Wind power prediction error decoupling analysis method, processor and storage medium
By decomposing the wind power prediction process into numerical weather prediction, wind-to-power conversion model, and prediction result correction, a fitting model was established and decoupling analysis was performed. This solved the problem of accuracy in wind power prediction, reduced the operating cost of the power system, and improved its security.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- STATE GRID HUNAN ELECTRIC POWER CO LTD ELECTRIC POWER SCI RES INST
- Filing Date
- 2022-10-28
- Publication Date
- 2026-05-12
AI Technical Summary
Existing wind power forecasting technologies are unable to accurately depict the inherent characteristics of wind power generation, resulting in errors in the forecast results. Furthermore, they do not consider the impact of changes in meteorological data on errors at each stage, leading to an increase in the reserve capacity of the power system and higher operating costs.
The wind power prediction process is decomposed into three stages: numerical weather prediction, wind-to-power conversion model, and prediction result correction. A wind power conversion fitting model is established, and the prediction results under real meteorological conditions and numerical weather prediction conditions are decoupled and analyzed to calculate the error of each stage and determine its proportion.
It enables precise location and decoupling of wind power forecasting errors, provides theoretical guidance for improving forecasting methods, reduces the operating costs of power systems, and improves safety and reliability.
Smart Images
Figure CN115526429B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of wind power prediction, and in particular to a method for decoupling analysis of wind power prediction errors, a processor, and a storage medium. Background Technology
[0002] Wind energy is an important renewable and clean energy source with abundant reserves and wide distribution, making wind power generation a crucial research direction in the development and utilization of renewable energy. However, wind resources in nature are characterized by randomness and intermittency. Random wind speed and direction cause significant fluctuations in the output power of wind turbines, posing a significant threat to the stability of power system voltage and frequency. When wind power exceeds a certain proportion of the power system, the safety and stability of the entire system face enormous challenges. To effectively absorb the intermittent and random nature of wind power and other renewable energy sources, the power system needs to increase additional reserve capacity, leading to an increase in the total operating cost of the power system. To reduce the configuration cost of power system reserve capacity and improve the overall safety and reliability of power system operation, accurately predicting wind power output is a critical problem that urgently needs to be solved in the field of power system operation and control.
[0003] Existing wind power forecasting technologies struggle to accurately characterize and represent the inherent properties of wind power generation, inevitably resulting in errors in the forecasts. Conducting an objective and comprehensive analysis and evaluation of these errors helps identify key influencing factors and weaknesses in the forecasting process, thereby guiding further improvements and optimizations to existing wind power forecasting technologies and significantly enhancing forecast accuracy. However, current wind power forecasting error analysis methods do not consider the changes in errors at each stage of the forecasting process due to variations in meteorological data and other conditions. Ignoring these error variation characteristics significantly reduces the reliability of existing wind power forecasting error analysis methods. Summary of the Invention
[0004] The purpose of this invention is to provide a method, processor, and storage medium for decoupling analysis of wind power prediction errors.
[0005] To achieve the above objectives, the first aspect of this invention provides a method for decoupling analysis of wind power prediction errors. The wind power prediction process includes a numerical weather prediction stage, a wind-to-power conversion model stage, and a prediction result correction stage. The method for decoupling analysis of wind power prediction errors includes:
[0006] The data required for wind power prediction error analysis is obtained from the historical operation data of wind farms. The data includes measured weather data, numerical weather forecast data, installed capacity, real-time operating capacity, planned operating capacity, wind power prediction data, and actual output power of wind farms.
[0007] Determine the first target error caused by the prediction result correction process;
[0008] The wind-to-power conversion model of the wind farm is fitted based on the required data to establish a wind power conversion fitting model;
[0009] The measured weather data of the wind farm is input into the wind power conversion fitting model to obtain the model output under real meteorological conditions;
[0010] The predicted wind power under real weather conditions is determined based on the installed capacity, planned operating capacity, and model output under real weather conditions.
[0011] The correction error of the forecast result under actual weather conditions is determined based on the real-time operating capacity, the planned operating capacity, and the wind power forecast power under actual weather conditions.
[0012] The wind-to-power conversion model error under real weather conditions is determined based on the predicted wind power under real weather conditions, the actual output power of the wind farm, and the correction error of the prediction results under real weather conditions.
[0013] Obtain the model output under numerical weather forecast conditions corresponding to the model output under real meteorological conditions;
[0014] The second objective error caused by the wind-to-power conversion model is determined based on the model output under real meteorological conditions, the model output under numerical weather forecast, and the wind-to-power conversion model error under real meteorological conditions.
