Wind power short-term power prediction method and device based on high-level meteorological data and medium

Abstract

The application provides a wind power short-term power prediction method and device based on high-level meteorological data and a medium, comprising: obtaining high-level meteorological forecast data of a wind farm to be predicted; using a pre-constructed wind farm power short-term prediction model to perform prediction processing on the high-level meteorological forecast data of the to-be-predicted time, to obtain a wind power short-term power prediction value of the to-be-predicted time; wherein the pre-constructed wind farm power short-term prediction model is constructed by a high-level meteorological mode forecast factor combination and an optimal weight combination determined by analyzing historical high-level meteorological forecast data of the wind farm. The application uses high-level meteorological data less disturbed to combine the wind farm power short-term prediction model to obtain a higher-precision wind power short-term power prediction value.

Description

Technical Field

[0001] The present invention relates to the field of new energy power generation technology, and in particular to a method, device and medium for predicting short-term wind power based on high-level meteorological data. Background Art

[0002] The wind farm's environment and local climate change are the primary factors influencing its output power. The random variability of wind speed and direction leads to fluctuating wind power output. Integrating large-scale wind power into the grid without knowing the output can negatively impact the grid's safe and stable operation. Power forecasting is an effective solution to this problem.

[0003] To improve forecast accuracy, more and more modeling methods have emerged, mainly focusing on two aspects: improving NWP forecast accuracy and reducing wind-to-power conversion errors.

[0004] Methods for improving NWP forecast accuracy include enhancing numerical forecast performance and numerical forecast post-processing. In numerical forecasting technology, in addition to the introduction of data assimilation and artificial intelligence analysis, there has also been rapid progress in refined forecasting, such as the development of regional grid prediction technology based on network architecture topology, the introduction of the WRF model, and the use of satellite data inversion technology to obtain measured meteorological data to compensate for the lack of meteorological observation data.

[0005] In numerical forecast post-processing, historical measured and forecasted surface meteorological data are often used to revise future numerical forecasts. For wind speed-power conversion models, a scatter plot of wind speed and power is typically fitted to produce a fitted curve, which serves as the wind speed-power conversion curve.

[0006] Existing technology uses historical measured and predicted surface meteorological data to revise future data forecasts, and obtains a fitting curve as a wind speed-power conversion curve by fitting a scatter plot of wind speed and power. However, the power prediction using this method is too rough and has low forecast accuracy. Summary of the Invention

[0007] To address the problem that the existing technology uses historical measured and forecasted surface meteorological data to revise future numerical forecasts and obtains a fitting curve as a wind speed-power conversion curve by fitting a scatter plot of wind speed and power, the power prediction by this method is too rough and the forecast accuracy is low. Therefore, the present invention provides a method for short-term wind power prediction based on high-level meteorological data, which includes:

[0008] Obtain high-level weather forecast data for the wind farm at the time of prediction;

[0009] Using a pre-built short-term wind farm power prediction model to perform prediction processing on the high-level meteorological forecast data at the time to be predicted, to obtain a short-term wind power prediction value at the time to be predicted;

[0010] The pre-built short-term wind farm power prediction model is constructed by analyzing historical high-level meteorological forecast data of the wind farm and determining the optimal high-level meteorological model forecast factor combination and the optimal weight combination.

[0011] Preferably, the construction of the wind farm power short-term prediction model includes:

[0012] Obtain historical high-level weather forecast data for wind farms;

[0013] Conduct feature analysis on historical high-level meteorological forecast data to determine high-level meteorological model forecast factors;

[0014] Determining target historical meteorological forecast data from the historical meteorological forecast data of the high-level meteorological model forecast factor using a similarity algorithm;

[0015] Analyzing the target historical weather forecast data using an ergodic method to determine an optimal high-level weather model forecast factor combination and an optimal weight combination;

[0016] A short-term wind power prediction model is constructed based on the optimal high-level meteorological model forecast factor combination and the optimal weight combination.

[0017] Preferably, the method of determining target historical meteorological forecast data from the historical meteorological forecast data of the high-level meteorological model forecast factor using a similarity algorithm includes:

[0018] Select the historical meteorological forecast data of each high-level meteorological model forecast factor at a certain moment as the current data;

[0019] According to the high-level meteorological model forecast factor corresponding to the current data, the historical meteorological forecast data closest to the current data is selected from the historical meteorological forecast data as the target historical meteorological forecast data.

