Wind power determination method and device, storage medium and electronic device

The proposed method using a convolutional neural network-based generator and multi-scale discriminator model improves wind power prediction accuracy by capturing local and global features, enhancing the reliability of wind power forecasting.

CN119090304APending Publication Date: 2024-12-06HUANENG MANZHOULI WIND POWER CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411169324.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-23
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

Existing wind power prediction methods fail to capture local detail features, leading to low accuracy in predicting wind power output due to the randomness and volatility of wind generation, which affects power grid stability and increases peak load pressure.

Method used

A method using a data set generated from wind turbine unit and environmental data, processed through a convolutional neural network-based generator and multi-scale discriminator model, incorporating global and local discriminators to improve prediction accuracy by capturing both local and global features.

Benefits of technology

Enhances wind power prediction accuracy by considering local and global consistency, addressing the limitations of previous methods and improving the reliability of wind power forecasting.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119090304A_ABST
    Figure CN119090304A_ABST
Patent Text Reader

Abstract

The invention discloses a wind power determination method and device, a storage medium and an electronic device.The method comprises the steps that wind power data generated when a wind generating set works are obtained, and the wind power data at least comprise set data of the wind generating set and environment data generated when the wind generating set works; a data set generated on the basis of the unit data and the environment data is input into an output power prediction model set for the wind generating set, wind power output by the output power prediction model is obtained, the data set comprises feature vectors corresponding to the unit data and the environment data, and the feature vectors correspond to the unit data and the environment data. The output power prediction model is obtained by training with the historical feature vector in the historical data set as an input sample and the historical wind power corresponding to the historical data set as an output sample, the technical problem that the prediction accuracy of the wind power is low is solved, and the prediction accuracy of the wind power is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of wind power generation, and in particular to a method and device for determining wind power, a storage medium and an electronic device. Background Art

[0002] At present, with the continuous growth of global energy demand, the development and utilization of renewable energy to meet the demand has become an important goal. As one of the forms of renewable energy, wind power generation has achieved rapid development in recent years. However, due to the strong randomness and volatility of wind power generation, the output wind power is affected by many factors, which makes the power prediction process complicated and difficult. These factors include meteorological factors such as wind speed, wind direction, temperature, air pressure, and natural environmental factors such as the geographical location, topography, and landform of the wind farm. In addition, as the proportion of wind power output by wind power generation in the power system gradually increases, the impact of the state of wind power generation on the stable operation of the power grid is also increasingly significant. It is certain that the volatility and uncertainty of the output wind power have brought challenges to the dispatching and operation of the power grid, and increased the peak load pressure and spare capacity demand of the power grid.

[0003] In the related art, the wind power prediction method based on regression prediction is to establish a mathematical relationship between wind power and influencing factors (such as wind speed, wind direction, temperature, air pressure, etc.), and use regression analysis technology to predict the future output wind power. The core of this method is to determine the functional relationship between each influencing factor and wind power, and to train and optimize this relationship model through historical data. However, such methods often ignore the influence of historical values ​​and do not make full use of the time dependence in time series data. Therefore, the related art further proposes a wind power prediction method based on time series, specifically using the historical information of time series data such as wind power or wind speed to predict future change trends. The core of this method is to use classic statistical models such as ARMA (Autoregressive Moving Average), ARIMA (Autoregressive Integrated Moving Average), or deep models such as LSTM (Long Short-Term Memory) to mine the time dependence and periodicity in time series data, and establish a prediction model based on this. However, such methods usually fail to capture local detail features, which will affect their prediction performance and reduce the accuracy of the prediction.

[0004] With regard to the technical problem of low wind power prediction accuracy in related technologies, no effective solution has been proposed so far.

[0005] Therefore, it is necessary to improve the related technology to overcome the above-mentioned defects in the related technology. Summary of the invention

[0006] The embodiments of the present application provide a method and device for determining wind power, a storage medium, and an electronic device to at least solve the technical problem of low wind power prediction accuracy.

[0007] According to one aspect of an embodiment of the present application, a method for determining wind power is provided, comprising: obtaining wind power data when a wind turbine is operating, wherein the wind power data at least includes unit data of the wind turbine and environmental data when the wind turbine is operating; inputting a data set generated based on the unit data and the environmental data into an output power prediction model set for the wind turbine to obtain wind power output by the output power prediction model, wherein the data set includes feature vectors corresponding to the unit data and the environmental data, and the output power prediction model is trained using historical feature vectors in a historical data set as input samples and using historical wind power corresponding to the historical data set as output samples.