[0015] The third objective error caused by the numerical weather prediction process is determined based on wind power forecast data, actual wind farm output power, the first objective error, and the second objective error; and
[0016] The proportion of each objective error is determined based on the first objective error, the second objective error, and the third objective error.
[0017] In this embodiment of the invention, the wind power prediction error decoupling analysis method further includes:
[0018] Based on the percentage of error for each target, the main factors causing errors in each stage of the wind power prediction process are analyzed.
[0019] In this embodiment of the invention, determining the first target error caused by the prediction result correction step includes:
[0020] The equivalent wind power forecast value is determined based on wind power forecast data, real-time operating capacity and planned operating capacity.
[0021] The first target error is obtained by subtracting the equivalent wind power prediction value from the wind power prediction data.
[0022] In this embodiment of the invention, the wind-to-power conversion model of the wind farm is fitted according to the required data to establish a wind power conversion fitting model, including:
[0023] The wind-to-power conversion model output is determined based on installed capacity, planned operating capacity, and wind power forecast data.
[0024] Numerical weather forecast data is used as input to the wind-to-power conversion model, and the data, together with the output of the wind-to-power conversion model, forms the training set and the test set.
[0025] XGBoost was used to train the training set to fit the wind-to-electricity conversion model;
[0026] After training is completed, the test set is used to determine whether the fitted wind-to-power conversion model meets the accuracy requirements;
[0027] Modify the initial parameters of XGBoost and repeat the training on the training set until the accuracy requirement is met;
[0028] The wind-to-power conversion model that meets the accuracy requirements is used as the wind power conversion fitting model.
[0029] In this embodiment of the invention, the wind-to-electricity conversion model includes one of the following:
[0030] Wind-to-power conversion models based on BP neural networks, wind-to-power conversion models based on LSTM recurrent neural networks, and wind-to-power conversion models based on CNN-LSTM hybrid neural networks.
[0031] In this embodiment of the invention, determining the predicted wind power under actual weather conditions based on the installed capacity, planned operating capacity, and model output under actual weather conditions includes:
[0032] The predicted wind power under actual meteorological conditions is calculated using the following formula:
[0033]
[0034] Among them, P wind It is the predicted wind power under actual meteorological conditions, C′ o C is the planned startup capacity, F is the installed capacity, and f is the planned startup capacity. wind,m It is the model output under real weather conditions.
[0035] In this embodiment of the invention, the correction error of the prediction result under actual weather conditions is determined based on the real-time operating capacity, the planned operating capacity, and the predicted wind power under actual weather conditions, including:
[0036] The correction error for the forecast results under actual meteorological conditions is calculated using the following formula:
[0037]
[0038] Among them, E wind,r It is the correction error of the forecast result under actual meteorological conditions, C o This is the real-time boot capacity, C′ o This is the planned startup capacity, P wind It is the predicted wind power output under actual weather conditions.
[0039] In this embodiment of the invention, the wind-to-power conversion model error under real weather conditions is determined based on the predicted wind power under real weather conditions, the actual output power of the wind farm, and the correction error of the prediction results under real weather conditions, including:
[0040] The error of the wind-to-electricity conversion model under real meteorological conditions is calculated using the following formula:
[0041] P wind =P m +E wind,m +E wind,r
[0042] Among them, P wind It is the predicted wind power output under actual weather conditions, P m E is the actual output power of the wind farm. wind,m It is the error of the wind-to-power conversion model under real meteorological conditions, and E wind,r It is the correction error of the prediction results under actual meteorological conditions.
[0043] In this embodiment of the invention, the second target error caused by the wind-to-power conversion model is determined based on the model output under real meteorological conditions, the model output under numerical weather forecasting, and the wind-to-power conversion model error under real meteorological conditions, including:
[0044] The second objective error is calculated using the following formula:
[0045] E m +f m =E wind,m +f wind, m
[0046] Among them, E m It is the second objective error, f m This is the model output from numerical weather prediction, E wind,m It is the error of the wind-to-power conversion model under real meteorological conditions, and f wind,m It is the model output under real weather conditions.
[0047] In this embodiment of the invention, the third target error caused by the numerical weather prediction process is determined based on wind power prediction data, actual wind farm output power, a first target error, and a second target error, including:
[0048] The third objective error is calculated using the following formula:
[0049] P p =P m +E n +E m +E r
[0050] Among them, P p This is wind power forecast data, E r It is the first target error, E m It is the second objective error, and E n This is the third objective error.
[0051] A second aspect of the present invention provides a processor configured to perform the above-described wind power prediction error decoupling analysis method.
[0052] A third aspect of the present invention provides a machine-readable storage medium storing instructions that, when executed by a processor, cause the processor to implement the aforementioned wind power prediction error decoupling analysis method.
