Wind power prediction method, processor, system and storage medium

Abstract

The present application relates to the field of wind power generation technology, and to a method, processor, system, and storage medium for predicting wind power. The method comprises: obtaining historical meteorological data and historical wind power data of a wind farm in historical years; determining a first time corresponding to a maximum value and a second time corresponding to a minimum value in the historical wind speed data; using the time corresponding to the minimum wind speed data before the first time as a first node, and using the time corresponding to the maximum wind speed data after the second time as a second node; using the time point of the transition from spring to summer and the time point of the transition from summer to autumn in the historical year as a third node and a fourth node, respectively; dividing the historical year into four quarters according to the time corresponding to the above nodes; training an LSTM neural network model for each quarter, and importing the meteorological data of the wind farm at a target time into the LSTM neural network model of the quarter corresponding to the target time to obtain the predicted wind power of the wind farm at the target time.

Description

Technical Field

[0001] The present application relates to the field of wind power generation technology, and in particular to a wind power prediction method, processor, system and storage medium. Background Art

[0002] As wind power technology matures, the proportion of wind turbines generating electricity is gradually increasing. During wind power generation, wind power is significantly affected by meteorological factors such as wind speed, temperature, and wind direction, resulting in significant volatility and uncertainty. Therefore, accurate wind power forecasting is necessary to facilitate the development of power system generation plans.

[0003] In existing technology, LSTM neural network models are often trained using historical meteorological data and wind power data. The target meteorological data is then imported into the LSTM neural network model to output the predicted wind power at that target time. However, the accuracy of the predicted wind power output generated by this method is still low and cannot meet current wind power forecasting needs. Summary of the Invention

[0004] The purpose of the embodiments of the present application is to provide a wind power prediction method, processor, system and storage medium.

[0005] In order to achieve the above-mentioned objectives, the first aspect of the present application provides a wind power prediction method, which includes:

[0006] Obtain historical meteorological data of the wind farm in historical years and historical wind power data corresponding to the historical meteorological data, wherein the historical meteorological data includes historical wind speed data;

[0007] Determine a first time corresponding to a maximum wind speed value and a second time corresponding to a minimum wind speed value in the historical wind speed data;

[0008] The time corresponding to the minimum wind speed data before the first time is used as the first node, and the time corresponding to the maximum wind speed data after the second time is used as the second node;

[0009] The time point when spring and summer alternate in historical years is taken as the third node, and the time point when summer and autumn alternate in historical years is taken as the fourth node;

[0010] The historical year is divided into four quarters according to the time corresponding to the first node, the second node, the third node and the fourth node;

[0011] The historical meteorological data and historical wind power data are divided into four data sets according to four quarters;

[0012] Import each data set into the LSTM neural network model for training to obtain an LSTM neural network model corresponding to each quarter;

[0013] The meteorological data of the wind farm at the target time is imported into the LSTM neural network model of the quarter corresponding to the target time to obtain the predicted wind power of the wind farm at the target time.

[0014] In an embodiment of the present application, each data set is imported into the LSTM neural network model for training to obtain an LSTM neural network model corresponding to each quarter, including: importing the historical meteorological data in each data set into the LSTM neural network model to obtain the predicted wind power output by the LSTM neural network model; comparing the predicted wind power with the historical wind power data; adjusting the LSTM neural network model according to the comparison results to train the LSTM neural network model.

[0015] In an embodiment of the present application, the LSTM neural network model is adjusted according to the comparison results to train the LSTM neural network model, including: obtaining the absolute value of the difference between the predicted wind power and the historical wind power; when the absolute value is higher than or equal to the preset difference, adjusting the parameters of the LSTM neural network model; when the absolute value is lower than the preset difference, determining that the predicted wind power output by the LSTM neural network model is accurate; when the output accuracy of the LSTM neural network model reaches a preset threshold, determining that the LSTM neural network model training is completed.

[0016] In an embodiment of the present application, the prediction method further includes: after obtaining the historical wind speed data, outputting a filtered wind speed curve of the historical wind speed changing with time through a Kalman filter algorithm model; and determining the maximum wind speed and the minimum wind speed based on the wind speed curve.

