Generation power and electric quantity prediction method of wind power plant and electronic equipment

Abstract

The embodiment of the invention provides a generation power and electric quantity prediction method of a wind power plant and electronic equipment. The method comprises the following steps: acquiring target data, wherein the target data comprises short-term data, medium-term data and long-term data; respectively inputting the short-term data, the medium-term data and the long-term data into a first sub-model, a second sub-model and a third sub-model to obtain a first prediction result, a second prediction result and a third prediction result; and according to the first prediction result, the second prediction result and the third prediction result, determining the generation power and the electric quantity of the wind power plant. According to the embodiment of the invention, short-term data, medium-term data and long-term data are analyzed at the same time, change modes on different time scales are captured, and the generated power of the wind power plant and the comprehensiveness of electric quantity prediction are improved; by considering the data change of multiple time scales, the machine learning model can keep stable accuracy in different time ranges, and adapts to short-term fluctuation and long-term trend.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of electric power technology, and in particular to a method and electronic equipment for predicting the generated power and amount of electricity of a wind farm. Background Art

[0002] As a clean and renewable energy source, wind power has developed rapidly in recent years. However, wind energy is intermittent and volatile, which brings challenges to the power and electricity forecasting of wind farms.

[0003] At present, the power and electricity prediction methods of wind farms mainly include physical model method, statistical model method and artificial intelligence model method. The physical model method is based on wind energy resources and wind turbine characteristics, but is limited by the complexity of the model and the accuracy of meteorological data, and the prediction accuracy is limited; the statistical model method is mainly based on time series analysis, but lacks comprehensive consideration of internal factors and external environmental factors of wind farms, and the prediction accuracy is low; artificial intelligence model methods, such as neural networks and support vector machines, although they have good nonlinear fitting capabilities, require a large amount of training data and have poor model interpretability.

[0004] In view of this, how to accurately predict the power and electricity of wind farms with good interpretability has become a technical problem that needs to be solved urgently. Summary of the invention

[0005] In a first aspect, an embodiment of the present invention provides a method for predicting power generation of a wind farm, comprising:

[0006] Acquire target data, the target data including: short-term data, medium-term data and long-term data, the short-term data, medium-term data and long-term data are wind turbine operation data, meteorological forecast data and external environment data of the wind farm within a first preset time period, a second preset time period and a third preset time period respectively, the third preset time period is longer than the second preset time period, and the second preset time period is longer than the first preset time period;

[0007] Inputting the short-term data, the medium-term data and the long-term data into the first sub-model, the second sub-model and the third sub-model respectively to obtain a first prediction result, a second prediction result and a third prediction result;

[0008] The generated power and amount of electricity of the wind farm are determined according to the first prediction result, the second prediction result and the third prediction result.

[0009] In a possible implementation, inputting the short-term data, the medium-term data, and the long-term data into the first sub-model, the second sub-model, and the third sub-model, respectively, to obtain the first prediction result, the second prediction result, and the third prediction result includes:

[0010] Performing first processing on the short-term data, the medium-term data and the long-term data respectively, wherein the first processing includes cleaning, denoising and / or normalization processing;

[0011] The short-term data, the medium-term data and the long-term data after the first processing are respectively input into the first sub-model, the second sub-model and the third sub-model to obtain a first prediction result, a second prediction result and a third prediction result.

[0012] In a possible implementation, inputting the short-term data, the medium-term data, and the long-term data after the first processing into the first sub-model, the second sub-model, and the third sub-model, respectively, to obtain a first prediction result, a second prediction result, and a third prediction result includes:

[0013] Input the short-term data, the mid-term data and the long-term data after the first processing into a support vector machine model respectively, and obtain a first result, a second result and a third result respectively, the first result includes a first sub-result and a second sub-result, the second result includes a third sub-result and a fourth sub-result, the third result includes a fifth sub-result and a sixth sub-result, the first sub-result represents that the short-term data after the first processing under the result is valid short-term data, the second sub-result represents that the short-term data after the first processing under the result is invalid short-term data, the third sub-result represents that the mid-term data after the first processing under the result is valid mid-term data, the fourth sub-result represents that the mid-term data after the first processing under the result is invalid mid-term data, the fifth sub-result represents that the long-term data after the first processing under the result is valid long-term data, and the sixth sub-result represents that the long-term data after the first processing under the result is invalid long-term data;

[0014] The effective short-term data, the effective medium-term data and the effective long-term data are respectively input into the first sub-model, the second sub-model and the third sub-model to obtain a first prediction result, a second prediction result and a third prediction result.

[0015] In a possible implementation, inputting the valid short-term data, the valid mid-term data, and the valid long-term data into the first sub-model, the second sub-model, and the third sub-model, respectively, to obtain the first prediction result, the second prediction result, and the third prediction result includes:

[0016] A first feature, a second feature, a third feature, a fourth feature and a fifth feature are extracted from the effective short-term data, the effective medium-term data and the effective long-term data respectively, wherein the first feature includes: wind turbine model, impeller diameter, rated power, operating status and fault information, the second feature includes: wind speed, wind direction, temperature, air pressure, humidity and rainfall, the third feature includes: power load feature, the power load feature includes: power load demand and load fluctuation, the fourth feature includes: date, time point, season and day of the week, and the fifth feature includes: geographical location of the wind farm and surrounding topography;

[0017] The first feature, second feature, third feature, fourth feature and fifth feature extracted from the valid short-term data, the valid medium-term data and the valid long-term data respectively are input into the first sub-model, the second sub-model and the third sub-model respectively to obtain a first prediction result, a second prediction result and a third prediction result.

[0018] In a possible implementation, the first feature, the second feature, the third feature, the fourth feature, and the fifth feature extracted from the effective short-term data, the effective mid-term data, and the effective long-term data are respectively input into the first sub-model, the second sub-model, and the third sub-model to obtain the first prediction result, the second prediction result, and the third prediction result, including:

[0019] The weights of the first feature, the second feature, the third feature, the fourth feature and the fifth feature extracted from the valid short-term data, the valid mid-term data and the valid long-term data are adjusted respectively to obtain the sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature from the valid short-term data, the valid mid-term data and the valid long-term data;

[0020] The sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature of the valid short-term data, the valid medium-term data and the valid long-term data are respectively input into the first sub-model, the second sub-model and the third sub-model to obtain a first prediction result, a second prediction result and a third prediction result.

[0021] In a possible implementation, the sixth feature, the seventh feature, the eighth feature, the ninth feature, and the tenth feature of the effective short-term data, the effective mid-term data, and the effective long-term data are respectively input into the first sub-model, the second sub-model, and the third sub-model to obtain the first prediction result, the second prediction result, and the third prediction result, including:

[0022] Performing a second processing on the first data, the second data and the third data respectively, wherein the second processing includes feature selection, feature dimension reduction and / or feature combination, the first data is the sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature for the effective short-term data, the second data is the sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature for the effective medium-term data, and the third data is the sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature for the effective long-term data;

[0023] The first data, the second data and the third data after the second processing are respectively input into the first sub-model, the second sub-model and the third sub-model to obtain a first prediction result, a second prediction result and a third prediction result.

