Wind resource power season prediction method, system and equipment based on numerical mode

Abstract

The invention relates to the technical field of climate prediction, and discloses a wind resource power season prediction method, system and equipment based on a numerical mode, and the method comprises the steps: obtaining preset data, carrying out the preprocessing of the preset data, obtaining input data, assimilating the input data into a climate system mode, and generating an initial field, the climate system mode comprises an atmospheric component, an ocean component, a land surface component and a sea ice component which realize data interaction through a coupler, and the atmospheric component adopts a spherical cubic grid power frame based on a finite volume. The initial field is disturbed, set numerical prediction is carried out, a historical return test and a real-time seasonal prediction test are designed and operated, and finally a wind resource power seasonal prediction result is obtained. And finally, reliable product prediction support and decision basis are provided for making an operation and maintenance plan of the wind power plant, power market transaction and power generation benefit measurement and calculation of the wind power plant.

Description

Technical Field

[0001] The present invention relates to the technical field of climate prediction, and in particular to a method, system and device for seasonal prediction of wind resource power based on numerical models. Background Art

[0002] As an important component of renewable energy, wind resources have been widely used and developed rapidly in the transformation of the world's energy system under the situation of mitigating and responding to climate change due to their clean and safe characteristics. The volatility, intermittency and seasonal changes of wind resources often lead to changes in wind power output, which have many impacts on the power quality, safe and stable operation and economic benefits of the power system. At present, the prediction of wind resources is mainly focused on ultra-short-term and short-term scale predictions, which are used to achieve wind power generation and operation scheduling in the current and next few days, but there is still a lack of prediction information on a longer time scale. Seasonal prediction is to predict the future climate one to several months in advance. For the wind power field, the prediction of this time scale can provide a basis for the formulation of operation and maintenance plans for wind farms, the calculation of wind farm power generation benefits, power market transactions, and disaster prevention and mitigation.

[0003] With the widespread application of numerical models in meteorological forecasting, it has become possible to use numerical models to establish a dynamic forecasting system. At present, the mainstream international seasonal forecasting systems are all built based on numerical models. In recent years, due to the improvement of climate models, assimilation technology and ensemble forecasting technology, the predictability of the climate system on a seasonal scale has also been steadily improving. However, some problems have still arisen. For example, the current seasonal forecasting system often focuses on temperature and precipitation, and there are fewer seasonal forecasting systems for wind resources, which are a major demand of the wind power industry. The dynamic forecasting system based on numerical models has certain defects in the selection of atmospheric model dynamic framework, parameterization uncertainty of the climate system, observation data quality and effective range, and insufficient understanding of complex sea-air interactions under climate change, resulting in low computational efficiency, low prediction accuracy, and large uncertainty in the prediction system, which limits the accuracy of wind resource dynamic forecasting. Summary of the invention

[0004] In view of this, the present invention provides a method, system and device for seasonal prediction of wind resource power based on numerical models to solve the technical problem that wind resource power prediction is not accurate enough.

[0005] In a first aspect, the present invention provides a method for seasonal prediction of wind resource power based on a numerical model, comprising: acquiring preset data and preprocessing the preset data to obtain input data; assimilating the input data into a climate system model to generate an initial field, wherein the climate system model includes an atmospheric component, an ocean component, a land component and a sea ice component that realize data interaction through a coupler, and the atmospheric component adopts a spherical cubic grid dynamic framework based on a finite volume; performing numerical prediction based on the initial field to obtain a seasonal prediction result of wind resource power.

[0006] The method for seasonal prediction of wind resource power based on numerical model of the present invention obtains input data by acquiring preset data and preprocessing the preset data, assimilates the input data into the climate system model, and generates an initial field, wherein the climate system model includes an atmospheric component, an ocean component, a land component and a sea ice component that realize data interaction through a coupler, the atmospheric component adopts a spherical cubic grid dynamic framework based on a finite volume, performs numerical prediction according to the initial field, and obtains a seasonal prediction result of wind resource power, and constructs a climate system model using atmospheric components, ocean components, land components and sea ice components, and uses a spherical cubic grid based on a finite volume method as a dynamic framework in the atmospheric component. On the one hand, the spherical cubic grid can better realize the accurate characterization and simulation of physical processes on the spherical surface, and at the same time, the grid on the spherical surface is regular and uniform, which reduces the complexity of calculation on the spherical surface, making the model calculation simpler and more efficient, and on the other hand, the finite volume method can improve the stability of numerical model calculation, improve the accuracy and reliability of model simulation, and at the same time, it has high computational efficiency when solving partial differential equations, and has great advantages in solving problems under complex boundary conditions. This enables the prediction system based on the finite volume spherical cubic grid as the power framework to achieve efficient computing in a parallel computing environment without losing computing accuracy, thereby improving the accuracy of wind resource power forecasts, and ultimately providing reliable forecast product support and decision-making basis for wind farm operation and maintenance plans, electricity market transactions, and wind farm power generation benefit estimation.

[0007] Optionally, assimilating the input data into the climate system model to generate an initial field includes: based on a time-lag perturbation method, adding perturbations at different times in the process of assimilating the input data into the climate system model to generate a number of different initial fields; correspondingly, performing numerical predictions based on the initial fields to obtain seasonal prediction results of wind resource power includes: performing numerical predictions based on different initial fields respectively to form an ensemble forecast with a number of ensemble members; and performing data post-processing on the ensemble forecast to obtain seasonal prediction results of wind resource power.