[0053] In wind power forecasting, wind-to-power conversion models are often trained using numerical weather prediction (NMR) data, and the correction of prediction results depends on the output of these models. Therefore, directly comparing the prediction results under NMR with those under actual weather conditions, while allowing for the resolution of errors caused by NMR, fails to accurately decouple the wind-to-power conversion model from the correction of prediction results. The technical solution provided in this invention, based on prediction results under both actual and NMR conditions, and comprehensively considering the impact of changes in weather conditions on each stage, can accurately decouple the error-causing stages. Furthermore, it calculates the error proportion of each stage, providing theoretical guidance for improving wind farm forecasting methods.
[0054] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description
[0055] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings:
[0056] Figure 1A schematic flowchart illustrating the wind power prediction error decoupling analysis method according to an embodiment of the present invention is shown.
[0057] Figure 2 The diagram illustrates the internal structure of a computer device according to an embodiment of this application. Detailed Implementation
[0058] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.
[0059] To address the shortcomings of existing wind power forecasting error analysis methods that do not consider the changes in error at each stage of prediction after changes in meteorological data and other conditions, this invention proposes a decoupling analysis method for wind power forecasting errors that takes into account the characteristics of error changes. This provides a more reliable basis for improving existing wind power forecasting methods and identifying key influencing factors of forecasting errors.
[0060] As the overall inventive concept, firstly, the wind power prediction process is decomposed into three key stages: numerical weather prediction, wind-to-power conversion model, and prediction result correction. Secondly, a fitting model is constructed for the wind-to-power conversion model stage, fitting the output of the wind-to-power conversion model under real meteorological conditions. Based on this, considering the prediction results under real meteorological conditions and numerical weather prediction conditions, and taking into account the error variations in each prediction stage, the prediction errors of the same stage under different conditions are solved separately, and an error decoupling equation is established. Finally, based on the error decoupling equation, the errors of each stage are solved to achieve precise location of weak links in wind power prediction.
[0061] Figure 1 A schematic flowchart illustrating a wind power prediction error decoupling analysis method according to an embodiment of the present invention is shown. Specifically, in this embodiment, the wind power prediction error decoupling analysis method may include the following steps.
[0062] In step S101, the data required for wind power prediction error analysis is obtained from the historical operation data of the wind farm. The data includes measured weather data of the wind farm, numerical weather forecast data, installed capacity C, and real-time operating capacity C. o Planned startup capacity C′ o Wind power forecast data P p Actual output power P of wind farm m .
[0063] Specifically, measured weather data for wind farms can include data collected by data acquisition devices (such as wind measurement towers). In one example, measured weather data for wind farms can include measured wind speeds (e.g., wind speeds at different altitudes, such as 10 meters, 70 meters, etc.) at multiple times within a period of time (e.g., a month, a quarter, a year, or many years). In further examples, measured weather data for wind farms can also include measured wind direction, measured air temperature, measured humidity, measured air pressure, etc.
[0064] Numerical weather forecast data can include numerical weather forecast data for the location of the wind farm, and can include a variety of meteorological factors, such as predicted wind speed, predicted wind direction, predicted temperature, predicted humidity, and predicted air pressure at the wind farm.
[0065] In one example, numerical weather prediction data may include predicted wind speeds from wind farms (e.g., predicted wind speeds at different altitudes, such as 10 meters, 70 meters, etc.). In further examples, numerical weather prediction data may also include predicted wind direction, predicted air temperature, predicted humidity, and predicted air pressure from wind farms.
[0066] Numerical Weather Prediction (NWP) is a method that uses large computers to perform numerical calculations based on actual atmospheric conditions and certain initial and boundary conditions to solve the fluid dynamics and thermodynamic equations describing the evolution of weather, and to predict the state of atmospheric motion and weather phenomena in the future for a certain period of time.
[0067] Wind power forecast data P p The wind power prediction data (predicted power) can be predicted by the wind farm's prediction model or by the highway prediction service provider, and the wind farm side will report the stored wind power prediction data (predicted power) to the dispatch side server.
[0068] Actual output power P of wind farm m It can be obtained through on-site measurement using a power measuring device.
[0069] In step S102, the first target error E caused by the prediction result correction step is determined. r .
[0070] Specifically, based on the wind power prediction process, the wind farm power prediction process can be divided into three key stages: numerical weather prediction, wind-to-power conversion model, and prediction result correction. The errors of these three key stages are E... n E m E r The predicted power of a wind farm can be expressed as:
[0071] P p =P m +E n +E m +E r Formula (1)
[0072] Based on wind power prediction data P p Real-time operating capacity of wind farm C o and planned startup capacity C' o Calculate the equivalent wind power prediction value P under accurate (real-time) operating capacity conditions. cap As shown in formula (2):
[0073]
[0074] Wind power forecast data P p With the predicted equivalent wind power value P cap Subtraction yields the error in the prediction result correction (i.e., the first target error) E. r As shown in formula (3):
[0075]
[0076] In step S103, the wind-to-electricity conversion model of the wind farm is fitted according to the required data to establish a wind-to-electricity conversion fitting model.