[0024] In a possible implementation, the first sub-model is a random forest model;

[0025] The step of inputting the first data, the second data and the third data after the second processing into the first sub-model, the second sub-model and the third sub-model respectively to obtain a first prediction result, a second prediction result and a third prediction result comprises:

[0026] Inputting the first data after the second processing into at least one decision tree of the random forest model to obtain a prediction result of each decision tree;

[0027] Vote on the prediction results of all decision trees to get the first prediction result.

[0028] In a possible implementation, the second sub-model is a gradient boosting tree;

[0029] The step of inputting the first data, the second data and the third data after the second processing into the first sub-model, the second sub-model and the third sub-model respectively to obtain a first prediction result, a second prediction result and a third prediction result comprises:

[0030] Inputting the second processed second data into each weak learner of the gradient boosting tree to obtain a prediction result of each weak learner;

[0031] The prediction results of all weak learners are weighted and summed to obtain the second prediction result.

[0032] In a possible implementation, determining the power generation and amount of electricity of the wind farm according to the first prediction result, the second prediction result, and the third prediction result includes:

[0033] The first prediction result, the second prediction result and the third prediction result are multiplied by a first weight, a second weight and a third weight respectively, and the first weight is greater than the second weight and the third weight;

[0034] Taking a weighted average of the first prediction result, the second prediction result and the third prediction result respectively multiplied by the first weight, the second weight and the third weight, to obtain the short-term wind farm power generation and electricity; or

[0035] The first prediction result, the second prediction result and the third prediction result are multiplied by a fourth weight, a fifth weight and a sixth weight respectively, and the fifth weight is greater than the fourth weight and the sixth weight;

[0036] Taking a weighted average of the first prediction result, the second prediction result and the third prediction result respectively multiplied by a fourth weight, a fifth weight and a sixth weight, to obtain the power generation and the amount of electricity of the wind farm in the medium term; or

[0037] The first prediction result, the second prediction result and the third prediction result are multiplied by a seventh weight, an eighth weight and a ninth weight respectively, and the ninth weight is greater than the seventh weight and the eighth weight;

[0038] The first prediction result, the second prediction result and the third prediction result multiplied by the seventh weight, the eighth weight and the ninth weight respectively are weighted averaged to obtain the long-term wind farm power generation and electricity.

[0039] In a second aspect, an embodiment of the present invention provides an electronic device, including:

[0040] A memory and a processor, wherein the processor and the memory communicate with each other via a bus; the memory stores program instructions executable by the processor, and the processor calls the program instructions to execute the method described in the first aspect and each step in various possible implementations.

[0041] In a third aspect, an embodiment of the present invention provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the first aspect and each step in various possible implementations.

[0042] In a fourth aspect, an embodiment of the present invention provides a computer program product comprising instructions, and when the computer program product is run on a computer, the method described in the first aspect is executed by the computer.

[0043] The embodiments of the present invention disclose the following technical effects:

[0044] 1) The embodiment of the present invention simultaneously analyzes short-term data, medium-term data and long-term data, captures the change patterns of data at different time scales, and improves the comprehensiveness of wind farm power generation and electricity forecasting; by considering the changes in data at multiple time scales, the machine learning model can maintain stable accuracy in different time ranges and adapt to short-term fluctuations and long-term trends;

[0045] 2) The embodiment of the present invention integrates multi-source data, namely, wind turbine operation data of the wind farm, meteorological forecast data and external environment data of the wind farm, to provide more comprehensive input information for the machine learning model, thereby improving the prediction accuracy of the wind farm's power generation and electricity.

[0046] 3) Collect wind turbine operation data from all wind turbines in the wind farm, conduct unified analysis and modeling, rather than predicting each wind turbine separately. The centralized prediction method makes full use of the integrity of the data, can capture the complex relationships between wind turbines within the wind farm, and improve the accuracy and robustness of the prediction of wind farm power and electricity;

[0047] 4) Dynamically adjust the weights of features (first feature, second feature, third feature, fourth feature and fifth feature) according to the changes in wind turbine operation data, meteorological forecast data and external environment data of the wind farm, so that the machine learning model can adapt to changes in different time and environment. Traditional feature weights are usually fixed and difficult to cope with environmental changes. Dynamically adjusting feature weights allows the machine learning model to maintain high prediction accuracy under changing conditions;

[0048] 5) The embodiment of the present invention integrates the prediction results of multiple models (i.e., random forest model, gradient boosting tree and neural network model), that is, weighted average or voting is performed on the first prediction result, the second prediction result and the third prediction result to obtain the short-term wind farm power and electricity, the medium-term wind farm power and electricity and the long-term wind farm power and electricity, rather than using the prediction results of a single model, so that the prediction results are more accurate and the prediction accuracy of the wind farm power and electricity is improved.

[0049] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0051] Figure 1A system architecture diagram applicable to the embodiment of the present invention;

[0052] Figure 2 A schematic diagram of a flow chart of a method for predicting power and electricity generation of a wind farm provided by an embodiment of the present invention;

[0053] Figure 3 A schematic block diagram of a device for predicting power generation and electricity quantity of a wind farm provided by an embodiment of the present invention;

[0054] Figure 4 A schematic block diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0056] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The singular forms "a", "said" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings.

[0057] It should be understood that the term "and / or" used in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0058] The word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrases "if determining" or "if monitoring (stated condition or event)" may be interpreted as "when determining" or "in response to determining" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)", depending on the context.

[0059] At present, the physical model method can be used to predict the power generation and electricity of wind farms based on wind energy resources and wind turbine characteristics. However, due to the complexity of the model and the accuracy of meteorological data, the prediction accuracy is limited. The statistical model method can also be used to predict the power generation and electricity of wind farms based on time series analysis. However, there is a lack of comprehensive consideration of internal factors and external environmental factors of wind farms, and the prediction accuracy is low. Artificial intelligence model methods, such as neural networks and support vector machines, can also be used to predict the power generation and electricity of wind farms. Although they have good nonlinear fitting capabilities, they require a large amount of training data and the model has poor interpretability.

[0060] In view of this, the embodiment of the present invention provides a new idea. In order to facilitate the understanding of the present application, the system architecture based on the embodiment of the present invention is first described. Figure 1 An exemplary system architecture to which the present invention can be applied is shown. Figure 1 As shown in , the system architecture may include a terminal device and a wind farm power generation and electricity quantity prediction device located at a server end.