[0017] In an embodiment of the present application, the prediction method further includes: when wind speed data at multiple heights are included at the same historical moment, taking the average value of the wind speed data at the multiple heights as the wind speed data at the historical moment.

[0018] In an embodiment of the present application, the prediction method further includes: after obtaining historical wind speed data and historical wind power data, determining a linear regression curve between the historical wind speed data and the historical wind power; determining an offset between each historical wind power data and the linear regression curve, wherein the historical wind power data is the historical wind speed data and the historical wind power data corresponding to the historical wind speed data; when the offset is greater than a preset offset, determining that the historical wind power data is abnormal data; and excluding the abnormal data to obtain the historical wind speed data and historical wind power data after the abnormality is excluded.

[0019] A second aspect of the present application provides a processor configured to execute the above-mentioned prediction method.

[0020] A third aspect of the present application provides a wind power prediction system, comprising:

[0021] A historical data acquisition module is used to acquire historical meteorological data of the wind farm in historical years and historical wind power data corresponding to the historical meteorological data, wherein the historical meteorological data includes historical wind speed data;

[0022] A quarterly division module is used to determine a first time corresponding to a maximum wind speed value and a second time corresponding to a minimum wind speed value in historical wind speed data; to use the time corresponding to the minimum wind speed data before the first time as a first node, and the time corresponding to the maximum wind speed data after the second time as a second node; to use the time point when the spring and summer seasons of the historical year alternate as a third node, and to use the time point when the summer and autumn seasons of the historical year alternate as a fourth node; and to divide the historical year into four quarters according to the time points corresponding to the first, second, third, and fourth nodes;

[0023] The neural network training module is used to divide the historical meteorological data and historical wind power data into four data sets according to four quarters; each data set is imported into the LSTM neural network model for training to obtain an LSTM neural network model corresponding to each quarter;

[0024] The neural network prediction module is used to import the meteorological data of the wind farm at the target time into the LSTM neural network model of the quarter corresponding to the target time to obtain the predicted wind power of the wind farm at the target time.

[0025] In an embodiment of the present application, the prediction system further includes: a filtering module for outputting a filtered wind speed curve of historical wind speed changes over time through a Kalman filter algorithm model after acquiring historical wind speed data.

[0026] A fourth aspect of the present application provides a machine-readable storage medium having instructions stored thereon. When the instructions are executed by a processor, the processor is configured to execute the above-mentioned prediction method.

[0027] This application improves the existing prediction method based on the rule that wind speed varies greatly with the seasons, but changes less in the same season, and can obtain more accurate predicted wind power values. However, the seasons change at different times in different regions, and the traditional four-season division will result in poor results in predicting wind power. Therefore, this application determines two seasonal alternation points based on the extreme values near the maximum wind speed in the historical years of the wind farm, combines the two traditional seasonal alternation time points, divides the year into four quarters, and trains an LSTM neural network model for each quarter. When predicting wind power for the target moment, the meteorological data at the target moment is imported into the LSTM neural network model of the quarter corresponding to the target moment, and a more accurate wind power can be output.

[0028] Other features and advantages of the embodiments of the present application will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the following detailed description, they are used to explain the embodiments of the present application but do not constitute a limitation on the embodiments of the present application. In the accompanying drawings:

[0030] Figure 1 The following schematically shows a flow chart of a method for predicting wind power according to an embodiment of the present application;

[0031] Figure 2 A wind speed curve showing the change of historical wind speed over time according to an embodiment of the present application is schematically shown;

[0032] Figure 3 The following schematically shows a structural block diagram of a wind power prediction system according to an embodiment of the present application;

[0033] Figure 4 The internal structure diagram of a computer device according to an embodiment of the present application is schematically shown. DETAILED DESCRIPTION

[0034] To make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the specific implementation methods described herein are only used to illustrate and explain the embodiments of the present application and are not used to limit the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0035] Figure 1 The following schematically shows a flow chart of a method for predicting wind power according to an embodiment of the present application. Figure 1As shown, in one embodiment of the present application, a method for predicting wind power is provided, for example, including the following steps:

[0036] S102 : Acquire historical meteorological data of the wind farm in historical years and historical wind power data corresponding to the historical meteorological data, wherein the historical meteorological data includes historical wind speed data.