[0077] In order to E n E m Decoupling is performed, and the wind-to-power conversion model of the wind farm is fitted based on the data in step S101. Specifically, examples of wind-to-power conversion models may include one of the following: a wind-to-power conversion model based on a BP neural network, a wind-to-power conversion model based on an LSTM (Long Short-Term Memory) recurrent neural network, and a wind-to-power conversion model based on a CNN-LSTM hybrid neural network.
[0078] A wind-to-electricity conversion model is a mathematical model that describes the relationship between wind resource meteorological elements and the active power of wind power generation equipment. In actual production, due to the influence of weather conditions, generator performance, and other factors, wind resource meteorological elements and electricity often exhibit a complex mapping relationship. In order to ensure the accuracy of power prediction, a wind-to-electricity conversion fitting model can be established for a specific wind farm and a specific time period to better meet the requirements of power prediction.
[0079] Specifically, step S103 may include the following sub-steps.
[0080] In sub-step S1031, the installed capacity C and the planned startup capacity C′ are used. oCompared with wind power forecast data P p Calculate the output f of the wind-to-electricity conversion model m As shown in formula (4):
[0081]
[0082] In step S1032, numerical weather prediction data is selected as the input for the wind-to-power conversion model (hereinafter referred to as model input), f m As the output of the wind-to-power conversion model, the model input and output data are normalized. Numerical weather forecast data (e.g., predicted wind speed, and optionally, predicted wind direction, predicted temperature, predicted humidity, predicted air pressure, etc.) for a predetermined period (e.g., one month, one quarter, one year, or many years) are selected from historical wind farm operation data, along with the calculated corresponding wind-to-power conversion model output f. m This is used to construct the dataset. The dataset can be divided into a training set and a test set according to a certain ratio;
[0083] In step S1033, the training set is trained to fit the wind power conversion model. Specifically, XGBoost can be used to train the training set.
[0084] In step S1034, after training is completed (e.g., after a set number of training iterations are reached), the test set is used to determine whether the fitted wind-to-power conversion model meets the accuracy requirements.
[0085] In step S1035, if the accuracy requirements are not met, the initial parameters of XGBoost can be modified, and step S1034 can be repeated until the fitted wind-to-power conversion model meets the accuracy requirements. The fitted wind-to-power conversion model that meets the accuracy requirements can be used as the established wind power conversion fitting model.
[0086] The wind power conversion fitting model establishment (or fitting) scheme provided in this embodiment of the invention can spontaneously obtain an accurate wind power conversion model when it is impossible or difficult to obtain an existing wind power conversion model.
[0087] In step S104, the measured weather data of the wind farm is input into the wind power conversion fitting model to obtain the model output f under the actual meteorological conditions. wind,m .
[0088] In step S105, based on the installed capacity C and the planned operating capacity C′ o and the model output f under real meteorological conditions wind,m Determine the predicted wind power P under actual meteorological conditions wind ;
[0089] Specifically, measured weather data from the wind farm (such as data collected by a wind tower, such as measured wind speed) can be input into the wind power conversion fitting model to obtain the model output f under real meteorological conditions. wind,m And the predicted wind power P under real meteorological conditions can be calculated according to formula (5). wind :
[0090]
[0091] In step S106, based on the real-time power-on capacity C o Planned startup capacity C′ o And the correction error E of the wind power forecast under actual weather conditions to determine the forecast result under actual weather conditions. wind,r .
[0092] Specifically, under real meteorological conditions, since there are no errors caused by numerical weather forecasts, the predicted wind power P is... wind The prediction error only includes the error E from the wind-to-power conversion model. wind,m Error E in the correction of prediction results wind,r Therefore:
[0093] P wind =P m +E wind,m +E wind,r Formula (6)
[0094] Furthermore, similar to formula (3):
[0095]
[0096] In step S107, the wind power forecast P is based on the actual weather conditions. wind Actual output power P of wind farm m And the correction error E of the prediction results under actual meteorological conditions wind,r Determine the wind-to-power conversion model error E under real meteorological conditions. wind,m .
[0097] Specifically, the error E calculated according to formula (7) wind,r Substituting into formula (6), we can obtain the wind-to-power conversion model error E under real meteorological conditions. wind,m .
[0098] In step S108, the model output f under real meteorological conditions is obtained. wind,m The corresponding numerical weather prediction model output f m .
