelligence according to claim 1, characterized in that: The preprocessing in S1 includes smoothing, denoising and normalization operations.

3. The artificial intelligence-based offshore wind power extreme environment parameter calculation method according to claim 1 is characterized in that: The time series model in S1 is selected from any one of a recurrent neural network, a long short-term memory network, a convolutional neural network, and a neural network framework based on an attention mechanism.

4. The artificial intelligence-based offshore wind power extreme environment parameter calculation method according to claim 1 is characterized in that: The maximum likelihood estimation method in S2 calculates the probability distribution parameters by knowing the probability distribution of wind speed, wave height and wave period, specifically: ； ； in, are the probability distribution function parameters, Is a distribution with A sample of values, is the probability distribution function, is the probability, For probability Maximum value.

5. The method for calculating extreme environmental parameters of offshore wind power based on artificial intelligence according to claim 1, characterized in that: The calculation method for the joint probability distribution of the three parameters of wind speed, wave height and wave period in S2 is: ； in, is the three-parameter joint probability distribution of wind speed, wave height and wave period, is the marginal distribution of wind speed, is the conditional probability distribution of wave height under wind speed conditions, is the conditional probability of the wave period under wind speed conditions and wave height conditions.

6. The artificial intelligence-based offshore wind power extreme environment parameter calculation method according to claim 5, characterized in that: Marginal distribution of the wind speed It obeys the two-parameter Weibull distribution, and its cumulative probability distribution is calculated as follows: ； in, is the shape parameter, is the scale parameter, is the average wind speed.

7. The artificial intelligence-based offshore wind power extreme environment parameter calculation method according to claim 5, characterized in that: Conditional probability distribution of wave height under the stated wind speed conditions It obeys the two-parameter Weibull distribution, and its cumulative probability distribution is calculated as follows: ； in, is the shape parameter that varies with wind speed, is a scale parameter that varies with wind speed, The average wave height.

8. The artificial intelligence-based offshore wind power extreme environment parameter calculation method according to claim 5, characterized in that: The probability of conditional wave cycle under the wind speed and wave height It obeys the lognormal distribution, and its cumulative distribution probability is calculated as follows: ； in, is the location parameter of the lognormal distribution that varies with wind speed and wave height, is the shape parameter of the lognormal distribution that varies with wind speed and wave height.

9. The method for calculating extreme environmental parameters of offshore wind power based on artificial intelligence according to claim 1 is characterized in that: The Rosenblatt transformation in S3 is used to transform the wind speed probability distribution, wave height conditional probability distribution and wave period conditional probability distribution into the standard normal parameter space. The process is as follows: ； ； ； in, is the average wind speed, is the average wave height, is the wave period, for The corresponding standard normal distribution variable, for The corresponding standard normal distribution variable, for The corresponding standard normally distributed variable.

10. The artificial intelligence-based offshore wind power extreme environment parameter calculation method according to claim 9 is characterized in that: The Iform method in S3 is used to establish the sphere radius corresponding to the return period contour: ； in, To establish a normal distribution, is the regression cycle, is the observation interval.

Images