[0020] Preferably, the method of analyzing the target historical weather forecast data using the ergodic method to determine the optimal high-level weather model forecast factor combination and the optimal weight combination includes:

[0021] Using the actual short-term wind power value corresponding to the target historical weather forecast data as the short-term wind power forecast value corresponding to the current data;

[0022] Constructing a sample set from current data of each high-level meteorological model forecast factor, a predicted value of short-term wind power corresponding to the current data, and an actual value of short-term wind power corresponding to the current data;

[0023] Dividing the sample set into a training set and a test set;

[0024] By combining the high-level meteorological model forecast factors in the training set using an ergodic method and assigning weights to the high-level meteorological model forecast factors in each combination, an optimal high-level meteorological model forecast factor combination and an optimal weight combination are selected;

[0025] The combined scheme prediction value obtained by the optimal high-level meteorological model forecast factor combination and the optimal weight combination is verified through the test set, and a wind power short-term prediction model is constructed by the verified optimal high-level meteorological model forecast factor combination and the optimal weight combination.

[0026] Preferably, the selecting the optimal high-level meteorological model prediction factor combination and the optimal weight combination by combining the high-level meteorological model prediction factors in the training set using an ergodic method and assigning weights to the high-level meteorological model prediction factors in each combination comprises:

[0027] The optimal high-level meteorological model forecast factors in the training set are used to construct a combination scheme using the ergodic method, and the weights of each actual observation value in each combination scheme are assigned using the ergodic method.

[0028] The measured value of each combination scheme is obtained by multiplying the actual observed value in each combination scheme by its respective weight;

[0029] The combination scheme and weight combination corresponding to the combination scheme prediction value closest to the actual value of the short-term wind power in the training set are respectively used as the combination of optimal high-level meteorological model forecast factors and the optimal weight combination.

[0030] Preferably, the verification of the predicted value of the combination scheme obtained by the optimal high-level meteorological model forecast factor combination and the optimal weight combination through the test set includes:

[0031] Selecting actual observation values ​​corresponding to the optimal high-level meteorological model forecast factor combination from the test set;

[0032] Assign weights to the selected actual observations according to the best weight combination, and obtain the combined scheme prediction value of each group of schemes by multiplying the selected actual observations by the assigned weights.

[0033] Subtracting the actual short-term wind power value in the test set from the predicted value of the combination scheme to obtain an error value;

[0034] Determine whether the error value is less than the error threshold; if so, verify the optimal weight combination and the optimal high-level meteorological model forecast factor combination, and construct a short-term wind power prediction model based on the optimal high-level meteorological model forecast factor combination and the optimal weight combination; otherwise, fail the verification.

[0035] Preferably, the method of using a pre-built short-term wind farm power prediction model to perform prediction processing on the high-level meteorological forecast data at the time to be predicted to obtain the short-term wind power prediction value at the time to be predicted includes:

[0036] Inputting the high-level meteorological forecast data into a short-term wind farm power forecast model, and extracting historical meteorological forecast data at similar times by the short-term wind farm power forecast model according to each high-level meteorological model forecast factor in the optimal high-level meteorological model forecast factor combination;

[0037] Calculating the distance between the historical weather forecast data at the similar time and the forecast data of each high-level weather model forecast factor at the time to be predicted, and selecting the actual value of wind power short-term corresponding to the time of the historical weather forecast data having the smallest distance to the forecast data of each high-level weather model forecast factor at the time to be predicted as the value corresponding to each high-level weather model forecast factor;

[0038] Multiplying the value corresponding to each high-level meteorological model forecast factor by the weight of each high-level meteorological model forecast factor in the optimal weight combination to obtain a short-term wind power forecast value of each high-level meteorological model forecast factor;

[0039] The short-term wind power forecast values ​​of the respective high-level meteorological model forecast factors in the optimal high-level meteorological model forecast factor combination are summed up as the short-term wind power forecast value of the wind farm at the time to be forecasted.