[0008] In an exemplary embodiment, before inputting the data set generated based on the unit data and the environmental data into an output power prediction model set for the wind turbine generator set, the method also includes: generating a data set based on the unit data and the environmental data, including: performing data cleaning operations on the unit data and the environmental data to obtain cleaned data; normalizing the cleaned data to obtain normalized data; extracting a first eigenvector corresponding to the unit data from the normalized data, and extracting a second eigenvector corresponding to the environmental data; and determining the data set based on the first eigenvector and the second eigenvector.

[0009] In an exemplary embodiment, the method further includes: determining a model structure constructed for the output power prediction model, wherein the model structure includes a generator and a discriminator, the generator adopts a network structure of a convolutional neural network, and the discriminator includes a global discriminator, a local discriminator and a statistical indicator discriminator.

[0010] In an exemplary embodiment, the method further includes: initializing the parameters of the generator, the global discriminator, the local discriminator and the statistical indicator discriminator, and performing batch training using historical feature vectors; for each training batch, training the generator while fixing the discriminator, and training the discriminator while fixing the generator.

[0011] In an exemplary embodiment, during the process of training the generator, the method further includes: obtaining a global discriminator feedback value generated by the global discriminator when training the generator, a local discriminator feedback value generated by the local discriminator when training the generator, and a statistical indicator discriminator feedback value generated by the statistical indicator discriminator when training the generator, and determining a total loss value of the generator according to the global discriminator feedback value, the local discriminator feedback value, and the statistical indicator discriminator feedback value, and minimizing the total loss value of the generator by updating the parameters of the generator; wherein the total loss value of the generator is expressed as follows: L G is the total loss value of the generator, is the global discriminator feedback value, is the local discriminator feedback value, is the feedback value of the statistical indicator discriminator, λ1 is the weight coefficient corresponding to the feedback value of the local discriminator, and λ2 is the weight coefficient corresponding to the feedback value of the statistical indicator discriminator.

[0012] In an exemplary embodiment, during the process of training the discriminator, the method further includes: for the global discriminator, obtaining a first loss value during the training of the global discriminator, and minimizing the first loss value by updating the training parameters of the global discriminator; for the local discriminator, obtaining a second loss value during the training of the local discriminator, and minimizing the second loss value by updating the training parameters of the global discriminator.

[0013] According to another aspect of an embodiment of the present application, a device for determining wind power is also provided, including: an acquisition module, used to acquire wind power data when a wind turbine is working, wherein the wind power data at least includes unit data of the wind turbine and environmental data when the wind turbine is working; an acquisition module, used to input a data set generated based on the unit data and the environmental data into an output power prediction model set for the wind turbine, to obtain wind power output by the output power prediction model, wherein the data set includes feature vectors corresponding to the unit data and the environmental data, and the output power prediction model is trained using historical feature vectors in a historical data set as input samples and using historical wind power corresponding to the historical data set as output samples.

[0014] According to another aspect of the embodiments of the present application, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the above-mentioned method for determining wind power when running.

[0015] According to another aspect of the embodiments of the present application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the method for determining wind power through the computer program.

[0016] According to another aspect of the present application, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in any one of the above method embodiments are implemented.

[0017] Through this application, wind power data of a wind turbine generator set when it is working is obtained, wherein the wind power data at least includes the unit data of the wind turbine generator set and the environmental data of the wind turbine generator set when it is working; a data set generated based on the unit data and the environmental data is input into an output power prediction model set for the wind turbine generator set to obtain the wind power output by the output power prediction model, wherein the data set includes feature vectors corresponding to the unit data and the environmental data, and the output power prediction model is trained using historical feature vectors in a historical data set as input samples and historical wind power corresponding to the historical data set as output samples. This application takes into account local and global consistency, and uses an output power prediction model to accurately predict the wind power of a wind turbine generator set. The prediction of wind power is of great significance, and can solve the technical problem of low accuracy in wind power prediction, thereby improving the accuracy in wind power prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0019] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0020] Figure 1 It is a hardware structure block diagram of a mobile terminal of a method for determining wind power in an embodiment of the present application;