[0037] S104: Determine a first time corresponding to a maximum wind speed value and a second time corresponding to a minimum wind speed value in the historical wind speed data.

[0038] S106: The time corresponding to the minimum wind speed data before the first time is used as the first node, and the time corresponding to the maximum wind speed data after the second time is used as the second node.

[0039] S108, taking the time point when the spring and summer seasons alternate in the historical year as the third node, and taking the time point when the summer and autumn seasons alternate in the historical year as the fourth node.

[0040] S110 , dividing the historical year into four quarters according to the time corresponding to the first node, the second node, the third node, and the fourth node.

[0041] S112, dividing the historical meteorological data and the historical wind power data into four data sets according to four quarters.

[0042] S114: Import each data set into the LSTM neural network model for training to obtain an LSTM neural network model corresponding to each quarter.

[0043] S116: Importing the meteorological data of the wind farm at the target time into the LSTM neural network model of the quarter corresponding to the target time to obtain the predicted wind power of the wind farm at the target time.

[0044] The wind power output by a wind turbine may be unstable due to factors such as wind speed and wind direction, causing the power grid system to need to predict the wind power output by the wind farm during the power dispatch process to ensure the stable operation of the power grid. In the prior art, an LSTM neural network model is often used to predict the wind power at the target time. However, the wind speed and wind direction in different seasons vary greatly, resulting in inaccurate predicted wind power output by the LSTM neural network model, which cannot meet the prediction needs of the power grid. In the embodiment of the present application, the data of historical years are divided into four data sets according to the quarter. An LSTM neural network model is trained for each quarter using the data of historical years, and the meteorological data at the target time is input into the LSTM neural model of the quarter corresponding to the target time to obtain the predicted wind power of the wind farm at the target time.

[0045] Specifically, the processor can obtain historical meteorological data for the wind farm in historical years and wind power data corresponding to the historical meteorological data. The historical meteorological data may include historical wind speed data, historical humidity data, historical wind direction data, historical temperature data, and historical air pressure data. The processor can determine a first time corresponding to the maximum wind speed in the historical wind speed data and a second time corresponding to the minimum wind speed. Seasonal transitions are strongly correlated with the maximum wind speed, and time points near the maximum wind speed can be identified as seasonal transition nodes. In this application, the time point corresponding to the minimum wind speed data before the first time point is used as the first node, and the time point corresponding to the maximum wind speed data after the second time point is used as the second node. The processor uses the time point at which the spring and summer seasons in the historical year transition as the third node, and the time point at which the summer and autumn seasons in the historical year transition as the fourth node. Based on the times corresponding to the four nodes, the historical year can be divided into four quarters. The processor can divide the historical meteorological data and historical wind power data for the historical year into four data sets according to the four quarters. Furthermore, the processor can import each data set into an LSTM neural network model for training to obtain an LSTM neural network model corresponding to each quarter. The processor can import the meteorological data of the wind farm at the target time into the LSTM neural network model of the quarter corresponding to the target time to obtain the predicted wind power of the wind farm at the target time.

[0046] For example, the processor can obtain historical meteorological data from January 1, 2018, to December 31, 2018, and historical wind power data corresponding to the historical meteorological data for a wind farm. The processor can determine a first time corresponding to the maximum wind speed and a second time corresponding to the minimum wind speed in the historical wind speed data. The processor can use the time corresponding to the minimum wind speed data before the first time as the first node and the time corresponding to the maximum wind speed data after the second time as the second node. The time point at which the spring and summer seasons transition in the historical year is used as the third node, and the time point at which the summer and autumn seasons transition in the historical year is used as the fourth node. For example, the processor determines from the historical wind speed data that the time corresponding to the maximum wind speed is November 22, 2018, and the time corresponding to the minimum wind speed is March 5, 2018. The processor can determine that the time corresponding to the first minimum wind speed before November 22, 2018 is November 20, 2018, and the time point corresponding to the first maximum wind speed after March 5, 2018 is March 10, 2018. The first node is midnight on November 20, 2018, and the second node is midnight on March 10, 2018. The third node corresponds to the transition between spring and summer at midnight on June 1, 2018, and the fourth node corresponds to the transition between summer and autumn at midnight on September 1, 2018. 2018 is divided into four quarters based on these four nodes. Winter is from midnight on January 1, 2018, to midnight on March 10, 2018, and from midnight on November 20, 2018, to midnight on December 31, 2018. Spring is from midnight on March 10, 2018, to midnight on June 1, 2018, summer is from midnight on June 1, 2018, to midnight on September 1, 2018, and autumn is from midnight on September 1, 2018, to midnight on November 22, 2018. The processor divides the historical meteorological data and wind power data into four datasets based on the four quarters. Each dataset is imported into the LSTM neural network model for training to produce an LSTM neural network model corresponding to each quarter. The processor can also use the four nodes of the historical year to divide the years after the historical year into quarters, and then divide the years after the historical year into four quarters. To determine the predicted wind power for February 1, 2019, the processor can import the meteorological data for February 1, 2019 into the winter LSTM neural network model, so that the winter LSTM neural network model outputs the predicted wind power for February 1, 2019.