[0099] Here, "corresponding" in the context of the model output under real meteorological conditions versus the model output under numerical weather forecasts refers to the values under the same reference object under two different conditions. For example, assuming a wind farm has a measured wind speed of x at a certain reference time, inputting it into a wind power conversion fitting model yields a model output of f. wind,m (x), then the corresponding numerical weather prediction model output can be calculated according to formula (4). At this time, the wind power prediction data P in formula (4) p This is the wind power prediction data obtained at that reference time.
[0100] In step S109, the model output f is based on the actual meteorological conditions. wind,m Model output f under numerical weather prediction m And the wind-to-power conversion model error E under real meteorological conditions wind,m Determine the second objective error E caused by the wind-to-power conversion model components. m .
[0101] Specifically, when considering only the errors in the wind-to-power conversion model, the power prediction error E of the wind-to-power conversion model... m and E wind,m The wind-to-power conversion model output f is given by input data and input data respectively, based on numerical weather prediction data and real meteorological data. m f wind,m The following relationship exists:
[0102] E m +f m =E wind,m +f wind,m Formula (8)
[0103] In step S110, based on the wind power prediction data P p Actual output power P of wind farm m First target error E r and the second target error E m Determine the third objective error E caused by the numerical weather prediction process. n .
[0104] Specifically, the third target error E can be calculated according to formula (1). n .
[0105] An error decoupling equation can be established based on the formulas in the above steps:
[0106]
[0107] The first target error E can be calculated using this error decoupling equation. r Second target error E mThird target error E n .
[0108] In step S111, based on the first target error E r Second target error E m and the third target error E n Determine the percentage of error for each objective.
[0109] Specifically, the proportion of each target error can be calculated according to formula (9):
[0110]
[0111]
[0112]
[0113] Among them, R n R m R r These represent the percentages of the third objective error, the second objective error, and the first objective error, respectively.
[0114] After determining the proportion of each target error, the method may further include a step of analyzing the main factors causing errors in each stage of the wind power prediction process based on the proportion of each target error. Specifically, for R... n R m R r Compare, if R n If R is the maximum value, then numerical weather prediction is the main factor causing the error; if R... m If R is at its maximum value, then the wind-to-power conversion model is the main factor causing the error; if R... r If the value is the maximum, then the correction of the prediction result becomes the main factor causing the error.
[0115] It is important to emphasize that because the input meteorological data affects the training of the actual wind farm prediction model, which in turn affects the output of the "wind-to-power conversion model," and further affects the correction of the prediction results, it is not appropriate to simply assume that the error (second error) E caused by the wind-to-power conversion model under numerical weather prediction is the only factor to be considered. m Error E between the wind-to-power conversion model and actual meteorological conditions wind,m Equivalent (i.e., E) m =E wind,m The error caused by the correction process of prediction results under numerical weather prediction (first target error) E r Error E caused by the correction process for forecast results under actual weather conditions wind,r Equivalent (i.e., E) r =E wind,rIn this case, the results of error analysis are not reasonable enough.
[0116] A specific embodiment of the present invention is based on data from the Baiyunxian Wind Farm in Chenzhou, Hunan Province (113.5576°E, 25.3931°N) from August to December 2021, with a sampling interval of 15 minutes and a total of 96 samples taken per day. Table 1 shows the error decoupling results and error percentages in this specific embodiment.
[0117]
[0118] Table 1
[0119] This invention provides a processor configured to execute the wind power prediction error decoupling analysis method of any of the above embodiments.
[0120] Specifically, the processor can be configured as follows:
[0121] The data required for wind power prediction error analysis is obtained from the historical operation data of wind farms. The data includes measured weather data, numerical weather forecast data, installed capacity, real-time operating capacity, planned operating capacity, wind power prediction data, and actual output power of wind farms.
[0122] Determine the first target error caused by the prediction result correction process;
[0123] The wind-to-power conversion model of the wind farm is fitted based on the required data to establish a wind power conversion fitting model;
[0124] The measured weather data of the wind farm is input into the wind power conversion fitting model to obtain the model output under real meteorological conditions;
[0125] The predicted wind power under real weather conditions is determined based on the installed capacity, planned operating capacity, and model output under real weather conditions.
[0126] The correction error of the forecast result under actual weather conditions is determined based on the real-time operating capacity, the planned operating capacity, and the wind power forecast power under actual weather conditions.
[0127] The wind-to-power conversion model error under real weather conditions is determined based on the predicted wind power under real weather conditions, the actual output power of the wind farm, and the correction error of the prediction results under real weather conditions.