[0008] The time-delayed perturbation method is used to generate different initial fields, which increases the diversity of predictions. The ensemble forecast method can be used to evaluate the uncertainty of the prediction results and improve the reliability and accuracy of the prediction.

[0009] Optionally, the time-lag-based perturbation method adds perturbations at different moments in the process of assimilating the input data into the climate system model to generate a number of different initial fields, including: determining the maximum number of ensemble members based on the assimilation time period and the time integration step; selecting the number of ensemble members based on the maximum number of ensemble members, and determining the interval frequency of the time integration step required for the perturbation based on the number of ensemble members and the maximum number of ensemble members; adding perturbations at the moments corresponding to each ensemble member in the process of assimilating the input data into the climate system model to generate a number of different initial fields corresponding to the ensemble member, wherein the moment corresponding to the ensemble member is the product of the interval frequency, the time integration step and the sequence number of the ensemble member.

[0010] By determining the maximum number of ensemble members and the frequency of the time integration steps required for the perturbation, the initial field is systematically generated. This method can more precisely control the uncertainty in the prediction process and improve the accuracy of the prediction.

[0011] Optionally, the model is started for 3 years to achieve equilibrium through internal adjustments, and a historical return experiment with 24 ensemble members is designed to verify the prediction level of the wind resource seasonal prediction system, and finally a real-time seasonal prediction of wind resources with multiple ensemble members is carried out.

[0012] The data post-processing of the ensemble forecast to obtain a wind resource dynamic seasonal forecast result includes: dividing the ensemble forecast into a preset number of spherical surfaces of to-be-processed data based on a finite volume spherical cubic grid dynamic framework; interpolating and merging the preset number of spherical surfaces of to-be-processed data into global longitude and latitude grid data through an interpolation conversion program; extracting variables corresponding to wind resources from the global longitude and latitude grid data, and averaging the variables corresponding to different ensemble members to obtain output data; and outputting the output data.

[0013] Through interpolation conversion and data post-processing, the ensemble forecast is converted into global latitude and longitude grid data, making the prediction results more intuitive and easy to understand. At the same time, by averaging the variables of different ensemble members, random errors can be reduced and the stability of the prediction results can be improved.

[0014] Optionally, after performing data post-processing on the ensemble forecast to obtain a seasonal forecast result of wind resource power, the method further includes: calculating a time correlation coefficient, a normalized root mean square error, and a mean absolute error in a deterministic forecast detection index according to the seasonal forecast result of wind resource power, and the calculation formula is:

[0015]

[0016] In the formula, ACC represents the time correlation coefficient, nRMSE represents the normalized root mean square error, MAE represents the mean absolute error, i represents time, N represents the total number of time samples, and F i represents the seasonal forecast result of wind resource power, O i Represents reference data, and They represent the seasonal prediction results of wind resource power and the climate state corresponding to the reference data respectively; the relative operating characteristic method is used to evaluate the probability forecast skill score to obtain the hit rate and false alarm rate, where the hit rate and false alarm rate calculation formula are:

[0017]

[0018] In the formula, H R m Indicates hit rate, FAR m represents the false alarm rate, j represents the number of bins, M represents the number of set members, YO j and NO j They respectively represent the number of times the corresponding observations occurred and the number of times they did not occur in all individual cases of the distribution of set members corresponding to the bin number j in the seasonal forecast results of wind resource power; the deterministic forecast detection of the seasonal forecast results of wind resource power is performed according to the time correlation coefficient, the normalized root mean square error and the mean absolute error, and the probabilistic forecast detection of the seasonal forecast results of wind resource power is performed according to the hit rate and the false alarm rate.

[0019] By calculating the deterministic forecast detection index and the probabilistic forecast skill score, the forecast results can be quantitatively evaluated to help users understand the accuracy and reliability of the forecast and provide decision support for the development and utilization of wind resources.

[0020] Optionally, after performing a deterministic forecast test on the wind resource power seasonal forecast results according to the time correlation coefficient, the normalized root mean square error and the mean absolute error, and performing a probabilistic forecast test on the wind resource power seasonal forecast results according to the hit rate and the false alarm rate, it also includes: visually displaying the deterministic forecast test results and the probabilistic forecast test.

[0021] This method enables non-professional users to intuitively understand the prediction results, improving the acceptability and ease of use of the prediction results.

[0022] Optionally, assimilating the input data into a climate system model to generate an initial field comprises: based on an incremental analysis update process, using a time-varying relaxation approximation method to assimilate the input data as initial conditions into a climate system model to generate an initial field.

[0023] By adopting the time-varying relaxation approximation method, the initial conditions can be adjusted more flexibly on the basis of incremental analysis, adapting to the dynamic changes of the climate system model and improving the accuracy of the initial field.

[0024] Optionally, the input data includes historical input data and current input data. Correspondingly, the numerical prediction is performed according to the initial field to obtain the wind resource power seasonal prediction result, including: performing a historical return test according to the initial field corresponding to the historical input data to obtain the wind resource power seasonal prediction result for the historical time period; performing a real-time prediction test according to the initial field corresponding to the current input data to obtain the wind resource power seasonal prediction result for the future time period.