[0047] Through the above method, the present application improves the existing prediction method based on the rule that wind speed varies greatly with the seasons, but changes less in the same season, and can obtain more accurate predicted wind power values. However, the seasons change at different times in different regions, and the effect of predicting wind power according to the traditional four-season division will be poor. Therefore, the present application determines two seasonal alternation points based on the extreme values near the maximum wind speed in the historical years of the wind farm, combines the two traditional seasonal alternation time points, divides the year into four quarters, and trains an LSTM neural network model for each quarter. When predicting wind power for the target moment, the meteorological data at the target moment is imported into the LSTM neural network model of the quarter corresponding to the target moment, and a more accurate wind power can be output.

[0048] In one embodiment, importing each data set into an LSTM neural network model for training to obtain an LSTM neural network model corresponding to each quarter includes: importing the historical meteorological data in each data set into the LSTM neural network model to obtain a predicted wind power output by the LSTM neural network model; comparing the predicted wind power with the historical wind power data; and adjusting the LSTM neural network model based on the comparison results to train the LSTM neural network model. The processor may import the historical meteorological data in each data set into the LSTM neural network model to obtain a predicted wind power output by the LSTM neural network model. The processor may compare the predicted wind power with the historical wind power data and, if it is determined that the difference between the output predicted wind power and the historical wind power is large, adjust the LSTM neural network model to train the LSTM neural network model.

[0049] In one embodiment, adjusting the LSTM neural network model based on the comparison results to train the LSTM neural network model includes: obtaining the absolute value of the difference between the predicted wind power and the historical wind power; adjusting the parameters of the LSTM neural network model if the absolute value is greater than or equal to a preset difference; determining that the predicted wind power output by the LSTM neural network model is accurate if the absolute value is less than the preset difference; and determining that the LSTM neural network model training is complete if the accuracy rate of the LSTM neural network model output reaches a preset threshold. A processor may obtain the absolute value of the difference between the predicted wind power and the historical wind power; if the absolute value is greater than or equal to the preset difference, the processor may determine that the predicted wind power output by the LSTM neural network model is inaccurate and adjust the parameters of the LSTM neural network model. If the absolute value is less than the preset difference, the processor may determine that the predicted wind power output by the LSTM neural network model is accurate. If the accuracy rate of the LSTM neural network model output reaches a preset threshold, the processor may determine that the LSTM neural network model training is complete and no further adjustments to the LSTM neural network model are made.

[0050] In one embodiment, the prediction method further includes: after acquiring the historical wind speed data, outputting a filtered wind speed curve of the historical wind speed over time using a Kalman filter algorithm model; and determining the maximum wind speed and the minimum wind speed based on the wind speed curve. After the processor acquires the historical wind speed data, the processor may output a filtered wind speed curve of the historical wind speed over time using a Kalman filter algorithm model, and the processor may determine the maximum wind speed and the minimum wind speed based on the wind speed curve.

[0051] In one embodiment, the prediction method further includes: when wind speed data for multiple heights is included at the same historical moment, taking the average of the wind speed data for the multiple heights as the wind speed data for the historical moment. When the processor obtains wind speed data for multiple heights at the same historical moment, the processor may take the average of the wind speed data for the multiple heights as the wind speed data for the historical moment.