[0128] Obtain the model output under numerical weather forecast conditions corresponding to the model output under real meteorological conditions;
[0129] The second objective error caused by the wind-to-power conversion model is determined based on the model output under real meteorological conditions, the model output under numerical weather forecast, and the wind-to-power conversion model error under real meteorological conditions.
[0130] The third objective error caused by the numerical weather prediction process is determined based on wind power forecast data, actual wind farm output power, the first objective error, and the second objective error; and
[0131] The proportion of each objective error is determined based on the first objective error, the second objective error, and the third objective error.
[0132] In this embodiment of the invention, the processor can also be configured to:
[0133] Based on the percentage of error for each target, the main factors causing errors in each stage of the wind power prediction process are analyzed.
[0134] In this embodiment of the invention, determining the first target error caused by the prediction result correction step includes:
[0135] The equivalent wind power forecast value is determined based on wind power forecast data, real-time operating capacity and planned operating capacity.
[0136] The first target error is obtained by subtracting the equivalent wind power prediction value from the wind power prediction data.
[0137] In this embodiment of the invention, the wind-to-power conversion model of the wind farm is fitted according to the required data to establish a wind power conversion fitting model, including:
[0138] The wind-to-power conversion model output is determined based on installed capacity, planned operating capacity, and wind power forecast data.
[0139] Numerical weather forecast data is used as input to the wind-to-power conversion model, and the data, together with the output of the wind-to-power conversion model, forms the training set and the test set.
[0140] XGBoost was used to train the training set to fit the wind-to-electricity conversion model;
[0141] After training is completed, the test set is used to determine whether the fitted wind-to-power conversion model meets the accuracy requirements;
[0142] Modify the initial parameters of XGBoost and repeat the training on the training set until the accuracy requirement is met;
[0143] The wind-to-power conversion model that meets the accuracy requirements is used as the wind power conversion fitting model.
[0144] In this embodiment of the invention, the wind-to-electricity conversion model includes one of the following:
[0145] Wind-to-power conversion models based on BP neural networks, wind-to-power conversion models based on LSTM recurrent neural networks, and wind-to-power conversion models based on CNN-LSTM hybrid neural networks.
[0146] In this embodiment of the invention, determining the predicted wind power under actual weather conditions based on the installed capacity, planned operating capacity, and model output under actual weather conditions includes:
[0147] The predicted wind power under actual meteorological conditions is calculated using the following formula:
[0148]
[0149] Among them, P wind It is the predicted wind power under actual meteorological conditions, C′ o C is the planned startup capacity, F is the installed capacity, and f is the planned startup capacity. wind,m It is the model output under real weather conditions.
[0150] In this embodiment of the invention, the correction error of the prediction result under actual weather conditions is determined based on the real-time operating capacity, the planned operating capacity, and the predicted wind power under actual weather conditions, including:
[0151] The correction error for the forecast results under actual meteorological conditions is calculated using the following formula:
[0152]
[0153] Among them, E wind,r It is the correction error of the forecast result under actual meteorological conditions, C o This is the real-time boot capacity, C′ o This is the planned startup capacity, P wind It is the predicted wind power output under actual weather conditions.
[0154] In this embodiment of the invention, the wind-to-power conversion model error under real weather conditions is determined based on the predicted wind power under real weather conditions, the actual output power of the wind farm, and the correction error of the prediction results under real weather conditions, including:
[0155] The error of the wind-to-electricity conversion model under real meteorological conditions is calculated using the following formula:
[0156] P wind =P m +E wind,m +E wind,r
[0157] Among them, P wind It is the predicted wind power output under actual weather conditions, P m E is the actual output power of the wind farm.wind,m It is the error of the wind-to-power conversion model under real meteorological conditions, and E wind,r It is the correction error of the prediction results under actual meteorological conditions.
[0158] In this embodiment of the invention, the second target error caused by the wind-to-power conversion model is determined based on the model output under real meteorological conditions, the model output under numerical weather forecasting, and the wind-to-power conversion model error under real meteorological conditions, including:
[0159] The second objective error is calculated using the following formula:
[0160] E m +f m =E wind,m +f wind,m
[0161] Among them, E m It is the second objective error, f m This is the model output from numerical weather prediction, E wind,m It is the error of the wind-to-power conversion model under real meteorological conditions, and f wind,m It is the model output under real weather conditions.
[0162] In this embodiment of the invention, the third target error caused by the numerical weather prediction process is determined based on wind power prediction data, actual wind farm output power, a first target error, and a second target error, including:
[0163] The third objective error is calculated using the following formula:
[0164] P p =P m +E n +E m +E r
[0165] Among them, P p This is wind power forecast data, E r It is the first target error, E m It is the second objective error, and E n This is the third objective error.