[0025] Conducting historical return tests and real-time prediction tests separately can evaluate the prediction performance of the model in different time periods, which helps to further optimize model parameters and improve prediction accuracy.

[0026] In a second aspect, the present invention provides a wind resource power prediction system based on a numerical model, comprising: a data preprocessing module, used to obtain preset data and preprocess the preset data to obtain input data; an initial field generation module, used to assimilate the input data into a climate system model to generate an initial field, wherein the climate system model includes an atmospheric component, an ocean component, a land component and a sea ice component that realize data interaction through a coupler, and the atmospheric component adopts a spherical cubic grid power framework based on a finite volume; a wind resource prediction module, used to perform numerical prediction based on the initial field to obtain a seasonal prediction result of wind resource power.

[0027] In a third aspect, the present invention provides a computer device comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the numerical model-based wind resource power seasonal prediction method of the above-mentioned first aspect or any corresponding embodiment thereof by executing the computer instructions. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0029] Figure 1 It is a flow chart of a method for seasonal prediction of wind resource power based on a numerical model in an embodiment of the present invention;

[0030] Figure 2 is a schematic diagram of a flow chart of preprocessing preset data in an embodiment of the present invention;

[0031] Figure 3 is a schematic diagram of a process for generating an initial field in an embodiment of the present invention;

[0032] Figure 4 is a schematic flow chart of a relaxation approximation method in an embodiment of the present invention;

[0033] Figure 5 It is a schematic diagram of the process of generating ensemble forecast in the implementation of the present invention;

[0034] Figure 6 It is a schematic diagram of the design of the historical return experiment and the real-time prediction experiment in the implementation of the present invention;

[0035] Figure 7 It is a schematic diagram of the process of data post-processing in the implementation of the present invention;

[0036] Figure 8 It is a design schematic diagram of visual display in the implementation of the present invention;

[0037] Fig. 9 is a structural block diagram of a wind resource power prediction system based on a numerical model in an embodiment of the present invention;

[0038] Fig.10 It is a schematic diagram of the hardware structure of a computer device in an embodiment of the present invention. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0040] With the widespread application of numerical models in meteorological forecasting, it has become possible to use numerical models to establish dynamic forecasting systems. At present, the mainstream seasonal forecasting systems are all built based on numerical models. In recent years, due to the improvement of climate models, assimilation technology and ensemble forecasting technology, the predictability of the climate system on the seasonal scale has also been steadily improved, but some problems have still emerged.

[0041] First, most of the current seasonal prediction systems are built to focus on temperature, precipitation, large-scale tropical circulation, etc., but other climate service areas (such as wind resources) are still lacking.

[0042] Secondly, today's seasonal forecasting system still faces certain challenges in business applications, and has high requirements for the degree of refinement, timeliness, and update and iteration capabilities of the forecasting system. For example, when obtaining higher-resolution forecast results, on the one hand, the horizontal resolution of the model can be adjusted, but the computing resources consumption is extremely large. On the other hand, local high resolution can be achieved through statistical or dynamic downscaling, but there will also be phenomena such as systematic deviations and boundary conditions of different models when nesting numerical models for downscaling, thereby amplifying the forecast error.

[0043] In addition, the inherent bias of the model is due to the current lack of understanding of the complex climate system and the inadequate description of the physical equations in the real Earth system, such as the atmosphere, ocean, land, and the interactions between the various components. This is a common problem for all numerical models.

[0044] Based on this, the present invention proposes a method for seasonal prediction of wind resource power based on a numerical model to solve at least one of the above problems.

[0045] According to an embodiment of the present invention, an embodiment of a method for seasonal prediction of wind resource power based on a numerical model is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0046] In this embodiment, a method for seasonal prediction of wind resource power based on numerical model is provided, which can be used in mobile terminals such as computers, mobile phones, etc. Figure 1 As shown, the process includes the following steps:

[0047] Step S101, obtaining preset data and preprocessing the preset data to obtain input data.

[0048] Specifically, the preset data includes historical and real-time observation data and reanalysis data. Through shell, cshell, python and NCL scripts, the historical and real-time observation data and reanalysis data are collected and processed into a form that can drive the climate system model. Figure 2 As shown in the figure, the specific implementation process includes:

[0049] (1) Download and collect observational and reanalysis data as well as necessary data for driving climate system models (e.g. global carbon dioxide concentration, topographic data, etc.);

[0050] (2) Convert observation and reanalysis data into NC format latitude and longitude grid data;

[0051] (3) Extract the variables required to drive the climate system model (including air temperature, meridional wind, zonal wind, specific humidity, surface pressure, sea level pressure, geopotential height, and surface temperature), and rename the variable names according to the requirements of the climate system model.

[0052] (4) The vertical pressure coordinates of the variables with vertical dimension are converted into vertical mixed coordinates.

[0053] Step S102, assimilate the input data into the climate system model to generate an initial field, wherein the climate system model includes an atmospheric component, an ocean component, a land component and a sea ice component that realize data interaction through a coupler, and the atmospheric component adopts a spherical cubic grid dynamic framework based on a finite volume.

[0054] Specifically, Figure 3 and Figure 4 As shown, based on the incremental analysis update process, the time-varying relaxation approximation method (IAU-TVnudging) is used to assimilate the input data into the climate system model as initial conditions to generate the initial field.