[0052] For example, Figure 2 As shown in the figure, the wind speed curve of historical wind speed changes over time output by the Kalman filter algorithm model. In the figure, the horizontal axis is the sample number of the historical wind speed data. Each sample number corresponds to a historical time point. The sample numbers 1779 and 17106 in the figure are the summer and autumn alternation time points of adjacent years. Therefore, the wind speed data corresponding to sample numbers 1779 to 17106 are the wind speed data of one year in history. The vertical axis is the average wind speed at different heights. When the wind speed data at different heights are included at the same historical moment, the average value of the historical wind speed can be obtained to make the data more accurate. The processor can obtain the wind speed data from the sample numbers 1779 and 17106. Figure 2 The wind speed curve in the graph determines that the historical wind speed corresponding to sample number 4180 is the maximum wind speed, and the historical wind speed corresponding to sample number 10177 is the minimum wind speed. The time point corresponding to the minimum wind speed data before the first time point is used as the first node, and the time point corresponding to the maximum wind speed data after the second time point is used as the second node. The processor can use the time point corresponding to sample point A as the first node, and the time point corresponding to point B as the second node. The spring-summer transition time point corresponds to sample numbers 12448 on the graph, and the summer-autumn transition time point corresponds to sample numbers 1779 and 17106 on the graph. Based on these time points, the processor can divide the historical sample data into four data sets. Each data set is imported into an LSTM neural network model for training to obtain an LSTM neural network model corresponding to each season. The processor imports the meteorological data of the wind farm at the target time into the LSTM neural network model corresponding to the season at the target time, thereby obtaining the predicted wind power of the wind farm at the target time.

[0053] In one embodiment, the prediction method further includes: after obtaining the historical wind speed data and the historical wind power data, determining the linear regression curve between the historical wind speed data and the historical wind power; determining the offset between each piece of historical wind power data and the linear regression curve, wherein the historical wind power data is the historical wind speed data and the historical wind power data corresponding to the historical wind speed data; when the offset is greater than the preset offset, determining that the historical wind power data is abnormal data; excluding the abnormal data to obtain the historical wind speed data and the historical wind power data after the abnormality is eliminated. Because there is a strong correlation between wind speed and wind power, the processor can determine the linear regression curve between the historical wind speed data and the historical wind power after obtaining the historical wind speed data and the historical wind power data. The processor can determine the offset between each piece of historical wind power data and the linear regression curve. For example, the determined historical wind power data coordinates are (a, b), and the determined linear regression curve is y=cx 2 +d, then the offset between the historical wind power data and the linear regression curve is the distance between the coordinates of the historical wind power data and the linear regression curve. If the offset is greater than the preset offset, the processor determines that the historical wind power data is abnormal and excludes the abnormal data to obtain historical wind speed data and historical wind power data after the abnormality is eliminated. For example, if a wind turbine is shut down for maintenance or extreme weather conditions, the wind power output is zero, and the processor can determine that this data is abnormal and exclude it.

[0054] In one specific embodiment, a prediction method includes: a processor obtaining historical meteorological data and historical wind power data corresponding to a wind farm in historical years, wherein the historical meteorological data includes historical wind speed data. If wind speed data is obtained at multiple heights at a given historical moment, the processor may use the average of the wind speed data at the multiple heights as the wind speed data at the given historical moment. After obtaining the historical wind speed data and historical wind power data, the processor may determine a linear regression curve between the historical wind speed data and the historical wind power data, and determine an offset between each piece of historical wind power data and the linear regression curve, wherein the historical wind power data includes the historical wind speed data and the historical wind power data corresponding to the historical wind speed data. If the offset is greater than a preset offset, the processor may determine that the historical wind power data is abnormal data and exclude the abnormal data to obtain abnormally excluded historical wind speed data and historical wind power data. The processor may output a filtered wind speed curve of the historical wind speed over time using a Kalman filter algorithm model, and determine a maximum wind speed and a minimum wind speed based on the wind speed curve. The processor may determine a first time corresponding to the maximum wind speed and a second time corresponding to the minimum wind speed in the historical wind speed data. The processor may use the time corresponding to the minimum wind speed data before the first time as the first node, and the time corresponding to the maximum wind speed data after the second time as the second node. The processor may use the time point at which the spring and summer seasons transition in the historical year as the third node, and the time point at which the summer and autumn seasons transition in the historical year as the fourth node. The processor may divide the historical year into four quarters based on the times corresponding to the first, second, third, and fourth nodes. The processor may divide the historical meteorological data and historical wind power data into four data sets according to the four quarters. The historical meteorological data in each data set is imported into the LSTM neural network model to obtain the predicted wind power output by the LSTM neural network model. The processor may obtain the absolute value of the difference between the predicted wind power and the historical wind power. If the absolute value is greater than or equal to a preset difference, the processor may adjust the parameters of the LSTM neural network model. If the absolute value is less than the preset difference, the processor determines that the predicted wind power output by the LSTM neural network model is accurate. If the accuracy of the LSTM neural network model output reaches a preset threshold, the processor determines that the LSTM neural network model training is complete. The processor imports the meteorological data of the wind farm at the target time into the LSTM neural network model of the quarter corresponding to the target time to obtain the predicted wind power of the wind farm at the target time.