[0040] In another aspect, the present invention further provides a wind power short-term power prediction device based on high-level meteorological data, comprising an acquisition module for acquiring high-level meteorological forecast data of a wind farm at a time to be predicted;

[0041] A prediction module is used to use a pre-built short-term wind farm power prediction model to perform prediction processing on the high-level meteorological forecast data at the time to be predicted, so as to obtain a short-term wind power prediction value at the time to be predicted;

[0042] The pre-built short-term wind farm power prediction model is constructed by analyzing historical high-level meteorological forecast data of the wind farm and determining a combination of high-level meteorological model forecast factors and an optimal weight combination.

[0043] Preferably, it also includes a model building module for building a short-term wind farm power prediction model;

[0044] The model building module includes:

[0045] The data acquisition submodule is used to obtain historical high-level weather forecast data for wind farms;

[0046] The analysis submodule is used to perform feature analysis on historical high-level meteorological forecast data and select high-level meteorological model forecast factors;

[0047] a combination determination submodule, configured to determine target historical meteorological forecast data from the historical meteorological forecast data of the high-level meteorological model forecast factors using a similarity algorithm, and to analyze the target historical meteorological forecast data using an ergodic method to determine an optimal high-level meteorological model forecast factor combination and an optimal weight combination;

[0048] The summary submodule is used to construct a short-term wind power prediction model based on the optimal high-level meteorological model forecast factor combination and the optimal weight combination.

[0049] Preferably, the combination determination submodule is specifically used for:

[0050] Select the historical meteorological forecast data of each high-level meteorological model forecast factor at a certain moment as the current data;

[0051] selecting, from the historical weather forecast data, the historical weather forecast data closest to the current data as the target historical weather forecast data, according to the high-level weather model forecast factor corresponding to the current data;

[0052] Using the actual value of the short-term wind power corresponding to the target historical weather forecast data as the predicted value of the short-term wind power corresponding to the current data;

[0053] Constructing a sample set from current data of each high-level meteorological model forecast factor, a predicted value of short-term wind power corresponding to the current data, and an actual value of short-term wind power corresponding to the current data;

[0054] Dividing the sample set into a training set and a test set;

[0055] By combining the high-level meteorological model forecast factors in the training set using an ergodic method and assigning weights to the high-level meteorological model forecast factors in each combination, an optimal high-level meteorological model forecast factor combination and an optimal weight combination are selected;

[0056] The combined scheme prediction value obtained by the optimal high-level meteorological model forecast factor combination and the optimal weight combination is verified through the test set, and a wind power short-term prediction model is constructed by the verified optimal high-level meteorological model forecast factor combination and the optimal weight combination.

[0057] In another aspect, the present invention further provides a computer device comprising: one or more processors;

[0058] The processor is configured to store one or more programs;

[0059] When the one or more programs are executed by the one or more processors, the above-mentioned method for short-term wind power prediction based on high-level meteorological data is implemented.

[0060] In another aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed, the method for short-term wind power prediction based on high-level meteorological data as described above is implemented.

[0061] Compared with the prior art, the present invention has the following beneficial effects:

[0062] 1. The present invention provides a method for forecasting short-term wind power based on high-level meteorological data, comprising obtaining high-level meteorological forecast data for a wind farm at a time to be forecasted; and using a pre-established short-term wind power forecast model to forecast the high-level meteorological forecast data for the time to be forecasted, thereby obtaining a short-term wind power forecast value for the time to be forecasted. The pre-established short-term wind power forecast model is constructed by analyzing historical high-level meteorological forecast data for the wind farm and determining a combination of high-level meteorological model forecast factors and an optimal weight combination. The present invention utilizes less-interfered high-level meteorological data in combination with the short-term wind power forecast model to obtain a more accurate short-term wind power forecast value.

[0063] 2. The present invention adopts a high-level meteorological element similarity algorithm and a traversal method to determine the high-level meteorological model forecast factor combination and the optimal weight combination, so that the selected high-level meteorological model forecast factor is more reasonable. Compared with the existing technology that only considers a single image factor, the short-term wind power forecast value calculated by this method is more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 This is a flow chart of the method for short-term wind power prediction based on high-level meteorological data of the present invention;

[0065] Figure 2 This is a flowchart of offline training and online prediction of the present invention;

[0066] Figure 3 Schematic diagram of the prediction results of the present invention. DETAILED DESCRIPTION

[0067] At present, the performance of numerical forecasting has entered a bottleneck period. Model post-processing mostly refers to the correction of the numerical forecast wind speed of the forecast part from the surface numerical forecast data such as surface wind speed, wind direction, temperature, humidity, and pressure, combined with historical wind tower observation data, and then the wind-to-power conversion is performed through some methods to predict the power.