[0061] The terminal devices may include, but are not limited to, smart mobile terminals, smart home devices, wearable devices, PCs (Personal Computers), etc. Smart mobile terminals may include mobile phones, tablet computers, laptop computers, PDAs (Personal Digital Assistants), Internet cars, etc. Smart home devices may include smart TVs, smart refrigerators, etc. Wearable devices may include smart watches, smart glasses, virtual reality devices, augmented reality devices, mixed reality devices (i.e., devices that can support virtual reality and augmented reality), etc.

[0062] The device for predicting the power generation and amount of electricity of a wind farm may use the method for predicting the power generation and amount of electricity of a wind farm provided in the embodiment of the present invention to predict the power generation and amount of electricity of the wind farm.

[0063] The above-mentioned wind farm power generation and electricity prediction device can be set on a single server, or on a server group consisting of multiple servers, or on a cloud server. A cloud server, also known as a cloud computing server or cloud host, is a host product in the cloud computing service system to solve the defects of difficult management and weak service scalability in traditional physical hosts and virtual private servers (VPS). Figure 1 In addition to the illustrated architecture, the wind farm power generation and electricity quantity prediction device may also be arranged on a computer terminal with strong computing capability.

[0064] Figure 2A schematic diagram of a flow chart of a method for predicting power and electricity of a wind farm provided by an embodiment of the present invention. The method can be performed by Figure 1 The power generation and electricity quantity prediction device of the wind farm in the system shown is executed. Figure 2 As shown in , the method may include the following steps:

[0065] Step 201: Acquire target data, which includes short-term data, medium-term data and long-term data. The short-term data, medium-term data and long-term data are wind turbine operation data, meteorological forecast data and external environment data of the wind farm within a first preset time period, a second preset time period and a third preset time period respectively. The third preset time period is greater than the second preset time period, and the second preset time period is greater than the first preset time period.

[0066] Step 202: input short-term data, medium-term data and long-term data into the first sub-model, the second sub-model and the third sub-model respectively to obtain a first prediction result, a second prediction result and a third prediction result.

[0067] Step 203: Determine the power generation and amount of electricity of the wind farm according to the first prediction result, the second prediction result and the third prediction result.

[0068] It can be seen from the above process that the embodiment of the present invention captures the change patterns on different time scales by simultaneously analyzing short-term data, medium-term data and long-term data, thereby improving the comprehensiveness of the prediction of the power generation and electricity of the wind farm; by considering the data changes on multiple time scales, the machine learning model can maintain stable accuracy in different time ranges and adapt to short-term fluctuations and long-term trends; the embodiment of the present invention integrates multi-source data, namely the wind turbine operation data of the wind farm, meteorological forecast data and the external environment data of the wind farm, to provide more comprehensive input information for the machine learning model, thereby improving the prediction accuracy of the power generation and electricity of the wind farm; the wind turbine operation data of all wind turbines in the wind farm are collected for unified analysis and modeling, rather than predicting each individual wind turbine separately. The centralized prediction method makes full use of the integrity of the data, can capture the complex relationships between the wind turbines within the wind farm, and improves the prediction accuracy and robustness of the power generation and electricity of the wind farm. The embodiment of the present invention collects wind turbine operating data from all wind turbines in the wind farm and performs unified analysis and modeling, rather than predicting each individual wind turbine separately. The centralized prediction method makes full use of the integrity of the data, can capture the complex relationships between wind turbines within the wind farm, and improves the accuracy and robustness of the prediction of the wind farm's generated power and electricity.

[0069] The following is a detailed description of each step in the above process and the effects that can be further produced in conjunction with the embodiments of the present invention. It should be noted that the limitations such as "first" and "second" involved in the embodiments of the present invention do not have limitations in terms of size, order, and quantity, and are only used to distinguish them in name, for example, "first prediction result" and "second prediction result" are used to distinguish two different prediction results.

[0070] First, in combination with an embodiment of the present invention, the above step 201, i.e., "obtaining target data, the target data includes: short-term data, medium-term data and long-term data, the short-term data, medium-term data and long-term data are respectively the wind turbine operation data, meteorological forecast data and external environment data of the wind farm within a first preset time period, a second preset time period and a third preset time period, the third preset time period is greater than the second preset time period, and the second preset time period is greater than the first preset time period" is described in detail.

[0071] In an embodiment of the present invention, multiple time scale data are obtained as target data, including short-term data, medium-term data and long-term data. Among them, the short-term data, medium-term data and long-term data are the wind turbine operation data, meteorological forecast data and external environment data of the wind farm in the first preset time period, the second preset time period and the third preset time period respectively, which overcomes the problem of low accuracy of using a single data to predict the power generation of the wind farm. Through multi-dimensional data fusion, the machine learning model is enhanced to understand and predict the operation status of the wind farm. Among them, the third preset time period is greater than the second preset time period, and the second preset time period is greater than the first preset time period. Exemplarily, the first preset time period is 1 hour, focusing on the time changes within the day, such as the specific time point of the day, day / night, etc. Hourly periodic features are introduced to capture the periodic changes in power generation within a day. These features help capture the impact of changes in factors such as wind conditions and light on power generation in the short term; the second preset time period is 1 day, and the weekly periodicity and monthly periodicity features reflect periodic patterns such as weekdays / weekends, the beginning of the month / the end of the month. Consider that holiday factors may affect electricity demand. These features help capture the impact of factors such as medium-term climate change and electricity demand patterns; the third preset time period is 1 month, focusing on changes in longer time spans, such as years, quarters, and days in a year. Monthly periodic features are introduced to reflect large-scale patterns of climate change throughout the year. These features help capture the impact of factors such as long-term climate change, equipment aging, and policy changes. In addition, the embodiment of the present invention collects wind turbine operating data for all wind turbines in a wind farm, conducts unified analysis and modeling, rather than predicting individual wind turbines separately. The centralized prediction method makes full use of the integrity of the data, can capture the complex relationships within the wind farm, and improves the accuracy and robustness of the prediction of the wind farm's generated power and electricity.

[0072] In the embodiment of the present invention, high-quality sample data is generated by using technologies such as generative adversarial networks to enrich the sample data set and improve the generalization ability of the model. Through intelligent data enhancement, the problem of insufficient sample data is solved, especially in the face of extreme situations, the robustness of the model can be significantly improved.

[0073] In the embodiment of the present invention, the digital twin technology can also be used to build a virtual model of the wind farm, and the sample data used for model training can be enhanced with simulation data, and simulation tests can be performed under extreme conditions. The digital twin technology can provide high-fidelity simulation data, helping the model to maintain high prediction accuracy even under extreme conditions and foresee potential risks.