[0055] In the above technical solution, this application improves the existing prediction method based on the rule that wind speed varies greatly with the seasons, but changes less in the same season, so as to obtain a more accurate predicted wind power value. However, the seasons change at different times in different regions, and the effect of predicting wind power according to the traditional four-season division will be poor. Therefore, this application determines two seasonal alternation points based on the extreme values near the maximum wind speed in the historical years of the wind farm, combines the two traditional seasonal alternation time points, divides the year into four quarters, and trains an LSTM neural network model for each quarter. When predicting wind power for the target moment, the meteorological data at the target moment is imported into the LSTM neural network model of the quarter corresponding to the target moment, and a more accurate wind power can be output.

[0056] Figure 1 FIG. 1 is a flow chart of a method for predicting wind power in one embodiment. It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. In addition, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0057] In one embodiment, Figure 3 As shown, a wind power prediction system 300 is provided, including a historical data acquisition module 301, a quarterly division module 302, a neural network training module 303 and a neural network prediction module 304, wherein:

[0058] The historical data acquisition module 301 is used to acquire historical meteorological data of the wind farm in historical years and historical wind power data corresponding to the historical meteorological data, wherein the historical meteorological data includes historical wind speed data.

[0059] The quarterly division module 302 is used to determine the first time corresponding to the maximum wind speed and the second time corresponding to the minimum wind speed in the historical wind speed data; the time corresponding to the minimum wind speed data before the first time is used as the first node, and the time corresponding to the maximum wind speed data after the second time is used as the second node; the time point when the spring and summer seasons of the historical year alternate as the third node, and the time point when the summer and autumn seasons of the historical year alternate as the fourth node; the historical year is divided into four quarters according to the time corresponding to the first node, the second node, the third node and the fourth node.

[0060] The neural network training module 303 is used to divide the historical meteorological data and the historical wind power data into four data sets according to four quarters; each data set is imported into the LSTM neural network model for training to obtain an LSTM neural network model corresponding to each quarter.

[0061] The neural network prediction module 304 is used to import the meteorological data of the wind farm at the target time into the LSTM neural network model of the quarter corresponding to the target time to obtain the predicted wind power of the wind farm at the target time.

[0062] In one embodiment, the prediction system further includes: a filtering module for outputting a filtered wind speed curve showing changes in historical wind speed over time through a Kalman filter algorithm model after acquiring historical wind speed data.

[0063] The wind power prediction system includes a processor and a memory. The above-mentioned historical data acquisition module, quarterly division module, neural network training module and neural network prediction module are all stored in the memory as program units, and the processor executes the above-mentioned program modules stored in the memory to realize corresponding functions.

[0064] The processor includes a core, which retrieves the corresponding program unit from the memory. One or more cores can be set, and the wind power prediction method is implemented by adjusting the core parameters.

[0065] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0066] An embodiment of the present application provides a machine-readable storage medium having a program stored thereon, which implements the above-mentioned wind power prediction method when executed by a processor.

[0067] An embodiment of the present application provides a processor, which is configured to run a program, wherein the program executes the above-mentioned wind power prediction method when running.