[0166] This invention provides a machine-readable storage medium storing instructions that, when executed by a processor, cause the processor to implement the wind power prediction error decoupling analysis method of any of the above embodiments.
[0167] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 2As shown. The computer device includes a processor A01, a network interface A02, a memory (not shown), and a database (not shown) connected via a system bus. The processor A01 provides computing and control capabilities. The memory includes internal memory A03 and a non-volatile storage medium A04. The non-volatile storage medium A04 stores an operating system B01, a computer program B02, and a database (not shown). The internal memory A03 provides an environment for the operation of the operating system B01 and the computer program B02 stored in the non-volatile storage medium A04. The database stores data related to repayment change parameters and repayment information. The network interface A02 communicates with external terminals via a network connection. When executed by the processor A01, the computer program B02 implements a method for adjusting a repayment plan.
[0168] Those skilled in the art will understand that Figure 2 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0169] This application provides a device including a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it implements the wind power prediction error decoupling analysis method of any of the above embodiments.
[0170] In wind power forecasting, wind-to-power conversion models are often trained using numerical weather prediction (NMR) data, and the correction of prediction results depends on the output of these models. Therefore, directly comparing the prediction results under NMR with those under actual weather conditions, while allowing for the resolution of errors caused by NMR, fails to accurately decouple the wind-to-power conversion model from the correction of prediction results. The technical solution provided in this invention, based on prediction results under both actual and NMR conditions, and comprehensively considering the impact of changes in weather conditions on each stage, can accurately decouple the error-causing stages. Furthermore, it calculates the error proportion of each stage, providing theoretical guidance for improving wind farm forecasting methods.
[0171] Based on the wind power prediction process, the wind farm power prediction process can be divided into three key stages: numerical weather prediction, wind-to-power conversion model, and prediction result correction. When meteorological conditions change, it will affect not only numerical weather prediction but also the wind-to-power conversion model and prediction result correction. However, existing wind power prediction methods do not consider that the errors in each stage of wind power prediction will change as meteorological data and other conditions change.
[0172] The differences between the wind power prediction method of this invention and the prior art include:
[0173] (1) For numerical weather prediction errors, existing methods directly solve the predicted power based on different meteorological conditions; however, the embodiments of the present invention further consider the impact of the wind-to-power conversion model and the prediction result correction on numerical weather prediction errors during the solution process.
[0174] (2) For the wind-to-power conversion model link error, the existing method is to solve it based on the actual power, predicted power, numerical weather prediction link error and prediction result correction link error; while the embodiment of the present invention is to solve it based on the equation (Formula 8) related to the wind-to-power conversion model error.
[0175] (3) In order to improve the error decoupling equation, the embodiments of the present invention have added a calculation formula for the error of the prediction result correction link under real meteorological conditions.
[0176] Therefore, the technical solution provided by the embodiments of the present invention fully considers this error variation characteristic. By supplementing the estimation of the error variation of the wind-to-power conversion model link and the prediction result correction link under actual meteorological conditions, a more comprehensive and reasonable decoupling analysis of the error of each link of wind power prediction is carried out. The obtained error analysis results can provide more reliable guidance for the improvement and optimization of wind power prediction technology.
[0177] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0178] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0179] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0180] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0181] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0182] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.
[0183] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, 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 technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.
[0184] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.
[0185] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.
Claims
1. A method for decoupling analysis of wind power prediction errors, characterized in that, The wind power prediction process includes numerical weather prediction, wind-to-power conversion modeling, and prediction result correction. The wind power prediction error decoupling analysis method includes: The data required for wind power prediction error analysis is obtained from the historical operation data of wind farms. The data includes the wind farm's measured weather data, numerical weather forecast data, installed capacity, real-time operating capacity, planned operating capacity, wind power prediction data, and actual output power of the wind farm. Determine the first target error caused by the prediction result correction step; The wind-to-power conversion model of the wind farm is fitted based on the required data to establish a wind power conversion fitting model; The measured weather data of the wind farm is input into the wind power conversion fitting model to obtain the model output under real meteorological conditions; The predicted wind power under real weather conditions is determined based on the installed capacity, the planned operating capacity, and the model output under real weather conditions. The correction error of the prediction result under actual weather conditions is determined based on the real-time operating capacity, the planned operating capacity, and the wind power prediction power under actual weather conditions. The wind-to-power conversion model error under real weather conditions is determined based on the predicted wind power under real weather conditions, the actual output power of the wind farm, and the correction error of the prediction results under real weather conditions. Obtain the model output under numerical weather forecast conditions corresponding to the model output under real meteorological conditions; The second objective error caused by the wind-to-power conversion model is determined based on the model output under real meteorological conditions, the model output under numerical weather forecast, and the wind-to-power conversion model error under real meteorological conditions. Based on the wind power prediction data, the actual output power of the wind farm, the first target error, and the second target error, a third target error caused by the numerical weather prediction process is determined; and The proportion of each target error is determined based on the first target error, the second target error, and the third target error; The determination of the second objective error caused by the wind-to-power conversion model stage based on the model output under real meteorological conditions, the model output under numerical weather forecasting, and the wind-to-power conversion model error under real meteorological conditions includes: The second target error is calculated using the following formula: in, It is the second target error. This is the model output under the numerical weather prediction. It is the error of the wind-to-power conversion model under real meteorological conditions, and It is the model output under real weather conditions; The model output under numerical weather prediction can be calculated using the following formula: in, f m This is the output of the wind-to-power conversion model under numerical weather prediction. For installed capacity, For planned startup capacity, This is wind power forecast data.