[0021] Figure 2 is a flow chart of a method for determining wind power according to an embodiment of the present application;

[0022] Figure 3 is a schematic diagram of a process for determining wind power according to an embodiment of the present application;

[0023] Figure 4is a structural schematic diagram of a generator according to an embodiment of the present application;

[0024] Figure 5 It is a structural block diagram of a device for determining wind power according to an embodiment of the present application. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] Next, some technical terms involved in the embodiments of this application are explained:

[0028] 1.ARMA (Autoregressive Moving Average): Autoregressive Moving Average model. This is a time series forecasting model that combines autoregressive and moving average parts.

[0029] 2.ARIMA (Autoregressive Integrated Moving Average): Autoregressive Integrated Moving Average Model. ARIMA is an extension of ARMA and is suitable for time series data with non-stationarity. It converts non-stationary time series into stationary series through difference operation, and then uses ARMA model for prediction.

[0030] 3. LSTM (Long Short-Term Memory): Long Short-Term Memory Network. LSTM is a special type of recurrent neural network (RNN) that can learn long-distance dependencies. It performs well in processing sequence data, such as natural language processing and time series prediction.

[0031] 4. GAN (Generative Adversarial Network): Generative Adversarial Network. GAN is a deep learning model consisting of a generator and a discriminator. The generator is responsible for generating new data instances, while the discriminator is responsible for distinguishing between real data and fake data generated by the generator. The two compete with each other to improve the quality of generated data.

[0032] 5. CNN (Convolutional Neural Network): Convolutional Neural Network. CNN is a deep learning model that is particularly suitable for processing data with grid structures such as images. It extracts local features through the convolution layer, then uses the pooling layer to reduce the feature dimension, and finally uses the fully connected layer for classification or other tasks.

[0033] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 FIG. 1 is a hardware structure block diagram of a mobile terminal of a method for determining wind power in an embodiment of the present application. Figure 1 As shown, the mobile terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor (Central Processing Unit, MCU) or a programmable logic device (Field Programmable Gate Array, FPGA)) and a memory 104 for storing data, wherein the mobile terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is for illustration only and does not limit the structure of the mobile terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0034] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the method for determining wind power in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0035] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0036] In this embodiment, a method for determining wind power is provided. Figure 2 is a flow chart of a method for determining wind power according to an embodiment of the present application, such as Figure 2 As shown, the process includes the following steps:

[0037] Step S202: Acquire wind power data when the wind turbine generator set is in operation, wherein the wind power data at least includes unit data of the wind turbine generator set and environmental data when the wind turbine generator set is in operation;

[0038] The environmental data may include, for example, wind speed, wind direction, ambient temperature and other data at different altitudes.

[0039] Step S204: Input the data set generated based on the unit data and the environmental data into the output power prediction model set for the wind turbine generator set to obtain the wind power output by the output power prediction model, wherein the data set includes feature vectors corresponding to the unit data and the environmental data, and the output power prediction model is trained using historical feature vectors in a historical data set as input samples and using historical wind power corresponding to the historical data set as output samples.

[0040] Through the above steps, the wind power data of the wind turbine generator set when it is working is obtained, wherein the wind power data at least includes the unit data of the wind turbine generator set and the environmental data of the wind turbine generator set when it is working; the data set generated based on the unit data and the environmental data is input into the output power prediction model set for the wind turbine generator set to obtain the wind power output by the output power prediction model, wherein the data set includes the feature vectors corresponding to the unit data and the environmental data, and the output power prediction model is trained with the historical feature vectors in the historical data set as input samples and the historical wind power corresponding to the historical data set as output samples. The present application takes into account local and global consistency, and uses the output power prediction model to accurately predict the wind power of the wind turbine generator set. The prediction of wind power is of great significance, and can solve the technical problem of low accuracy in wind power prediction, thereby improving the accuracy in wind power prediction.

[0041] In an exemplary embodiment, before executing step S204 to input the data set generated based on the unit data and the environmental data into the output power prediction model set for the wind turbine generator set, a data set can be further generated based on the unit data and the environmental data, and the specific steps include: performing data cleaning operations on the unit data and the environmental data to obtain cleaned data; normalizing the cleaned data to obtain normalized data; extracting a first eigenvector corresponding to the unit data from the normalized data, and extracting a second eigenvector corresponding to the environmental data; and determining the data set based on the first eigenvector and the second eigenvector.