[0068] Considering that surface meteorological data is not only influenced by atmospheric motion but also by terrain, surface roughness, environmental pollution, vegetation, and population density, high-altitude meteorology is less affected by the surface environment and has more regular atmospheric motion. Therefore, the relationship between high-altitude meteorology and wind power generation is more clearly reflected in the data. Therefore, the present invention proposes a power prediction method based on high-altitude meteorological data. This method uses low-interference high-altitude meteorological data to convert wind power to power, resulting in more accurate power prediction.

[0069] Example 1

[0070] The present invention provides a method for predicting short-term wind power based on high-level meteorological data, such as Figure 1 Shown, including:

[0071] Step 1: Obtain high-level meteorological forecast data for the wind farm at the time to be predicted;

[0072] Step 2: using a pre-built short-term wind farm power prediction model to perform prediction processing on the high-level meteorological forecast data at the time to be predicted, to obtain a short-term wind power prediction value at the time to be predicted;

[0073] The pre-built short-term wind farm power prediction model is constructed by analyzing historical high-level meteorological forecast data of the wind farm and determining a combination of high-level meteorological model forecast factors and an optimal weight combination.

[0074] Before step 1, it also includes: building a short-term wind farm power prediction model, which is trained and established based on a high-level meteorological element similarity algorithm. Figure 2 A detailed introduction is given to the construction of a short-term wind farm power prediction model.

[0075] Obtain historical high-level weather forecast data for wind farms;

[0076] Conduct feature analysis on historical high-level meteorological forecast data and select high-level meteorological model forecast factors;

[0077] Determining target historical meteorological forecast data from the historical meteorological forecast data of the high-level meteorological model forecast factor using a similarity algorithm;

[0078] Analyzing the target historical weather forecast data using an ergodic method to determine an optimal high-level weather model forecast factor combination and an optimal weight combination;

[0079] A short-term wind power prediction model is constructed based on the optimal high-level meteorological model forecast factor combination and the optimal weight combination.

[0080] Furthermore, the use of a similarity algorithm to determine target historical weather forecast data from the historical weather forecast data of the high-level weather model forecast factor includes:

[0081] Select the historical meteorological forecast data of each high-level meteorological model forecast factor at a certain moment as the current data;

[0082] According to the high-level meteorological model forecast factor corresponding to the current data, the historical meteorological forecast data closest to the current data is selected from the historical meteorological forecast data as the target historical meteorological forecast data.

[0083] Furthermore, the ergodic method is used to analyze the target historical weather forecast data to determine the optimal high-level weather model forecast factor combination and the optimal weight combination, including:

[0084] Using the actual value of the short-term wind power corresponding to the target historical weather forecast data as the predicted value of the short-term wind power corresponding to the current data;

[0085] Constructing a sample set from current data of each high-level meteorological model forecast factor, a predicted value of short-term wind power corresponding to the current data, and an actual value of short-term wind power corresponding to the current data;

[0086] Dividing the sample set into a training set and a test set;

[0087] By combining the high-level meteorological model forecast factors in the training set using an ergodic method and assigning weights to the high-level meteorological model forecast factors in each combination, an optimal high-level meteorological model forecast factor combination and an optimal weight combination are selected;

[0088] The combined scheme prediction value obtained by the optimal high-level meteorological model forecast factor combination and the optimal weight combination is verified through the test set, and a wind power short-term prediction model is constructed by the verified optimal high-level meteorological model forecast factor combination and the optimal weight combination.

[0089] The selection of similar historical moments here includes: calculating the similarity between the current moment and a certain historical moment, including multiple consecutive moments before and after the current moment, and also including each selected high-level meteorological element, which is in the form of a matrix, the inverse of the weighted Euclidean distance between the current moment and the historical moment.

[0090] Here we take temperature as an example to introduce the forecast factor of the high-level meteorological model. From the historical temperature forecast data, a historical meteorological forecast data similar to the current temperature forecast value is selected, and the actual value of the wind power short-term power corresponding to the similar historical meteorological forecast data is used as the predicted value of the wind power short-term power corresponding to the current temperature.