[0055] In the current version of the climate system model, only atmospheric data are assimilated as the initial field. The ocean, sea ice and land surface components in the climate system model will be adjusted following the atmospheric components due to the interaction process within the climate system model. The observation / reanalysis data are introduced into the current climate system model through the IAU-TVnudging method. The specific implementation method is as follows:

[0056] At each iteration of the model, the increment Δx is analyzed i By gradually adding the model integration process within a fixed time window and considering the time evolution of the assimilation increment within the assimilation window, the assimilation information is continuously introduced to gradually correct the initialization bias of the model and reduce the imbalance between the model and the assimilated observations.

[0057]

[0058] where x i represents any variable, subscript i represents time, and Represents the variable x i The use of a time-varying relaxation approximation method allows for more flexible adjustment of initial conditions based on incremental analysis, adapting to the dynamic changes in climate system models and improving the accuracy of initial fields.

[0059] Step S103, performing numerical prediction based on the initial field to obtain seasonal prediction results of wind resource power.

[0060] In the climate system model, numerical analysis methods are used to analyze and calculate the initial field to obtain the seasonal prediction results of wind resource dynamics.

[0061] The method for seasonal prediction of wind resource power based on numerical model in the embodiment of the present invention obtains input data by acquiring preset data and preprocessing the preset data, assimilates the input data into the climate system model, and generates an initial field, wherein the climate system model includes an atmospheric component, an ocean component, a land component, and a sea ice component that realize data interaction through a coupler, and the atmospheric component adopts a spherical cubic grid dynamic framework based on a finite volume, performs numerical prediction according to the initial field, and obtains a seasonal prediction result of wind resource power, and uses the atmospheric component, the ocean component, the land component, and the sea ice component to construct a climate system model, and uses a spherical cubic grid based on a finite volume method as a dynamic framework in the atmospheric component. On the one hand, the spherical cubic grid can better realize the accurate characterization and simulation of physical processes on the spherical surface, and at the same time, the grid on the spherical surface is regular and uniform, which reduces the complexity of calculation on the spherical surface, making the model calculation simpler and more efficient, and on the other hand, the finite volume method can improve the stability of numerical model calculation, improve the accuracy and reliability of model simulation, and at the same time, it has high computational efficiency when solving partial differential equations, and has great advantages in solving problems under complex boundary conditions. This enables the prediction system based on the finite volume spherical cubic grid as the power framework to achieve efficient computing in a parallel computing environment without losing computing accuracy, thereby improving the accuracy of wind resource power forecasts, and ultimately providing reliable forecast product support and decision-making basis for wind farm operation and maintenance plans, electricity market transactions, and wind farm power generation benefit estimation.

[0062] In one embodiment, step S102, assimilating the input data into the climate system model to generate an initial field, includes:

[0063] Step S1021, based on the time-delay perturbation method, perturbations are added at different times in the process of assimilating the input data into the climate system model to generate several different initial fields.

[0064] Correspondingly, step S103 performs numerical prediction based on the initial field to obtain the seasonal prediction result of wind resource power, including:

[0065] Step S1031, performing numerical predictions according to different initial fields to form an ensemble forecast having a number of ensemble members;

[0066] Step S1032, post-processing the ensemble forecast data to obtain the seasonal forecast result of wind resource power.

[0067] Specifically, step S1021 includes:

[0068] Step a1, determining the maximum number of set members according to the assimilation time period and the time integration step;

[0069] Step a2, selecting the number of set members based on the maximum number of set members, and determining the interval frequency of the time integration step required for the disturbance according to the number of set members and the maximum number of set members;

[0070] Step a3, adding disturbances at the time corresponding to each ensemble member in the process of assimilating the input data into the climate system model, generating several different initial fields corresponding to the ensemble members, where the time corresponding to the ensemble member is the product of the interval frequency, the time integration step and the serial number of the ensemble member.

[0071] It should be understood that the uncertainty of weather-climate forecasts is caused by the nonlinearity and complexity of the climate system, as well as the uncertainty of the initial values ​​and numerical models themselves. In order to reduce the uncertainty of the forecast results, it is inevitable to switch from a single-member deterministic forecast to a multi-member ensemble forecast. Based on the time-delay perturbation method, the initial field generated by the assimilation cycle from the start time of the initial operation of the model to a certain time is used. According to the model time integration step, multiple different initial fields are generated from this continuously evolving meteorological information, thereby forming an ensemble forecast with multiple ensemble members. Maximum number of ensemble members ENS max , the frequency of the time integration step required for the disturbance T d and the generation time T of each ensemble member in the climate system model ENS The calculation formula is as follows:

[0072]

[0073] T ENS =T d ×ENS i ×dt

[0074] Among them, ENS max is the maximum number of ensemble members, n is the number of assimilation days, dt is the time integration step, T d is the frequency of the time integration step required for the disturbance, ENS real is the number of set members actually set, T ENS For each ensemble member, the generation time in the pattern, ENS i The sequence number of the set member.

[0075] In one example, if Figure 5 As shown in the figure, when making seasonal forecasts, the initial field is obtained by assimilating the analysis field of one day at each start of the forecast. To ensure the stability of the calculation, the time integration step is selected as 1800s, so that the maximum number of ensemble members that can be generated is 48. Assuming that 24 ensemble members are generated, then T d=2, so the disturbance generation time from the 1st member to the 24th ensemble member (i.e. the generation time of the ensemble member in the climate system model) is 3600s, 7200s, 10800s to 86400s respectively.