[0042] This embodiment considers improving the generative adversarial network model to obtain an output power prediction model. The structure of the improved generative adversarial network model in this solution includes a generator and a discriminator. This improved model introduces a multi-scale discriminator and distribution difference constraints based on the idea of ​​GAN, which to a certain extent avoids the shortcomings of the original GAN ​​network predicting time series data lacking detailed features and being too smooth. This embodiment preprocesses the data and inputs the preprocessed data into the generative adversarial network for prediction, integrating the successful experience of the generative adversarial network in the field of computer vision and migrating the structure to the time series prediction problem. By improving the structure of the original adversarial network, it can take into account both the local and global features of the time series, thereby improving the prediction accuracy.

[0043] In an exemplary embodiment, further, the following steps can be implemented: determine a model structure constructed for the output power prediction model, wherein the model structure includes a generator and a discriminator, the generator adopts a network structure of a convolutional neural network, and the discriminator includes a global discriminator, a local discriminator and a statistical indicator discriminator.

[0044] In this embodiment, a generator is constructed by CNN, and multiple convolution kernels of different sizes are used to capture multi-scale information in the input historical time series. The input of the generator is the historical time series data, and the output is the time series data to be predicted.

[0045] In order to take into account both the local details and the overall trend of the generated sequence, this embodiment specifically sets three discriminators, which jointly evaluate the prediction data generated by the generator.

[0046] The first discriminator is the global discriminator, which receives the real time series data of the forecast period and the forecast data generated by the generator as input, and outputs a scalar value representing the probability that the forecast data is a real sample. This discriminator focuses on the accuracy of the overall trend.

[0047] The second is the local discriminator, which has the same input as the global discriminator, but outputs a 1×n vector, where n represents the n local segments into which the prediction data is segmented. The local discriminator evaluates each segment separately to ensure that the generated prediction data is indistinguishable from the real in each local area, thereby improving the local accuracy of the prediction.

[0048] The third is the statistical indicator discriminator, which focuses on the statistical characteristics of the generated data. The discriminator receives the generated time series data as input and outputs a 1×5 vector containing key statistical indicators such as the maximum, minimum, mean, median and variance of the generated data. By calculating the L2 distance between these statistical indicators and the corresponding indicators of the real data, the statistical indicator discriminator imposes additional constraints on the generator, making the generated prediction data closer to the real data at the statistical level, further improving the overall performance of the prediction.

[0049] Among them, L2 distance (also known as Euclidean distance or Euclidean distance) is a method of measuring the straight-line distance between two points in Euclidean space. In statistics and data analysis, L2 distance is often used to measure the similarity or difference between data points.

[0050] In an exemplary embodiment, further, the following technical solution is proposed: the parameters of the generator, the global discriminator, the local discriminator and the statistical indicator discriminator are initialized, and batch training is performed using historical feature vectors; for each training batch, the generator is trained while the discriminator is fixed, and the discriminator is trained while the generator is fixed.

[0051] In an exemplary embodiment, during the process of training the generator, the global discriminator feedback value generated by the global discriminator when training the generator, the local discriminator feedback value generated by the local discriminator when training the generator, and the statistical indicator discriminator feedback value generated by the statistical indicator discriminator when training the generator can also be obtained, and the total loss value of the generator is determined according to the global discriminator feedback value, the local discriminator feedback value, and the statistical indicator discriminator feedback value, and the total loss value of the generator is minimized by updating the parameters of the generator; wherein the total loss value of the generator is expressed as follows: L G is the total loss value of the generator, is the global discriminator feedback value, is the local discriminator feedback value, is the feedback value of the statistical indicator discriminator, λ1 is the weight coefficient corresponding to the feedback value of the local discriminator, and λ2 is the weight coefficient corresponding to the feedback value of the statistical indicator discriminator.

[0052] In an exemplary embodiment, in the process of training the discriminator, further, for the global discriminator, a first loss value during the training of the global discriminator is obtained, and the first loss value is minimized by updating the training parameters of the global discriminator; for the local discriminator, a second loss value during the training of the local discriminator is obtained, and the second loss value is minimized by updating the training parameters of the global discriminator.