[0068] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor A01, a network interface A02, a display screen A04, an input device A05 and a memory (not shown in the figure) connected via a system bus. Among them, the processor A01 of the computer device is used to provide computing and control capabilities. The memory of the computer device includes an internal memory A03 and a non-volatile storage medium A06. The non-volatile storage medium A06 stores an operating system B01 and a computer program B02. The internal memory A03 provides an environment for the operation of the operating system B01 and the computer program B02 in the non-volatile storage medium A06. The network interface A02 of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor A01, a method for predicting wind power is implemented. The display screen A04 of the computer device can be a liquid crystal display or an electronic ink display, and the input device A05 of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad provided on the computer device housing, or an external keyboard, touchpad or mouse.

[0069] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0070] An embodiment of the present application provides a device, comprising a processor, a memory, and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the following steps are implemented: obtaining historical meteorological data of a wind farm in historical years and historical wind power data corresponding to the historical meteorological data, wherein the historical meteorological data includes historical wind speed data; determining a first time corresponding to a maximum wind speed and a second time corresponding to a minimum wind speed in the historical wind speed data; using the time corresponding to the minimum wind speed data before the first time as a first node, and using the time corresponding to the maximum wind speed data after the second time as a second node; using the time point at which the spring and summer seasons alternate in the historical year as a third node, and using the time point at which the summer and autumn seasons alternate in the historical year as a fourth node; dividing the historical year into four quarters according to the times corresponding to the first node, the second node, the third node, and the fourth node; dividing the historical meteorological data and the historical wind power data into four data sets according to the four quarters; importing each data set into an LSTM neural network model for training to obtain an LSTM neural network model corresponding to each quarter; and importing the meteorological data of the wind farm at a target time into the LSTM neural network model of the quarter corresponding to the target time to obtain predicted wind power of the wind farm at the target time.

[0071] In one embodiment, each data set is imported into an LSTM neural network model for training to obtain an LSTM neural network model corresponding to each quarter, including: importing historical meteorological data in each data set into the LSTM neural network model to obtain predicted wind power output by the LSTM neural network model; comparing the predicted wind power with historical wind power data; and adjusting the LSTM neural network model according to the comparison results to train the LSTM neural network model.

[0072] In one embodiment, adjusting the LSTM neural network model according to the comparison results to train the LSTM neural network model includes: obtaining the absolute value of the difference between the predicted wind power and the historical wind power; when the absolute value is higher than or equal to a preset difference, adjusting the parameters of the LSTM neural network model; when the absolute value is lower than the preset difference, determining that the predicted wind power output by the LSTM neural network model is accurate; when the output accuracy of the LSTM neural network model reaches a preset threshold, determining that the LSTM neural network model training is completed.

[0073] In one embodiment, the prediction method further includes: after acquiring the historical wind speed data, outputting a filtered wind speed curve showing the historical wind speed changing with time through a Kalman filter algorithm model; and determining the maximum wind speed and the minimum wind speed according to the wind speed curve.

[0074] In one embodiment, the prediction method further includes: when wind speed data at multiple heights are included at the same historical moment, taking the average of the wind speed data at the multiple heights as the wind speed data at the historical moment.

[0075] In one embodiment, the prediction method further includes: after obtaining historical wind speed data and historical wind power data, determining a linear regression curve between the historical wind speed data and the historical wind power; determining an offset between each historical wind power data and the linear regression curve, wherein the historical wind power data is the historical wind speed data and the historical wind power data corresponding to the historical wind speed data; when the offset is greater than a preset offset, determining that the historical wind power data is abnormal data; and excluding the abnormal data to obtain the historical wind speed data and historical wind power data after the abnormality is eliminated.

[0076] 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.

[0077] 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.

[0078] 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.

[0079] 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 The steps for the function specified in one or more boxes.