[0076] After obtaining multiple different initial fields, numerical predictions are performed according to different initial fields to form an ensemble forecast with several ensemble members. Finally, the ensemble forecast is post-processed to obtain the seasonal forecast results of wind resource power.

[0077] In an embodiment of the present invention, a time-delayed perturbation method is used to generate different initial fields and then generate an ensemble forecast including multiple prediction results, which increases the diversity of the predictions. The uncertainty of the prediction results can be evaluated by means of ensemble forecasts, thereby improving the reliability and accuracy of the predictions.

[0078] Furthermore, if Figure 7 As shown, in step S1032, the ensemble forecast is post-processed to obtain the wind resource power seasonal forecast result, including:

[0079] Step b1, dividing the ensemble forecast into a preset number of spherical data to be processed based on a finite volume spherical cube grid dynamic framework;

[0080] Step b2, interpolating and merging the to-be-processed data of a preset number of spherical surfaces into global latitude and longitude grid data through an interpolation conversion program;

[0081] Step b3, extracting variables corresponding to wind resources from global latitude and longitude grid data, and averaging the variables corresponding to different set members to obtain output data;

[0082] Step b4, outputting the output data.

[0083] Specifically, the preset number is 6. The atmospheric component based on the finite volume spherical cubic grid dynamic framework divides the sphere into 6 spherical surfaces for calculation, and completes data exchange through the boundary. Finally, the calculation results of each spherical surface are output as separate data to be processed (including data for 6 hours and days). The process is shown in Figure 7:

[0084] (1) Through the interpolation conversion program, the unprocessed data of the six spheres are interpolated and merged into global longitude and latitude grid data;

[0085] (2) Extract the required variables, write the set members into one data, calculate the set average and output it;

[0086] (3) Calculate wind resource related variables, including 10m wind speed and direction, 100m wind speed and direction, and wind energy density as output data.

[0087] The embodiment of the present invention converts the ensemble forecast into global latitude and longitude grid data through interpolation conversion and data post-processing, making the prediction result more intuitive and easy to understand. At the same time, by averaging the variables of different ensemble members, random errors can be reduced and the stability of the prediction results can be improved.

[0088] Further, in step S1032, after performing data post-processing on the ensemble forecast to obtain the seasonal forecast result of wind resource power, the method further includes:

[0089] Step S1033, calculate the time correlation coefficient, normalized root mean square error and mean absolute error in the deterministic forecast detection index according to the seasonal forecast result of wind resource power, and the calculation formula is:

[0090]

[0091]

[0092] In the formula, ACC represents the time correlation coefficient, nRMSE represents the normalized root mean square error, MAE represents the mean absolute error, i represents time, N represents the total number of time samples, and F i represents the seasonal forecast result of wind resource power, O i Represents reference data, and They represent the seasonal prediction results of wind resource dynamics and the climate state corresponding to the reference data respectively.

[0093] Step S1034, using the Receiver Operating Characteristic (ROC) method to evaluate the probability forecast skill score, and obtain the hit rate and false alarm rate, wherein the hit rate and false alarm rate are calculated as follows:

[0094]

[0095] In the formula, H R m Indicates hit rate, FAR m represents the false alarm rate, j represents the number of bins, YO j and NO j They respectively represent the number of times the observations occurred and the number of times they did not occur in all cases of the distribution of ensemble members corresponding to the bin number j in the seasonal forecast results of wind resource power.

[0096] The probability forecast ROC contingency table is shown below, where M represents the number of set members, F represents the number of members for which the forecast event occurs, and MF represents the number of members for which the forecast event does not occur.

[0097] Number of warehouses Forecast ensemble membership distribution Observation Occurrences Observation not occurring times 1 F=0,MF=M <![CDATA[YO1]]> <![CDATA[NO1]]> 2 F=1,MF=M-1 <![CDATA[YO2]]> <![CDATA[NO2 <!-- 8 -->]]> 3 F=2,MF=M-2 <![CDATA[YO3]]> <![CDATA[NO3]]> … … … … M+1 F=M,MF=0 <![CDATA[YO M+1 ]]> <![CDATA[NO M+1 ]]>

[0098] Step S1035, performing deterministic forecast detection on the seasonal forecast results of wind resource power according to the time correlation coefficient, the normalized root mean square error and the mean absolute error, and performing probabilistic forecast detection on the seasonal forecast results of wind resource power according to the hit rate and the false alarm rate.

[0099] Specifically, the NCL script is used to calculate according to the above-mentioned test method to obtain the time correlation coefficient, normalized root mean square error, mean absolute error, hit rate and false alarm rate. The calculated results are compared with the observation or reanalysis data to complete the deterministic forecast detection and probabilistic forecast detection.

[0100] The deterministic forecast detection of wind resource power seasonal forecast results is carried out through the time correlation coefficient, normalized root mean square error and mean absolute error, and the probabilistic forecast detection of wind resource power seasonal forecast results is carried out according to the hit rate and false alarm rate. The forecast results can be quantitatively evaluated to help users understand the accuracy and reliability of the forecast and provide decision support for the development and utilization of wind resources.