[0074] The above step 202, i.e., "inputting short-term data, medium-term data and long-term data into the first sub-model, the second sub-model and the third sub-model respectively to obtain a first prediction result, a second prediction result and a third prediction result" is described in detail below in conjunction with an embodiment of the present invention.

[0075] As a possible implementation method, in order to obtain high-quality short-term data, medium-term data and long-term data, the short-term data, medium-term data and long-term data are respectively subjected to a first processing, the first processing including cleaning, denoising and / or normalization processing; the short-term data, medium-term data and long-term data after the first processing are respectively input into the first sub-model, the second sub-model and the third sub-model to obtain the first prediction result, the second prediction result and the third prediction result. Through the first processing, data noise is reduced, the stability and robustness of the machine learning model are improved, and at the same time, early warning is provided for operation to reduce potential operational risks.

[0076] As a possible implementation method, since there are invalid data in the short-term data, the medium-term data and the long-term data, in order to eliminate the invalid data, the short-term data, the medium-term data and the long-term data after the first processing are respectively input into the first sub-model, the second sub-model and the third sub-model, and the first prediction result, the second prediction result and the third prediction result are obtained. Specifically, the short-term data, the medium-term data and the long-term data after the first processing are respectively input into the support vector machine model to obtain the first result, the second result and the third result, respectively. The first result includes the first sub-result and the second sub-result, the second result includes the third sub-result and the fourth sub-result, the third result includes the fifth sub-result and the sixth sub-result, and the first sub-result represents the result under The short-term data after the first process is valid short-term data, the second sub-result represents that the short-term data after the first process under the result is invalid short-term data, the third sub-result represents that the medium-term data after the first process under the result is valid medium-term data, the fourth sub-result represents that the medium-term data after the first process under the result is invalid medium-term data, the fifth sub-result represents that the long-term data after the first process under the result is valid long-term data, and the sixth sub-result represents that the long-term data after the first process under the result is invalid long-term data; the valid short-term data, the valid medium-term data and the valid long-term data are respectively input into the first sub-model, the second sub-model and the third sub-model to obtain the first prediction result, the second prediction result and the third prediction result.

[0077] As a possible implementation, relevant features are extracted for effective short-term data, effective medium-term data, and effective long-term data. Specifically, the first feature, the second feature, the third feature, the fourth feature, and the fifth feature are extracted for effective short-term data, effective medium-term data, and effective long-term data, respectively. The first feature includes: wind turbine model, impeller diameter, rated power, operating status, and fault information. The second feature includes: wind speed, wind direction, temperature, air pressure, humidity, and rainfall. The third feature includes: power load features, and the power load features include: power load demand and load fluctuation. The fourth feature includes: date, time point, season, and day of the week. The fifth feature includes: geographical location of the wind farm and surrounding topography. The effective short-term data, effective medium-term data, and effective long-term data are input into the first sub-model, the second sub-model, and the third sub-model, respectively, and the first prediction result, the second prediction result, and the third prediction result are obtained. Specifically, the first feature, the second feature, the third feature, the fourth feature, and the fifth feature extracted for the effective short-term data, the effective medium-term data, and the effective long-term data are input into the first sub-model, the second sub-model, and the third sub-model, respectively, to obtain the first prediction result, the second prediction result, and the third prediction result.

[0078] It can be understood that the first feature, the second feature, the third feature, the fourth feature and the fifth feature extracted from the effective short-term data, the effective medium-term data and the effective long-term data are respectively input into the first sub-model, the second sub-model and the third sub-model, and before obtaining the first prediction result, the second prediction result and the third prediction result, the weights of the first feature, the second feature, the third feature, the fourth feature and the fifth feature extracted from the effective short-term data, the effective medium-term data and the effective long-term data are respectively adjusted to obtain the sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature for the effective short-term data, the effective medium-term data and the effective long-term data; the sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature for the effective short-term data, the effective medium-term data and the effective long-term data are respectively input into the first sub-model, the second sub-model and the third sub-model to obtain the first prediction result, the second prediction result and the third prediction result. The embodiment of the present invention dynamically adjusts the weights of the features (the first feature, the second feature, the third feature, the fourth feature and the fifth feature) according to the changes in the wind turbine operation data, the meteorological forecast data and the external environment data of the wind farm, so that the machine learning model can adapt to the changes in different times and environments. Traditional feature weights are usually fixed and difficult to cope with environmental changes. Dynamically adjusting feature weights enables machine learning models to maintain high prediction accuracy under changing conditions.

[0079] In one example, the weights of the first feature, the second feature, the third feature, the fourth feature, and the fifth feature are dynamically adjusted according to the calculated feature importance. Exemplarily, a model-based method can be used: such as a random forest model, by calculating the average impurity reduction of each feature in all trees to evaluate its importance. For the Gradient Boosting model, the number of times each feature is used as a split point can be calculated to evaluate its importance. The SHAP value method can also be used: the contribution of the feature value can be assigned to each prediction. It not only gives the global feature importance, but also explains a single prediction. It can also be based on a statistical method: calculate the Pearson correlation coefficient or mutual information between each feature and the target variable (such as power generation). These methods can capture linear and nonlinear relationships. The importance of permutation features can also be used: the importance of a feature is evaluated by randomly disrupting the value of the feature and then observing the degree of decline in model performance. After calculating the feature importance, the soft maximum function can be used to convert the feature importance into a weight.

[0080] As a possible implementation, in order to improve the model training accuracy, the embodiment of the present invention performs feature engineering. That is to say, the sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature for the effective short-term data, the effective medium-term data and the effective long-term data are respectively input into the first sub-model, the second sub-model and the third sub-model, and the first prediction result, the second prediction result and the third prediction result can be obtained specifically as follows: the first data, the second data and the third data are respectively subjected to the second processing, the second processing includes feature selection, feature dimension reduction and / or feature combination, the first data is the sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature for the effective short-term data, the second data is the sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature for the effective medium-term data, and the third data is the sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature for the effective long-term data; the first data, the second data and the third data after the second processing are respectively input into the first sub-model, the second sub-model and the third sub-model, and the first prediction result, the second prediction result and the third prediction result are obtained. It is understandable that feature engineering enables machine learning models to process complex data more efficiently, screen out the most valuable features for prediction results, reduce redundant information, and improve the computational efficiency and prediction performance of machine learning models. It is also understandable that in an embodiment of the present invention, feature selection can be achieved by calculating the contribution of each feature to the model prediction results during the training process of the random forest model. The higher the feature importance, the greater the influence of the feature on the prediction results of the random forest model. By observing the feature importance, it can be understood which features are the key factors affecting the first prediction results. Exemplarily, it can be found that features such as wind speed and wind direction have a greater impact on the first prediction results. Many advanced machine learning models are often regarded as "black boxes" and difficult to explain. This method emphasizes interpretability, increases users' trust in the random forest model, and also provides a strong basis for adjusting operational strategies. It is also understandable that feature dimensionality reduction is performed by principal component analysis method. The target data (including short-term data, medium-term data, and long-term data) is projected into a new coordinate system, and the new coordinate axis corresponds to the main components of the data, retaining the components with the largest variance. It is also understandable that multiple features are combined into new features by cross combination. For example, wind speed and wind direction are combined into wind energy density, and wind speed and impeller diameter are multiplied to obtain the swept area of ​​the impeller. In addition, feature engineering, model selection, and hyperparameter optimization are automatically performed through the automated machine learning platform, reducing manual intervention and improving development efficiency. Automated machine learning reduces the complexity of machine learning model development, allowing non-professionals to participate in model development and significantly shortening the development cycle.