[0080] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0081] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0082] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0083] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0084] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A method for predicting wind power, characterized in that: The prediction method comprises: Acquire historical meteorological data of the wind farm in historical years and historical wind power data corresponding to the historical meteorological data, wherein the historical meteorological data includes historical wind speed data; Determine a first time corresponding to a maximum wind speed value and a second time corresponding to a minimum wind speed value in the historical wind speed data; The time corresponding to the minimum wind speed data before the first time is used as the first node, and the time corresponding to the maximum wind speed data after the second time is used as the second node; The time point when the spring and summer seasons of the historical year alternate is used as the third node, and the time point when the summer and autumn seasons of the historical year alternate is used as the fourth node; Dividing the historical year into four quarters according to the time corresponding to the first node, the second node, the third node, and the fourth node; Dividing the historical meteorological data and the historical wind power data into four data sets according to the four quarters; Importing each data set into the LSTM neural network model for training to obtain the LSTM neural network model corresponding to each quarter; The meteorological data of the wind farm at the target time is imported into the LSTM neural network model of the quarter corresponding to the target time to obtain the predicted wind power of the wind farm at the target time.

2. The prediction method according to claim 1, characterized in that Importing each data set into the LSTM neural network model for training to obtain the LSTM neural network model corresponding to each quarter includes: Importing the historical meteorological data in each data set into the LSTM neural network model to obtain the predicted wind power output by the LSTM neural network model; comparing the predicted wind power with the historical wind power data; The LSTM neural network model is adjusted according to the comparison result to train the LSTM neural network model.

3. The prediction method according to claim 2, characterized in that The adjusting the LSTM neural network model according to the comparison result to train the LSTM neural network model includes: Obtaining an absolute value of a difference between the predicted wind power and the historical wind power; When the absolute value is greater than or equal to a preset difference, adjusting the parameters of the LSTM neural network model; When the absolute value is lower than the preset difference, determining that the predicted wind power output by the LSTM neural network model is accurate; When the output accuracy of the LSTM neural network model reaches a preset threshold, it is determined that the training of the LSTM neural network model is completed.

4. The prediction method according to claim 1, wherein: Also includes: After obtaining the historical wind speed data, a Kalman filter algorithm model is used to output a wind speed curve showing the filtered historical wind speed changing over time; The maximum wind speed and the minimum wind speed are determined according to the wind speed curve.

5. The prediction method according to claim 1, wherein: Also includes: When wind speed data at multiple heights are included at the same historical moment, the average value of the wind speed data at the multiple heights is used as the wind speed data at the historical moment.

6. The prediction method according to claim 1, characterized in that Also includes: After acquiring the historical wind speed data and the historical wind power data, determining a linear regression curve between the historical wind speed data and the historical wind power; Determining an offset between each piece of historical wind power data and the linear regression curve, wherein the historical wind power data is the historical wind speed data and the historical wind power data corresponding to the historical wind speed data; When the offset is greater than a preset offset, determining that the historical wind power data is abnormal data; The abnormal data is eliminated to obtain historical wind speed data and historical wind power data after the abnormalities are eliminated.

7. A processor, characterized in that: The method is configured to execute the prediction method according to any one of claims 1 to 6.

8. A wind power prediction system, characterized in that: include: A historical data acquisition module is used to acquire historical meteorological data of the wind farm in historical years and historical wind power data corresponding to the historical meteorological data, wherein the historical meteorological data includes historical wind speed data; A quarterly division module is configured to determine a first time corresponding to a maximum wind speed value and a second time corresponding to a minimum wind speed value in the historical wind speed data; use the time corresponding to the minimum wind speed data before the first time as a first node, and use the time corresponding to the maximum wind speed data after the second time as a second node; use the time point of the alternation between the spring and summer seasons of the historical year as a third node, and use the time point of the alternation between the summer and autumn seasons of the historical year as a fourth node; and divide the historical year into four quarters based on the times corresponding to the first node, the second node, the third node, and the fourth node; A neural network training module is used to divide the historical meteorological data and the historical wind power data into four data sets according to the four quarters; each data set is imported into an LSTM neural network model for training to obtain the LSTM neural network model corresponding to each quarter; A neural network prediction module is used to import the meteorological data of the wind farm at the target time into the LSTM neural network model of the quarter corresponding to the target time to obtain the predicted wind power of the wind farm at the target time.

9. The prediction system according to claim 8, characterized in that Also includes: The filtering module is used to output a wind speed curve of the filtered historical wind speed changing with time through the Kalman filter algorithm module after acquiring the historical wind speed data.

10. A machine-readable storage medium having instructions stored thereon, characterized in that: When the instruction is executed by a processor, the processor is configured to perform the prediction method according to any one of claims 1 to 6.

Images