[0101] Optionally, in step S1035, after performing a deterministic forecast detection on the seasonal forecast result of wind resource power according to the time correlation coefficient, the normalized root mean square error and the mean absolute error, and performing a probabilistic forecast detection on the seasonal forecast result of wind resource power according to the hit rate and the false alarm rate, the method further includes:

[0102] Step S1036, visually displaying the deterministic forecast detection results and the probabilistic forecast detection results.

[0103] like Figure 8 As shown in the figure, the visualization module is used to visualize the deterministic forecast detection results and the probabilistic forecast detection results. In addition, the seasonal forecast results of wind resource power can also be displayed through the visualization module at the same time. That is, the seasonal forecast results of wind resource power, the deterministic forecast detection results, and the probabilistic forecast detection results can be output as pictures for visualization at the same time. Through the visualization display, non-professional users can also intuitively understand the forecast results, which improves the acceptability and ease of use of the forecast results.

[0104] In one embodiment, the input data includes historical input data and current input data. Correspondingly, in step S103, numerical prediction is performed according to the initial field to obtain a seasonal prediction result of wind resource power, including:

[0105] Step S1031, performing a historical return test according to the initial field corresponding to the historical input data, and obtaining the seasonal prediction result of the wind resource power in the historical time period.

[0106] Step S1032, performing a real-time prediction test according to the initial field corresponding to the current input data, and obtaining a seasonal prediction result of wind resource power in a future time period.

[0107] Specifically, Figure 6 As shown in the figure, design a historical return experiment and a real-time prediction experiment. The steps of the historical return experiment include:

[0108] The data collection and preprocessing module completes the collection and preprocessing of historical observations and reanalysis data from 2000 to 2022, generates the initial field through data assimilation, and uses it as the initial value condition to drive the climate system model.

[0109] In order to ensure that the model obtains stable integrals and adjust the initial field of the model to make it closer to the actual observed value, a 3-year (2000-2002) spin-up was carried out before the historical return test. The model was adjusted to achieve a balance between the internal dynamic and thermal processes, reduce the impact of initial values ​​and model deviations on the prediction, and improve the accuracy and reliability of the prediction.

[0110] The historical return experiment is reported from the second last day of each month from 2003 to 2022, with one day of assimilation, the assimilation time step is 6 hours, the integration is 190 days, and there are 24 ensemble members. Therefore, there are 12×24 groups of experiments each year, and a total of 5760 runs.

[0111] The real-time prediction experiment steps include: completing data collection and preprocessing from 2023 to the present, and making real-time predictions twice a month (5th and 15th) starting from January 2023, predicting the next 180 days, consisting of 35-48 ensemble members, with a maximum of 12×48 sets of experiments per year (i.e. 576 runs).

[0112] Conducting historical return tests and real-time prediction tests separately can evaluate the prediction performance of the model in different time periods, which helps to further optimize model parameters and improve prediction accuracy.

[0113] A method for seasonal prediction of wind resource power based on a numerical model according to an embodiment of the present invention has the following advantages:

[0114] The climate system model can be used to predict wind resources on a global scale. The climate system model includes an atmospheric component built on a finite volume spherical cubic grid dynamic framework, which maintains the continuity and conservation of physical processes on the sphere, has high computational efficiency, and is scalable. In the future, the grid can be divided according to the simulation / prediction area and needs, and the grid can be encrypted in areas that require higher resolution, while reducing the errors caused by boundary conditions, thereby improving simulation accuracy;

[0115] The time-delayed perturbation method is used to generate different initial fields, and ensemble forecasts are performed based on these initial fields. The ensemble forecast method can be used to evaluate the uncertainty of the prediction results and improve the reliability and accuracy of the prediction.

[0116] The embodiment of the present invention also provides a wind resource power prediction system based on a numerical model, such as Fig. 9 As shown, including:

[0117] The data preprocessing module 901 is used to obtain preset data and preprocess the preset data to obtain input data;

[0118] An initial field generation module 902 is used to assimilate the input data into the climate system model to generate an initial field, wherein the climate system model includes an atmospheric component, an ocean component, a land component and a sea ice component that realize data interaction through a coupler, and the atmospheric component adopts a finite volume based spherical cubic grid dynamic framework;

[0119] The wind resource prediction module 903 is used to perform numerical prediction based on the initial field to obtain the wind resource dynamic seasonal prediction result.

[0120] The wind resource power prediction system based on the numerical model of the embodiment of the present invention obtains input data by acquiring preset data and preprocessing the preset data, assimilates the input data into the climate system model, and generates an initial field, wherein the climate system model includes an atmospheric component, an ocean component, a land component, and a sea ice component that realize data interaction through a coupler, and the atmospheric component adopts a spherical cubic grid dynamic framework based on a finite volume, performs numerical prediction according to the initial field, and obtains a seasonal prediction result of the wind resource power, and uses the atmospheric component, the ocean component, the land component, and the sea ice component to construct a climate system model, and uses a spherical cubic grid based on a finite volume method as a dynamic framework in the atmospheric component. On the one hand, the spherical cubic grid can better realize the accurate characterization and simulation of physical processes on the spherical surface, and at the same time, the grid on the spherical surface is regular and uniform, which reduces the complexity of calculation on the spherical surface, making the model calculation simpler and more efficient, and on the other hand, the finite volume method can improve the stability of the numerical model calculation, improve the accuracy and reliability of the model simulation, and at the same time, it has a high computational efficiency when solving partial differential equations, and has great advantages in solving problems under complex boundary conditions. This enables the prediction system based on the finite volume spherical cubic grid as the power framework to achieve efficient computing in a parallel computing environment without losing computing accuracy, thereby improving the accuracy of wind resource power forecasts, and ultimately providing reliable forecast product support and decision-making basis for wind farm operation and maintenance plans, electricity market transactions, and wind farm power generation benefit estimation.