[0091] Select a historical temperature forecast data that is similar to the current temperature forecast value from the historical temperature forecast data, including:

[0092] Select the temperature forecast value for multiple consecutive moments before the current moment and the historical temperature forecast values ​​corresponding to the same moment on different days;

[0093] The difference between the selected temperature historical forecast value and the current temperature forecast value is calculated, and the short-term actual wind power value corresponding to the temperature historical forecast value with the smallest difference is used as the wind power forecast value of the current temperature forecast value.

[0094] An example of constructing a sample set:

[0095] The distance between current high-level weather forecast data and historical high-level weather forecast data at the same and similar times on different days is calculated. The n similar historical times with the smallest distance are found. The actual observations corresponding to these n historical times are used as the member set, and the deterministic forecast result is obtained by weighted averaging. The high-level weather forecast data here specifically refers to the historical data of high-level weather model forecast factors. The deterministic forecast result is the sum of the actual observations in the member set multiplied by the weight of each actual observation, which is the short-term wind power forecast value. Distance is calculated using the conventional Euclidean distance.

[0096] Furthermore, the method of combining the high-level meteorological model prediction factors in the training set using the ergodic method and assigning weights to the high-level meteorological model prediction factors in each combination to select the optimal high-level meteorological model prediction factor combination and the optimal weight combination includes:

[0097] The optimal high-level meteorological model forecast factors in the training set are used to construct a combination scheme using the ergodic method, and the weights of each actual observation value in each combination scheme are assigned using the ergodic method.

[0098] The predicted value of each combination scheme is obtained by multiplying the actual observed value in each combination scheme by its respective weight;

[0099] The combination scheme corresponding to the combination scheme prediction value closest to the wind power short-term power prediction value and the wind power short-term power actual value in the training set is used as the combination of the optimal high-level meteorological model forecast factors and the best weight combination.

[0100] The weights are calculated using the traversal method to find the optimal weights. For example, if four meteorological elements are selected and the total coefficient is 10, then the total weight matching method is A(10, 4). The traversal method is used to select the set of weights that will give the best training effect.

[0101] Furthermore, historical high-level weather forecast data for the wind farm is obtained, including:

[0102] The historical data includes: high-level meteorological data and wind farm measured active power data, and the data time resolution is not less than 15 minutes.

[0103] High-level meteorological data include: TC925 / TC850 / TC700 / TC500 (temperature of 925 / 850 / 700 / 500 hPa), QS975 / QS850 / QS700 / QS500 (relative humidity of 975 / 850 / 700 / 500 hPa), U975 / V975 (horizontal and vertical wind speeds of 975 hPa), U925 / V925 (horizontal and vertical wind speeds of 925 hPa), U850 / V850 (horizontal and vertical wind speeds of 850 hPa), U700 / V700 (horizontal and vertical wind speeds of 700 hPa), U500 / V500 (horizontal and vertical wind speeds of 500 hPa), etc.

[0104] The high-level meteorological data are generated based on the WRF model, and the measured active power data are obtained from the wind farm monitoring system.

[0105] The acquisition of high-level meteorological forecast data for the wind farm at the time of prediction in step 1 specifically includes:

[0106] The predicted meteorological data is high-level meteorological data, which is generated based on the WRF model and has a data time resolution of no less than 15 minutes.

[0107] High-level meteorological data include: TC925 / TC850 / TC700 / TC500 (temperature of 925 / 850 / 700 / 500 hPa), QS975 / QS850 / QS700 / QS500 (relative humidity of 975 / 850 / 700 / 500 hPa), U975 / V975 (horizontal and vertical wind speeds of 975 hPa), U925 / V925 (horizontal and vertical wind speeds of 925 hPa), U850 / V850 (horizontal and vertical wind speeds of 850 hPa), U700 / V700 (horizontal and vertical wind speeds of 700 hPa), U500 / V500 (horizontal and vertical wind speeds of 500 hPa), etc.