[0081] The above step 203, namely "determining the generated power and amount of electricity of the wind farm according to the first prediction result, the second prediction result and the third prediction result" is described in detail below in conjunction with an embodiment of the present invention.

[0082] As a possible implementation method, when the first sub-model, the second sub-model and the third sub-model are respectively neural network models, the first data after the second processing is input into the neural network model to obtain a first prediction result, i.e., the power generation and electricity of the short-term wind farm; the second data after the second processing is input into the neural network model to obtain a second prediction result, i.e., the power generation and electricity of the medium-term wind farm; the third data after the second processing is input into the neural network model to obtain a third prediction result, i.e., the power generation and electricity of the long-term wind farm.

[0083] It should be noted that the embodiment of the present invention quantifies the uncertainty of the prediction by providing a probability distribution of the prediction results (the power generation and electricity of the wind farm obtained), so as to help operators better assess risks. This method enhances the understanding of the prediction results, especially when facing uncertainty, helping decision makers to better assess potential risks and benefits. Exemplarily, the probability distribution is: the probability of a short-term power generation of 10,000 kilowatts is 75%, the probability of a short-term power generation of 12,000 kilowatts is 60%, and the probability of a short-term power generation of 15,000 kilowatts is 50%. The short-term power generation corresponding to the maximum probability is taken as the final short-term power generation of the wind farm. Similarly, the power forecast of a short-term wind farm can also be obtained through probability distribution. The power generation of a medium-term wind farm, the power of a medium-term wind farm, the power generation of a long-term wind farm, and the power of a long-term wind farm can also be obtained through probability distribution.

[0084] As another possible implementation, in order to improve the accuracy of the prediction of the power generation and electricity of the wind farm, the embodiment of the present invention can integrate the prediction results of multiple sub-models instead of using the prediction results of a single sub-model. Specifically, when the first sub-model, the second sub-model and the third sub-model are a random forest model, a gradient boosting tree and a neural network model respectively, the first data after the second processing is input into at least one decision tree of the random forest model to obtain the prediction result of each decision tree; the prediction results of all decision trees are voted to obtain the first prediction result; the second data after the second processing is input into each weak learner of the gradient boosting tree to obtain the prediction result of each weak learner; the prediction results of all weak learners are weighted and summed to obtain the second prediction result; the third data after the second processing is input into the neural network model to obtain the third prediction result.

[0085] The first prediction result, the second prediction result and the third prediction result are multiplied by the first weight, the second weight and the third weight respectively, and the first weight is greater than the second weight and the third weight; the first prediction result, the second prediction result and the third prediction result multiplied by the first weight, the second weight and the third weight are weighted averaged to obtain the short-term wind farm power generation and electricity; or

[0086] The first prediction result, the second prediction result and the third prediction result are multiplied by the fourth weight, the fifth weight and the sixth weight respectively, and the fifth weight is greater than the fourth weight and the sixth weight; the first prediction result, the second prediction result and the third prediction result multiplied by the fourth weight, the fifth weight and the sixth weight are weighted averaged to obtain the mid-term wind farm power and electricity; or

[0087] The first prediction result, the second prediction result and the third prediction result are multiplied by the seventh weight, the eighth weight and the ninth weight respectively, and the ninth weight is greater than the seventh weight and the eighth weight; the first prediction result, the second prediction result and the third prediction result multiplied by the seventh weight, the eighth weight and the ninth weight respectively are weighted averaged to obtain the long-term wind farm power and electricity generation.

[0088] It should be noted that the embodiment of the present invention performs weighted averaging or voting of the prediction results of multiple sub-models, and by integrating the advantages of different models, it can better cope with the complex and changeable wind farm environment and improve the robustness and accuracy of the prediction.

[0089] It is understandable that the third data after the second processing is input into the neural network model to obtain the third prediction result. The neural network model (taking the long short-term memory network (LSTM) neural network model as an example) can be obtained by the following steps:

[0090] Construct an initial LSTM neural network model; input sample data (the third data after the second processing for model training) into the LSTM neural network model to obtain the LSTM model prediction value and prediction error, the prediction error being the difference between the LSTM neural network model prediction value and the actual wind farm power generation; calculate the root mean square error between the LSTM neural network model prediction value and the actual wind farm power generation, and adjust the neuron parameters of the initial LSTM neural network model using the random gradient descent method with batch according to the root mean square error; repeat steps 2 and 3 of the aforementioned steps to obtain the LSTM neural network model.

[0091] It can be seen that in the embodiment of the present invention, the deep learning model (such as LSTM) is combined with the traditional machine learning model (such as random forest model, gradient boosting tree) to give full play to the advantages of both. This hybrid integration takes advantage of the deep learning model in processing time series data while retaining the interpretability of the traditional model to improve the accuracy and interpretability of the prediction.

[0092] It should be noted that the model can be dynamically adjusted and optimized according to the sample data acquired in real time (the third data after the second processing for model training), and new data can be gradually added to the model for training, and the model parameters can be updated to adapt to the changing environment and conditions of the wind farm. Traditional models are usually fixed and difficult to adapt to environmental changes. Adaptive learning enables machine learning models to evolve continuously and always maintain a high degree of adaptability to the current environment, thereby improving the long-term effectiveness and sustained accuracy of machine learning models. In addition, by accessing the latest target data, the machine learning model can update the final prediction results in real time, providing immediate support for the scheduling and operation of the wind farm. Real-time prediction capabilities are crucial in wind farm operations, and can quickly respond to environmental changes, optimize resource allocation, and improve the stability and economy of the power grid. In addition, in an embodiment of the present invention, it is also possible to update in real time based on the acquired sample data through online learning, without the need to train the entire machine learning model.

[0093] It is also important to note that the knowledge of machine learning models that have been trained on data-rich wind farms can be transferred to new wind farms with less data, reducing model training time and improving accuracy. Transfer learning enables new or small wind farms to quickly deploy prediction models, reducing reliance on large amounts of data while improving prediction accuracy. In addition, meta-learning technology is used to quickly adapt to new wind farms or environmental conditions through a "learning how to learn" approach, improving the generalization ability of the model. Meta-learning enables the model to quickly adapt to new scenarios and data distributions, improving the flexibility and adaptability of the model in different environments.