[0121] Furthermore, the initial field generation module 901 includes:

[0122] The time-lagged perturbation module is used to add perturbations at different times during the assimilation of input data into the climate system model based on the time-lagged perturbation method to generate several different initial fields;

[0123] Correspondingly, the wind resource prediction module 903 includes:

[0124] An ensemble forecast module is used to perform numerical forecasts according to different initial fields to form an ensemble forecast with several ensemble members;

[0125] The data post-processing module is used to perform data post-processing on the ensemble forecast to obtain the seasonal forecast results of wind resource power.

[0126] Furthermore, the time-delay disturbance module includes:

[0127] A quantity determination module is used to determine the maximum number of set members according to the assimilation time period and the time integration step;

[0128] An interval frequency module, used for selecting the number of set members based on the maximum number of set members, and determining the interval frequency of the time integration step required for the disturbance according to the number of set members and the maximum number of set members;

[0129] The disturbance adding module is used to add disturbances at the time corresponding to each ensemble member in the process of assimilating the input data into the climate system model, and generate several different initial fields corresponding to the ensemble members, where the time corresponding to the ensemble member is the product of the interval frequency, the time integration step and the serial number of the ensemble member.

[0130] Furthermore, the data post-processing module includes:

[0131] A spherical partitioning module is used to divide the ensemble forecast into a preset number of spherical data to be processed based on a finite volume spherical cubic grid dynamic framework;

[0132] A difference conversion module is used to interpolate and merge the to-be-processed data of a preset number of spherical surfaces into global latitude and longitude grid data through an interpolation conversion program;

[0133] The ensemble average module is used to extract the variables corresponding to the wind resources from the global latitude and longitude grid data, and average the variables corresponding to different ensemble members to obtain the output data;

[0134] The data output module is used to output the output data.

[0135] Furthermore, the wind resource power prediction system based on the numerical model also includes:

[0136] The first calculation module is used to calculate the time correlation coefficient, normalized root mean square error and mean absolute error in the deterministic forecast detection index according to the seasonal forecast results of wind resource power. The calculation formula is:

[0137]

[0138] In the formula, ACC represents the time correlation coefficient, nRMSE represents the normalized root mean square error, MAE represents the mean absolute error, i represents time, N represents the total number of time samples, and F i represents the seasonal forecast result of wind resource power, O i Represents reference data, and They represent the climate states corresponding to the seasonal prediction results of wind resource dynamics and the reference data respectively;

[0139] The second calculation module is used to evaluate the probability forecast skill score by the relative operation characteristic method to obtain the hit rate and false alarm rate, wherein the hit rate and false alarm rate calculation formula are:

[0140]

[0141] In the formula, H R m Indicates hit rate, FAR m represents the false alarm rate, j represents the number of bins, M represents the number of set members, YO j and NO j They represent the number of times the corresponding observations occurred and did not occur in all cases of the distribution of the set members corresponding to the bin number j in the seasonal forecast results of wind resource power;

[0142] The forecast detection module is used to perform deterministic forecast detection on the seasonal forecast results of wind resource power according to the time correlation coefficient, the normalized root mean square error and the mean absolute error, and to perform probabilistic forecast detection on the seasonal forecast results of wind resource power according to the hit rate and the false alarm rate.

[0143] Furthermore, the wind resource power prediction system based on the numerical model also includes:

[0144] The visualization module is used to visualize the results of deterministic forecast detection and probabilistic forecast detection.

[0145] Furthermore, the initial field generation module includes:

[0146] The relaxation approximation module is used to assimilate the input data as initial conditions into the climate system model using a time-varying relaxation approximation method based on the incremental analysis update process to generate the initial field.

[0147] Furthermore, the input data includes historical input data and current input data. Correspondingly, the wind resource prediction module includes:

[0148] The historical return prediction module is used to perform historical return tests based on the initial field corresponding to the historical input data to obtain the seasonal prediction results of wind resource power in the historical time period;

[0149] The real-time prediction module is used to perform real-time prediction tests based on the initial field corresponding to the current input data to obtain the seasonal prediction results of wind resource power in the future time period.

[0150] The embodiment of the present invention also provides a schematic diagram of the structure of a computer device, such as Fig.10 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Fig.10 A processor 10 is taken as an example.

[0151] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0152] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.

[0153] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0154] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0155] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Fig.10 The example of connecting through bus is taken in the following.

[0156] The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0157] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0158] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0159] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for seasonal prediction of wind resource power based on numerical model, characterized in that: include: Acquire preset data and preprocess the preset data to obtain input data; Assimilating the input data into a climate system model to generate an initial field, wherein the climate system model includes an atmospheric component, an ocean component, a land surface component and a sea ice component that implement data interaction through a coupler, and the atmospheric component adopts a finite volume-based spherical cubic grid dynamic framework; A numerical prediction is performed based on the initial field to obtain a seasonal prediction result of wind resource power.