2. The wind power prediction error decoupling analysis method according to claim 1, characterized in that, Also includes: Based on the percentage of error for each target, the main factors causing errors in each stage of the wind power prediction process are analyzed.
3. The wind power prediction error decoupling analysis method according to claim 1, characterized in that, The determination of the first target error caused by the prediction result correction step includes: The equivalent wind power prediction value is determined based on the wind power prediction data, the real-time operating capacity, and the planned operating capacity. The first target error is obtained by subtracting the equivalent wind power prediction value from the wind power prediction data.
4. The wind power prediction error decoupling analysis method according to claim 1, characterized in that, The process of fitting the wind-to-power conversion model of the wind farm based on the required data to establish a wind power conversion fitting model includes: The wind-to-power conversion model output is determined based on the installed capacity, the planned operating capacity, and the wind power prediction data. The numerical weather forecast data is used as input to the wind-to-power conversion model, and the data, together with the output of the wind-to-power conversion model, form a training set and a test set. The training set was trained using XGBoost to fit the wind-to-electricity conversion model; After training is completed, the test set is used to determine whether the fitted wind-to-power conversion model meets the accuracy requirements. Modify the initial parameters of XGBoost and repeat the training on the training set until the accuracy requirement is met; The fitted wind-to-power conversion model that meets the accuracy requirements is used as the wind power conversion fitting model.
5. The wind power prediction error decoupling analysis method according to claim 4, characterized in that, The wind-to-power conversion model includes one of the following: Wind-to-power conversion models based on BP neural networks, wind-to-power conversion models based on LSTM recurrent neural networks, and wind-to-power conversion models based on CNN-LSTM hybrid neural networks.
6. The wind power prediction error decoupling analysis method according to claim 1, characterized in that, Determining the predicted wind power under actual weather conditions based on the installed capacity, the planned operating capacity, and the model output under actual weather conditions includes: The predicted wind power under actual meteorological conditions is calculated using the following formula: in, It is the predicted wind power output under actual weather conditions. The planned startup capacity, The installed capacity is as stated. It is the model output under real weather conditions.
7. The wind power prediction error decoupling analysis method according to claim 6, characterized in that, The step of determining the correction error of the prediction result under actual weather conditions based on the real-time operating capacity, the planned operating capacity, and the predicted wind power under actual weather conditions includes: The correction error for the forecast results under actual meteorological conditions is calculated using the following formula: in, It is a correction error for the forecast results under actual meteorological conditions. This refers to the real-time power-on capacity. This is the planned startup capacity. It is the predicted wind power output under actual weather conditions.
8. The wind power prediction error decoupling analysis method according to claim 7, characterized in that, The determination of the wind-to-power conversion model error under real weather conditions based on the predicted wind power under real weather conditions, the actual output power of the wind farm, and the correction error of the prediction results under real weather conditions includes: The error of the wind-to-electricity conversion model under real meteorological conditions is calculated using the following formula: in, It is the predicted wind power output under actual weather conditions. This is the actual output power of the wind farm. It is the error of the wind-to-power conversion model under real meteorological conditions, and It is the correction error of the prediction results under actual meteorological conditions.
9. The wind power prediction error decoupling analysis method according to claim 1, characterized in that, The step of determining the third target error caused by the numerical weather prediction process based on the wind power prediction data, the actual output power of the wind farm, the first target error, and the second target error includes: The third objective error is calculated using the following formula: in, This refers to the wind power prediction data. It is the error of the first target. It is the second target error, and This is the third target error. This is the actual output power of the wind farm.
10. A processor, characterized in that, It is configured to perform the wind power prediction error decoupling analysis method according to any one of claims 1 to 9.
11. A machine-readable storage medium, characterized in that, The machine-readable storage medium stores instructions that, when executed by a processor, cause the processor to implement the wind power prediction error decoupling analysis method according to any one of claims 1 to 9.