[0094] In the embodiments of the present invention, the final prediction results are displayed using intuitive charts and curves through visualization tools, so that users can more easily understand and analyze the prediction trends, patterns and anomalies. The complex final prediction results become more intuitive and actionable, helping decision makers to quickly identify trends and make wise decisions, and enhancing the practicality of machine learning models.

[0095] In the embodiment of the present invention, edge computing technology is used to delegate some computing tasks to local devices, achieving faster local prediction and response, and reducing the burden on the central server. Edge computing integration improves the response speed and robustness of the system, especially in the case of large network delays, and can provide more real-time prediction support.

[0096] As can be seen from the above content, the embodiment of the present invention simultaneously analyzes short-term data, medium-term data and long-term data, captures the change pattern of data at different time scales, and improves the comprehensiveness of the prediction of power generation and electricity of the wind farm; by considering the changes of data at multiple time scales, the machine learning model can maintain stable accuracy in different time ranges and adapt to short-term fluctuations and long-term trends; by integrating multi-source data, namely the wind turbine operation data of the wind farm, meteorological forecast data and the external environment data of the wind farm, more comprehensive input information is provided for the machine learning model, and the prediction accuracy of power generation and electricity of the wind farm is improved. The wind turbine operation data is collected for all wind turbines in the wind farm, and unified analysis and modeling are performed instead of predicting a single wind turbine separately. The centralized prediction method makes full use of the integrity of the data, can capture the complex interrelationships of wind turbines within the wind farm, and improves the prediction accuracy and robustness of power generation and electricity of the wind farm; according to the changes of wind turbine operation data, meteorological forecast data and external environment data of the wind farm, the weights of the features (first feature, second feature, third feature, fourth feature and fifth feature) are dynamically adjusted, so that the machine learning model can adapt to changes in different times and environments. Traditional feature weights are usually fixed and difficult to cope with environmental changes. Dynamically adjusting feature weights enables machine learning models to maintain high prediction accuracy under changing conditions; by integrating the prediction results of multiple models (i.e., random forest model, gradient boosting tree, and neural network model), that is, weighted averaging or voting the first prediction result, the second prediction result, and the third prediction result, the short-term wind farm power and electricity, the medium-term wind farm power and electricity, and the long-term wind farm power and electricity are obtained, rather than using the prediction results of a single model, so that the prediction results are more accurate and the prediction accuracy of the wind farm power and electricity is improved. In addition, the embodiments of the present invention are not limited to the power prediction of wind farms, but also extend to the prediction of electricity, with a wider range of applications. Traditional models mostly focus on single-objective prediction, while multi-objective prediction enables operators to fully understand the operating status of wind farms from multiple dimensions and provide richer operating information. The embodiments of the present invention can also use causal reasoning to improve the interpretability and prediction accuracy of the model by identifying the causal relationship that affects the power and electricity prediction of wind farms.

[0097] Based on the same concept as the above method embodiment, an embodiment of the present invention provides a device for predicting the generated power and amount of electricity of a wind farm. Figure 3 A schematic block diagram of a wind farm power generation and electricity quantity prediction device according to an embodiment is shown. The device can be arranged at Figure 1 The server side of the architecture shown in Figure 1. Figure 3 As shown, the device 300 may include: an acquisition unit 301, an obtaining unit 302 and a determination unit 303. The main functions of each component unit are as follows:

[0098] The acquisition unit 301 is used to acquire target data, and the target data includes: short-term data, medium-term data and long-term data, and the short-term data, medium-term data and long-term data are respectively the wind turbine operation data, meteorological forecast data and external environment data of the wind farm within a first preset time period, a second preset time period and a third preset time period, the third preset time period is greater than the second preset time period, and the second preset time period is greater than the first preset time period.

[0099] The obtaining unit 302 is used to input the short-term data, the medium-term data and the long-term data into the first sub-model, the second sub-model and the third sub-model respectively to obtain the first prediction result, the second prediction result and the third prediction result.

[0100] The determination unit 303 is used to determine the generated power and amount of electricity of the wind farm according to the first prediction result, the second prediction result and the third prediction result.

[0101] As an implementable manner, the obtaining unit 302 is specifically used to perform a first processing on the short-term data, the medium-term data and the long-term data respectively, wherein the first processing includes cleaning, denoising and / or normalization processing;

[0102] The short-term data, the medium-term data and the long-term data after the first processing are respectively input into the first sub-model, the second sub-model and the third sub-model to obtain a first prediction result, a second prediction result and a third prediction result.

[0103] As a feasible manner, the obtaining unit 302 is specifically used to input the short-term data, the medium-term data and the long-term data after the first processing into a support vector machine model respectively, to obtain a first result, a second result and a third result respectively, the first result includes a first sub-result and a second sub-result, the second result includes a third sub-result and a fourth sub-result, the third result includes a fifth sub-result and a sixth sub-result, the first sub-result represents that the short-term data after the first processing under the result is valid short-term data, the second sub-result represents that the short-term data after the first processing under the result is invalid short-term data, the third sub-result represents that the medium-term data after the first processing under the result is valid medium-term data, the fourth sub-result represents that the medium-term data after the first processing under the result is invalid medium-term data, the fifth sub-result represents that the long-term data after the first processing under the result is valid long-term data, and the sixth sub-result represents that the long-term data after the first processing under the result is invalid long-term data;

[0104] The effective short-term data, the effective medium-term data and the effective long-term data are respectively input into the first sub-model, the second sub-model and the third sub-model to obtain a first prediction result, a second prediction result and a third prediction result.

[0105] As an achievable manner, the obtaining unit 302 is specifically used to extract the first feature, the second feature, the third feature, the fourth feature and the fifth feature for the effective short-term data, the effective medium-term data and the effective long-term data, respectively, wherein the first feature includes: wind turbine model, impeller diameter, rated power, operating status and fault information, the second feature includes: wind speed, wind direction, temperature, air pressure, humidity and rainfall, the third feature includes: power load feature, the power load feature includes: power load demand and load fluctuation, the fourth feature includes: date, time point, season and day of the week, and the fifth feature includes: geographical location of the wind farm and surrounding topography;

[0106] The first feature, second feature, third feature, fourth feature and fifth feature extracted from the valid short-term data, the valid medium-term data and the valid long-term data respectively are input into the first sub-model, the second sub-model and the third sub-model respectively to obtain a first prediction result, a second prediction result and a third prediction result.