[0108] Step 2 uses a pre-built short-term wind farm power prediction model to perform prediction processing on the high-level meteorological forecast data at the time to be predicted to obtain a short-term wind power prediction value at the time to be predicted, specifically including:

[0109] Inputting the high-level meteorological forecast data into a short-term wind farm power forecast model, and extracting historical meteorological forecast data at similar times by the short-term wind farm power forecast model according to each high-level meteorological model forecast factor in the optimal high-level meteorological model forecast factor combination;

[0110] Calculate the distance between the historical weather forecast data at the similar time and the forecast data of each high-level weather model forecast factor at the time to be predicted, and select the actual value of wind power short-term corresponding to the time of the historical weather forecast data with the smallest distance to the forecast data of each high-level weather model forecast factor at the time to be predicted as the value corresponding to each high-level weather model forecast factor;

[0111] Multiplying the value corresponding to each high-level meteorological model forecast factor by the weight of each high-level meteorological model forecast factor in the optimal weight combination to obtain a short-term wind power forecast value of each high-level meteorological model forecast factor;

[0112] The short-term wind power forecast values ​​of the respective high-level meteorological model forecast factors in the optimal high-level meteorological model forecast factor combination are summed up as the short-term wind power forecast value of the wind farm at the time to be forecasted.

[0113] In summary, based on the predicted meteorological data and the previously trained optimal meteorological factor combination and optimal weighted wind power short-term prediction model, the short-term power of the wind farm is predicted.

[0114] Example 2

[0115] The present invention takes a certain wind farm as an example. The wind farm is located in a hilly area with uneven terrain. Sudden strong winds often occur, which makes prediction difficult and the prediction error is large. The accuracy of the predicted power needs to be improved urgently. The data time interval used in the present invention is: from 0:00 on May 3, 2020 to 0:00 on January 17, 2021, a total of 259 days, with a time resolution of 15 minutes, 96 points per day, and a total of 24,864 data. The first 200 days, a total of 19,200 data, were selected as the training set, the following 28 days were used as the test set, a total of 2,688 data, and the remaining 31 days, a total of 2,976 data, were used as the prediction set.

[0116] Since there are too many high-level meteorological elements and surface meteorological elements, and substituting too high-dimensional elements into the model will introduce noise, we first consider using recursive feature elimination combined with random forest for dimensionality reduction. The working principle of recursive feature elimination (RFE) is to build a model, recursively delete features, and build a model based on the remaining features. The model accuracy is used to determine which features or feature combinations contribute more to the model. The final number of selected features and which features are selected can be output.

[0117] After the factors are selected, the high-level meteorological element similarity method uses the simultaneous current forecast and historical forecast to calculate the "distance", find the six historical similar moments with the smallest "distance", and use the corresponding observation values ​​of the forecast quantities required at the six historical moments as the member set, and obtain the deterministic forecast result after weighted averaging.

[0118] The distance is calculated using conventional Euclidean distance, and the weight optimization is performed using the ergodic method to calculate the optimal weight (the ergodic method, if 0.1 is the minimum variable, each prediction factor can take 11 possible weights, including 0.0, 0.1, ..., 1.0). The final prediction results are as follows: Figure 3 As shown in the figure, it can be seen that compared with the surface numerical prediction data, the high-level numerical prediction data can obtain higher-precision predicted power.

[0119] Example 3

[0120] The present invention based on the same inventive concept also provides a wind power short-term power forecasting system based on high-level meteorological data, comprising:

[0121] An acquisition module is used to obtain high-level meteorological forecast data of the wind farm at the time to be predicted;

[0122] A prediction module, configured to bring the high-level meteorological forecast data at the time to be predicted into a pre-built short-term wind farm power prediction model to obtain a short-term wind power prediction value at the time to be predicted;

[0123] The pre-built short-term wind farm power prediction model is constructed by analyzing historical high-level meteorological forecast data of the wind farm and determining a combination of high-level meteorological model forecast factors and an optimal weight combination.

[0124] Preferably, it also includes a model building module for building a short-term wind farm power prediction model;

[0125] The model building module includes:

[0126] The data acquisition submodule is used to obtain historical high-level weather forecast data for wind farms;

[0127] The analysis submodule is used to perform feature analysis on historical high-level meteorological forecast data using recursive feature elimination combined with random forests to select high-level meteorological model forecast factors;

[0128] A combination determination submodule, configured to determine an optimal high-level meteorological model forecast factor combination and an optimal weight combination based on historical meteorological forecast data of the high-level meteorological model forecast factors in combination with a similarity algorithm and an ergodic method;

[0129] The summary submodule is used to construct a short-term wind power prediction model based on the optimal high-level meteorological model forecast factor combination and the optimal weight combination.