2. The method for seasonal prediction of wind resource power based on numerical model according to claim 1, characterized in that: Assimilating the input data into a climate system model to generate an initial field includes: Based on the time-lagged perturbation method, perturbations are added at different times during the assimilation of the input data into the climate system model to generate a number of different initial fields; Correspondingly, the numerical prediction is performed according to the initial field to obtain the seasonal prediction result of wind resource power, including: Performing numerical predictions according to different initial fields respectively to form an ensemble forecast having a plurality of ensemble members; The ensemble forecast is subjected to data post-processing to obtain a seasonal forecast result of wind resource power.

3. The method for seasonal prediction of wind resource power based on numerical model according to claim 2 is characterized in that: The time-lag perturbation method adds perturbations at different times during the assimilation of the input data into the climate system model to generate several different initial fields, including: Determine the maximum number of ensemble members based on the assimilation time period and time integration step; Selecting the number of set members based on the maximum number of set members, and determining the interval frequency of the time integration step required for the disturbance according to the number of set members and the maximum number of set members; In the process of assimilating the input data into the climate system model, disturbances are added at the time corresponding to each ensemble member to generate a number of different initial fields corresponding to the ensemble members, wherein the time corresponding to the ensemble member is the product of the interval frequency, the time integration step and the serial number of the ensemble member.

4. The method for seasonal prediction of wind resource power based on numerical model according to claim 3 is characterized in that: The post-processing of the ensemble forecast to obtain a seasonal forecast result of wind resource power includes: Based on a finite volume spherical cube grid dynamic framework, the ensemble forecast output result is divided into a preset number of spherical surfaces of data to be processed; Through the interpolation conversion program, the to-be-processed data of a preset number of spheres are interpolated and merged into global latitude and longitude grid data; Extracting variables corresponding to wind resources from the global latitude and longitude grid data, and outputting the data; The output data is outputted.

5. The method for seasonal prediction of wind resource power based on numerical model according to claim 2, characterized in that: After the ensemble forecast is post-processed to obtain the seasonal forecast result of wind resource power, the method further includes: According to the seasonal prediction results of wind resource power, the time correlation coefficient, the normalized root mean square error and the mean absolute error in the deterministic forecast detection index are calculated, and the calculation formula is: In the formula, ACC represents the time correlation coefficient, nRMSE represents the normalized root mean square error, MAE represents the mean absolute error, i represents time, N represents the total number of time samples, and F i represents the seasonal forecast result of wind resource power, O i Represents reference data, and They represent the climate states corresponding to the seasonal prediction results of wind resource dynamics and the reference data respectively; The relative operating characteristic method is used to evaluate the probability forecast skill score to obtain the hit rate and false alarm rate, where the hit rate and false alarm rate calculation formula are: In the formula, HR m Indicates hit rate, FAR m represents the false alarm rate, j represents the number of bins, M represents the number of set members, YO j and NO j They represent the number of times the corresponding observations occurred and did not occur in all cases of the distribution of the set members corresponding to the bin number j in the seasonal forecast results of wind resource power; The wind resource power seasonal prediction results are deterministically predicted and tested based on the time correlation coefficient, the normalized root mean square error and the mean absolute error, and the wind resource power seasonal prediction results are probabilistically predicted and tested based on the hit rate and the false alarm rate.

6. The method for seasonal prediction of wind resource power based on numerical model according to claim 5 is characterized in that: After performing deterministic forecast detection on the seasonal forecast result of the wind resource power according to the time correlation coefficient, the normalized root mean square error and the mean absolute error, and performing probabilistic forecast detection on the seasonal forecast result of the wind resource power according to the hit rate and the false alarm rate, the method further includes: Visualize the results of deterministic forecast detection and probabilistic forecast detection.

7. The method for seasonal prediction of wind resource power based on numerical model according to claim 1, characterized in that: Assimilating the input data into a climate system model to generate an initial field includes: Based on the incremental analysis update process, the input data are assimilated into the climate system model as initial conditions using a time-varying relaxation approximation method to generate an initial field.

8. The method for seasonal prediction of wind resource power based on numerical model according to claim 1, characterized in that: The input data includes historical input data and current input data. Correspondingly, the numerical prediction is performed according to the initial field to obtain the seasonal prediction result of wind resource power, including: Perform a historical return test according to the initial field corresponding to the historical input data to obtain a seasonal forecast result of wind resource power in a historical time period; A real-time prediction test is performed based on the initial field corresponding to the current input data to obtain a seasonal prediction result of wind resource power in a future time period.

9. A wind resource power prediction system based on numerical model, characterized in that: include: A data preprocessing module, used to obtain preset data and preprocess the preset data to obtain input data; An initial field generation module is used to assimilate the input data into a climate system model to generate an initial field, wherein the climate system model includes an atmospheric component, an ocean component, a land component and a sea ice component that realize data interaction through a coupler, and the atmospheric component adopts a finite volume-based spherical cubic grid dynamic framework; The wind resource prediction module is used to perform numerical prediction based on the initial field to obtain a wind resource dynamic seasonal prediction result.

10. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method for seasonal prediction of wind resource power based on a numerical model as described in any one of claims 1 to 9 by executing the computer instructions.

Images