[0107] As an implementable manner, the obtaining unit 302 is specifically used to adjust the weights of the first feature, the second feature, the third feature, the fourth feature and the fifth feature respectively extracted from the valid short-term data, the valid mid-term data and the valid long-term data, to obtain the sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature for the valid short-term data, the valid mid-term data and the valid long-term data;

[0108] The sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature of the valid short-term data, the valid medium-term data and the valid long-term data are respectively input into the first sub-model, the second sub-model and the third sub-model to obtain a first prediction result, a second prediction result and a third prediction result.

[0109] As an achievable manner, the obtaining unit 302 is specifically used to perform a second processing on the first data, the second data, and the third data, respectively, wherein the second processing includes feature selection, feature dimension reduction, and / or feature combination, the first data is the sixth feature, the seventh feature, the eighth feature, the ninth feature, and the tenth feature for the effective short-term data, the second data is the sixth feature, the seventh feature, the eighth feature, the ninth feature, and the tenth feature for the effective medium-term data, and the third data is the sixth feature, the seventh feature, the eighth feature, the ninth feature, and the tenth feature for the effective long-term data;

[0110] The first data, the second data and the third data after the second processing are respectively input into the first sub-model, the second sub-model and the third sub-model to obtain a first prediction result, a second prediction result and a third prediction result.

[0111] As a possible implementation method, the first sub-model is a random forest model;

[0112] The obtaining unit 302 is specifically used to input the first data after the second processing into at least one decision tree of the random forest model to obtain the prediction result of each decision tree; and vote on the prediction results of all decision trees to obtain the first prediction result.

[0113] As an implementable manner, the second sub-model is a gradient boosting tree. The obtaining unit 302 is specifically used to input the second processed second data into each weak learner of the gradient boosting tree to obtain the prediction result of each weak learner; and weighted sum the prediction results of all weak learners to obtain the second prediction result.

[0114] As an implementable manner, the determining unit 303 is specifically configured to multiply the first prediction result, the second prediction result, and the third prediction result by a first weight, a second weight, and a third weight, respectively, and the first weight is greater than the second weight and the third weight;

[0115] Taking a weighted average of the first prediction result, the second prediction result and the third prediction result respectively multiplied by the first weight, the second weight and the third weight, to obtain the short-term wind farm power generation and electricity; or

[0116] The first prediction result, the second prediction result and the third prediction result are multiplied by a fourth weight, a fifth weight and a sixth weight respectively, and the fifth weight is greater than the fourth weight and the sixth weight;

[0117] Taking a weighted average of the first prediction result, the second prediction result and the third prediction result respectively multiplied by a fourth weight, a fifth weight and a sixth weight, to obtain the power generation and the amount of electricity of the wind farm in the medium term; or

[0118] The first prediction result, the second prediction result and the third prediction result are multiplied by a seventh weight, an eighth weight and a ninth weight respectively, and the ninth weight is greater than the seventh weight and the eighth weight;

[0119] The first prediction result, the second prediction result and the third prediction result multiplied by the seventh weight, the eighth weight and the ninth weight respectively are weighted averaged to obtain the long-term wind farm power generation and electricity.

[0120] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The device embodiment described above is only schematic, wherein the unit described as a separate component may or may not be physically separated, and the component displayed as a unit may or may not be a physical unit, that is, it may be located in one place, or it may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative work.

[0121] In addition, an embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps of any one of the methods in the aforementioned method embodiments are implemented.

[0122] And an electronic device, comprising:

[0123] one or more processors; and

[0124] A memory associated with the one or more processors, the memory being used to store program instructions, wherein the program instructions, when read and executed by the one or more processors, execute the steps of the method described in any one of the aforementioned method embodiments.

[0125] An embodiment of the present invention further provides a computer program product, including a computer program, which implements the steps of any one of the methods in the aforementioned method embodiments when executed by a processor.

[0126] in, Figure 4 The architecture of the electronic device is exemplarily shown, which may include a processor 410, a video display adapter 411, a disk drive 412, an input / output interface 413, a network interface 414, and a memory 420. The processor 410, the video display adapter 411, the disk drive 412, the input / output interface 413, the network interface 414, and the memory 420 may be communicatively connected via a communication bus 430.

[0127] The processor 410 may be implemented by a general-purpose CPU, a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, etc., and is used to execute relevant programs to implement the technical solutions provided by the embodiments of the present invention.

[0128] The memory 420 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 420 can store an operating system 421 for controlling the operation of the electronic device 400, and a basic input and output system (BIOS) 422 for controlling the low-level operation of the electronic device 400. In addition, a web browser 423, a data storage management system 424, and a wind farm power generation and electricity prediction device 425, etc. can also be stored. The above-mentioned wind farm power generation and electricity prediction device 425 can be an application program that specifically implements the operations of the aforementioned steps in the embodiment of the present invention. In short, when the technical solution provided by the embodiment of the present invention is implemented by software or firmware, the relevant program code is stored in the memory 420 and is called and executed by the processor 410.

[0129] The input / output interface 413 is used to connect the input / output module to realize information input and output. The input / output module can be configured in the device as a component (not shown in the figure), or it can be externally connected to the device to provide corresponding functions. Among them, the input device may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device may include a display, a speaker, a vibrator, an indicator light, etc.

[0130] The network interface 414 is used to connect to a communication module (not shown) to achieve communication interaction between the device and other devices. The communication module can achieve communication through a wired mode (such as USB, network cable, etc.) or a wireless mode (such as mobile network, WIFI, Bluetooth, etc.).

[0131] The bus 430 comprises a pathway for transmitting information between the various components of the device (eg, the processor 410, the video display adapter 411, the disk drive 412, the input / output interface 413, the network interface 414, and the memory 420).

[0132] It should be noted that, although the above device only shows a processor 410, a video display adapter 411, a disk drive 412, an input / output interface 413, a network interface 414, a memory 420, a bus 430, etc., in the specific implementation process, the device may also include other components necessary for normal operation. In addition, it can be understood by those skilled in the art that the above device may also only include components necessary for implementing the solution of the present application, and does not necessarily include all the components shown in the figure.

[0133] The above described system and other embodiments are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art may understand and implement it without creative work.

[0134] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0135] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the embodiments of the present invention, rather than to limit them. Although the embodiments of the present invention have been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting the power generation and quantity of a wind farm, characterized in that: include: Acquire target data, the target data including: short-term data, medium-term data and long-term data, the short-term data, medium-term data and long-term data are wind turbine operation data, meteorological forecast data and external environment data of the wind farm within a first preset time period, a second preset time period and a third preset time period respectively, the third preset time period is longer than the second preset time period, and the second preset time period is longer than the first preset time period; Inputting the short-term data, the medium-term data and the long-term data into the first sub-model, the second sub-model and the third sub-model respectively to obtain a first prediction result, a second prediction result and a third prediction result; The generated power and amount of electricity of the wind farm are determined according to the first prediction result, the second prediction result and the third prediction result.