[0130] Preferably, the combination determination submodule is specifically used for:

[0131] Select the historical meteorological forecast data of each high-level meteorological model forecast factor at a certain moment as the current data;

[0132] Among the historical weather forecast data of similar times on different days for each high-level weather model forecast factor, find the historical weather forecast data with the smallest distance to the current data of the high-level weather model forecast factor;

[0133] The actual value of the short-term wind power corresponding to the historical weather forecast data with the smallest distance is used as the short-term wind power forecast value corresponding to the current data of the high-level weather model forecast factor;

[0134] Constructing a sample set from current data of each high-level meteorological model forecast factor, a predicted value of short-term wind power corresponding to the current data, and an actual value of short-term wind power corresponding to the current data;

[0135] Dividing the sample set into a training set and a test set;

[0136] Selecting the optimal high-level meteorological model forecast factor combination and the optimal weight combination through the training set using an ergodic method;

[0137] The selected optimal high-level meteorological model forecast factor combination and the optimal weight combination are verified through the test set, and a wind power short-term prediction model is constructed based on the verified optimal high-level meteorological model forecast factor combination and the optimal weight combination.

[0138] Example 4

[0139] Based on the same inventive concept, the present invention also provides a computer device, which includes a processor and a memory, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or may be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding function, so as to implement the steps of the wind power short-term power prediction method based on high-level meteorological data in the above embodiment.

[0140] Example 5

[0141] Based on the same inventive concept, the present invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It can be understood that the computer-readable storage medium here can include both built-in storage media in the computer device and, of course, extended storage media supported by the computer device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space. These instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the steps of the wind power short-term power prediction method based on high-level meteorological data in the above embodiment.

[0142] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0143] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0144] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0145] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0146] The above are merely embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention are included in the scope of the claims of the present invention to be approved.

Claims

1. A method for short-term wind power forecasting based on high-level meteorological data, characterized in that: include Obtain high-level weather forecast data for the wind farm at the time of prediction; Using a pre-built short-term wind farm power prediction model to perform prediction processing on the high-level meteorological forecast data at the time to be predicted, to obtain a short-term wind power prediction value at the time to be predicted; The pre-built short-term wind farm power forecast model is constructed by analyzing the historical high-level meteorological forecast data of the wind farm and determining the optimal high-level meteorological model forecast factor combination and the optimal weight combination; The construction of the short-term wind farm power forecasting model includes: Obtain historical high-level weather forecast data for wind farms; Conduct feature analysis on historical high-level meteorological forecast data and select high-level meteorological model forecast factors; Determining target historical meteorological forecast data from the historical meteorological forecast data of the high-level meteorological model forecast factor using a similarity algorithm; Analyzing and combining the target historical weather forecast data using an ergodic method, assigning weights to the high-level weather model forecast factors in each combination, and determining the optimal high-level weather model forecast factor combination and the optimal weight combination; A short-term wind power prediction model is constructed based on the optimal high-level meteorological model forecast factor combination and the optimal weight combination.

2. The method according to claim 1, wherein The method of using a similarity algorithm to determine target historical meteorological forecast data from the historical meteorological forecast data of the high-level meteorological model forecast factor includes: Select the historical meteorological forecast data of each high-level meteorological model forecast factor at a certain moment as the current data; According to the high-level meteorological model forecast factor corresponding to the current data, the historical meteorological forecast data closest to the current data is selected from the historical meteorological forecast data as the target historical meteorological forecast data.

3. The method according to claim 2, wherein The ergodic method is used to analyze and combine the target historical weather forecast data, and weights are assigned to the high-level weather model forecast factors in each combination to determine the optimal high-level weather model forecast factor combination and the optimal weight combination, including: Using the actual short-term wind power value corresponding to the target historical weather forecast data as the short-term wind power forecast value corresponding to the current data; Constructing a sample set from current data of each high-level meteorological model forecast factor, a predicted value of short-term wind power corresponding to the current data, and an actual value of short-term wind power corresponding to the current data; Dividing the sample set into a training set and a test set; By combining the high-level meteorological model forecast factors in the training set using an ergodic method and assigning weights to the high-level meteorological model forecast factors in each combination, an optimal high-level meteorological model forecast factor combination and an optimal weight combination are selected; The combined scheme prediction value obtained by the optimal high-level meteorological model forecast factor combination and the optimal weight combination is verified through the test set, and a wind power short-term prediction model is constructed by the verified optimal high-level meteorological model forecast factor combination and the optimal weight combination.

4. The method according to claim 3, wherein The method combines the high-level meteorological model prediction factors in the training set using an ergodic method, assigns weights to the high-level meteorological model prediction factors in each combination, and selects the optimal high-level meteorological model prediction factor combination and the optimal weight combination, including: The optimal high-level meteorological model forecast factors in the training set are used to construct a combination scheme using the ergodic method, and the weights of each actual observation value in each combination scheme are assigned using the ergodic method. The predicted value of each combination scheme is obtained by multiplying the actual observed value in each combination scheme by its respective weight; The combination scheme and weight combination corresponding to the combination scheme prediction value closest to the actual value of the short-term wind power in the training set are respectively used as the combination of optimal high-level meteorological model forecast factors and the optimal weight combination.

5. The method according to claim 3, wherein The verification of the combined scheme prediction value obtained by the optimal high-level meteorological model forecast factor combination and the optimal weight combination through the test set includes: Selecting actual observation values ​​corresponding to the optimal high-level meteorological model forecast factor combination from the test set; Assign weights to the selected actual observations according to the best weight combination, and obtain the combined scheme prediction value of each group of schemes by multiplying the selected actual observations by the assigned weights. Subtracting the actual short-term wind power value in the test set from the predicted value of the combination scheme to obtain an error value; Determine whether the error value is less than the error threshold; if so, verify the optimal weight combination and the optimal high-level meteorological model forecast factor combination, and construct a short-term wind power prediction model based on the optimal high-level meteorological model forecast factor combination and the optimal weight combination; otherwise, fail the verification.

6. The method according to claim 3, wherein The method of using a pre-built short-term wind farm power prediction model to perform prediction processing on the high-level meteorological forecast data at the time to be predicted to obtain a short-term wind power prediction value at the time to be predicted includes: Inputting the high-level meteorological forecast data into a short-term wind farm power forecast model, and extracting historical meteorological forecast data at similar times by the short-term wind farm power forecast model according to each high-level meteorological model forecast factor in the optimal high-level meteorological model forecast factor combination; Calculate the distance between the historical weather forecast data at the similar time and the forecast data of each high-level weather model forecast factor at the time to be predicted, and select the actual value of wind power short-term corresponding to the time of the historical weather forecast data with the smallest distance to the forecast data of each high-level weather model forecast factor at the time to be predicted as the value corresponding to each high-level weather model forecast factor; Multiplying the value corresponding to each high-level meteorological model forecast factor by the weight of each high-level meteorological model forecast factor in the optimal weight combination to obtain a short-term wind power forecast value of each high-level meteorological model forecast factor; The short-term wind power forecast values ​​of the respective high-level meteorological model forecast factors in the optimal high-level meteorological model forecast factor combination are summed up as the short-term wind power forecast value of the wind farm at the time to be forecasted.

7. A device for implementing the method for short-term wind power prediction based on high-level meteorological data as described in any one of claims 1 to 6, characterized in that: include An acquisition module is used to obtain high-level meteorological forecast data of the wind farm at the time to be predicted; A prediction module is used to use a pre-built short-term wind farm power prediction model to perform prediction processing on the high-level meteorological forecast data at the time to be predicted, so as to obtain a short-term wind power prediction value at the time to be predicted; The pre-built short-term wind farm power prediction model is constructed by analyzing historical high-level meteorological forecast data of the wind farm and determining a combination of high-level meteorological model forecast factors and an optimal weight combination.

8. A computer device, characterized in that: include: one or more processors; The processor is configured to store one or more programs; When the one or more programs are executed by the one or more processors, the method for short-term wind power prediction based on high-level meteorological data according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium, characterized in that A computer program is stored thereon, and when the computer program is executed, the method for short-term wind power prediction based on high-level meteorological data as claimed in any one of claims 1 to 6 is implemented.

Images