2. The method according to claim 1, characterized in that The step of inputting the short-term data, the medium-term data and the long-term data into the first sub-model, the second sub-model and the third sub-model respectively to obtain the first prediction result, the second prediction result and the third prediction result comprises: Performing first processing on the short-term data, the medium-term data and the long-term data respectively, wherein the first processing includes cleaning, denoising and / or normalization processing; The short-term data, the medium-term data and the long-term data after the first processing are respectively input into the first sub-model, the second sub-model and the third sub-model to obtain a first prediction result, a second prediction result and a third prediction result.

3. The method according to claim 2, characterized in that The step of inputting the short-term data, the medium-term data and the long-term data after the first processing into the first sub-model, the second sub-model and the third sub-model respectively to obtain a first prediction result, a second prediction result and a third prediction result comprises: Input the short-term data, the mid-term data and the long-term data after the first processing into a support vector machine model respectively, and obtain a first result, a second result and a third result respectively, the first result includes a first sub-result and a second sub-result, the second result includes a third sub-result and a fourth sub-result, the third result includes a fifth sub-result and a sixth sub-result, the first sub-result represents that the short-term data after the first processing under the result is valid short-term data, the second sub-result represents that the short-term data after the first processing under the result is invalid short-term data, the third sub-result represents that the mid-term data after the first processing under the result is valid mid-term data, the fourth sub-result represents that the mid-term data after the first processing under the result is invalid mid-term data, the fifth sub-result represents that the long-term data after the first processing under the result is valid long-term data, and the sixth sub-result represents that the long-term data after the first processing under the result is invalid long-term data; The effective short-term data, the effective medium-term data and the effective long-term data are respectively input into the first sub-model, the second sub-model and the third sub-model to obtain a first prediction result, a second prediction result and a third prediction result.

4. The method according to claim 3, characterized in that The step of inputting the effective short-term data, the effective medium-term data and the effective long-term data into the first sub-model, the second sub-model and the third sub-model respectively to obtain the first prediction result, the second prediction result and the third prediction result comprises: A first feature, a second feature, a third feature, a fourth feature and a fifth feature are extracted from the effective short-term data, the effective medium-term data and the effective long-term data respectively, wherein the first feature includes: wind turbine model, impeller diameter, rated power, operating status and fault information, the second feature includes: wind speed, wind direction, temperature, air pressure, humidity and rainfall, the third feature includes: power load feature, the power load feature includes: power load demand and load fluctuation, the fourth feature includes: date, time point, season and day of the week, and the fifth feature includes: geographical location of the wind farm and surrounding topography; The first feature, second feature, third feature, fourth feature and fifth feature extracted from the valid short-term data, the valid medium-term data and the valid long-term data respectively are input into the first sub-model, the second sub-model and the third sub-model respectively to obtain a first prediction result, a second prediction result and a third prediction result.

5. The method according to claim 4, characterized in that The first feature, the second feature, the third feature, the fourth feature and the fifth feature respectively extracted from the effective short-term data, the effective medium-term data and the effective long-term data are respectively input into the first sub-model, the second sub-model and the third sub-model to obtain a first prediction result, a second prediction result and a third prediction result, including: The weights of the first feature, the second feature, the third feature, the fourth feature and the fifth feature extracted from the valid short-term data, the valid mid-term data and the valid long-term data are adjusted respectively to obtain the sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature from the valid short-term data, the valid mid-term data and the valid long-term data; The sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature of the valid short-term data, the valid medium-term data and the valid long-term data are respectively input into the first sub-model, the second sub-model and the third sub-model to obtain a first prediction result, a second prediction result and a third prediction result.

6. The method according to claim 5, characterized in that The sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature of the effective short-term data, the effective medium-term data and the effective long-term data are input into the first sub-model, the second sub-model and the third sub-model respectively to obtain a first prediction result, a second prediction result and a third prediction result, including: Performing a second processing on the first data, the second data and the third data respectively, wherein the second processing includes feature selection, feature dimension reduction and / or feature combination, the first data is the sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature for the effective short-term data, the second data is the sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature for the effective medium-term data, and the third data is the sixth feature, the seventh feature, the eighth feature, the ninth feature and the tenth feature for the effective long-term data; The first data, the second data and the third data after the second processing are respectively input into the first sub-model, the second sub-model and the third sub-model to obtain a first prediction result, a second prediction result and a third prediction result.

7. The method according to claim 6, characterized in that The first sub-model is a random forest model; The step of inputting the first data, the second data and the third data after the second processing into the first sub-model, the second sub-model and the third sub-model respectively to obtain a first prediction result, a second prediction result and a third prediction result comprises: Inputting the first data after the second processing into at least one decision tree of the random forest model to obtain a prediction result of each decision tree; Vote on the prediction results of all decision trees to get the first prediction result.

8. The method according to claim 6, characterized in that The second sub-model is a gradient boosted tree; The step of inputting the first data, the second data and the third data after the second processing into the first sub-model, the second sub-model and the third sub-model respectively to obtain a first prediction result, a second prediction result and a third prediction result comprises: Inputting the second processed second data into each weak learner of the gradient boosting tree to obtain a prediction result of each weak learner; The prediction results of all weak learners are weighted and summed to obtain the second prediction result.

9. The method according to any one of claims 1 to 8, characterized in that: The determining the power generation power and the amount of electricity of the wind farm according to the first prediction result, the second prediction result and the third prediction result includes: The first prediction result, the second prediction result and the third prediction result are multiplied by a first weight, a second weight and a third weight respectively, and the first weight is greater than the second weight and the third weight; Taking a weighted average of the first prediction result, the second prediction result and the third prediction result respectively multiplied by the first weight, the second weight and the third weight, to obtain the short-term wind farm power generation and electricity; or The first prediction result, the second prediction result and the third prediction result are multiplied by a fourth weight, a fifth weight and a sixth weight respectively, and the fifth weight is greater than the fourth weight and the sixth weight; Taking a weighted average of the first prediction result, the second prediction result and the third prediction result respectively multiplied by a fourth weight, a fifth weight and a sixth weight, to obtain the power generation and the amount of electricity of the wind farm in the medium term; or The first prediction result, the second prediction result and the third prediction result are multiplied by a seventh weight, an eighth weight and a ninth weight respectively, and the ninth weight is greater than the seventh weight and the eighth weight; The first prediction result, the second prediction result and the third prediction result multiplied by the seventh weight, the eighth weight and the ninth weight respectively are weighted averaged to obtain the long-term wind farm power generation and electricity.

10. An electronic device, characterized in that: include: A memory and a processor, wherein the processor and the memory communicate with each other via a bus; The memory stores program instructions executable by the processor, and the processor can execute the method according to any one of claims 1 to 9 by calling the program instructions.

Images