[0053] Optionally, the first loss value is expressed as follows:

[0054] X is the independent variable.

[0055] Optionally, the second loss value is expressed as follows:

[0056] i is a positive integer.

[0057] Obviously, the embodiments described above are only some embodiments of the present application, not all embodiments. In order to better understand the above method, the above process is described below in conjunction with the embodiments, but it is not intended to limit the technical solutions of the embodiments of the present application, specifically:

[0058] In an optional embodiment, the present application considers local and global consistency, combined with Figure 3 The implementation flow shown in the figure describes in detail the prediction process of the output wind power. Specifically, the steps are as follows:

[0059] Step 1: Data preparation and preprocessing.

[0060] 1. Data collection: Collect unit data and meteorological data. The unit data obtained is the actual operation data of the unit. Meteorological data can be understood as environmental data such as wind speed, wind direction, ambient temperature, etc. at different altitudes.

[0061] 2. Data cleaning: remove missing values ​​and outliers, and normalize the data for input into the base learner.

[0062] 3. Feature extraction: Based on the Pearson correlation coefficient, the features related to wind power prediction are extracted from the original data. Through the above steps, a preprocessed data set (i.e., historical data set) is obtained, which contains multiple samples, each of which has a series of features related to wind power prediction.

[0063] Step 2: Model construction.

[0064] 1. Generator: The generator G uses a convolutional neural network (CNN) structure, which aims to capture multi-scale information in the input historical time series and generate forecast data for future time periods. The specific structure of the generator network is as follows: Figure 4 As shown in the figure, it includes convolutional layers, activation layers and pooling layers, and finally outputs the predicted sequence through the fully connected layer.

[0065] Input: Historical time series data X = {x1, x2, ..., x T}, where x t is the observed value at time t (such as wind power, wind speed, etc.).

[0066] Output: Forecast data for future time series in is the predicted value at time t.

[0067] 2. Discriminator: The structure of the discriminator can be any neural network structure, but its core function is to discriminate input samples and distinguish whether these samples come from the real data distribution or are generated by the generator.

[0068] The output layer of the discriminator outputs a probability value, indicating the probability that the input sample is a true sample. In the binary classification problem, the output layer is usually a single node, and the sigmoid activation function is used to compress the output value into the (0, 1) interval.

[0069] In this scheme, the output layer of the global discriminator is one node, and the output layer of the local discriminator is n nodes, which output a 1×n vector, in which each value represents the probability that the segment is a true sample.

[0070] Global Discriminator (D global ): Evaluates the authenticity and overall trend of the entire prediction period, and outputs the probability (scalar) that the predicted data is a true sample.

[0071] Local Discriminator (D local ): Evaluate the predicted data on a local segment basis and output a false-true discriminant vector (1×n) for each segment.

[0072] Statistical indicator discriminator (D stats ): Evaluate the statistical characteristics of the generated data and output a vector (1×5) containing key statistical indicators, constrained by L2 distance.

[0073] Step 3: Model training.

[0074] 1. Initialization: Initialize the parameters of the generator, global discriminator, local discriminator and statistical indicator discriminator.

[0075] 2. Circuit training.

[0076] First, for each training batch, fix the discriminator and train the generator: input historical time series data into the generator to generate predicted time series data.

[0077] The global discriminator feedback value is expressed as follows:

[0078] Where D global (G(z)) is the probability estimate of the global discriminator that the generated data is real. G(z) is the generated data.

[0079] The local discriminator feedback value is expressed as follows: the predicted data is divided into n segments, each segment is input into the local discriminator separately, and the local loss is calculated using cross entropy loss or similar loss

[0080]

[0081] Where D local (G(z) i ) is the discriminant output of the local discriminator for the i-th segment.

[0082] The feedback value of the statistical indicator discriminator is expressed as follows: Calculate the L2 distance between the statistical indicator vector of the generated data and the statistical indicator vector of the real data to get the statistical loss

[0083]

[0084] Among them, M ganj and M realj are the values ​​of the generated data and the real data on the jth statistical indicator respectively.

[0085] The total loss value of the generator is expressed as follows:

[0086]

[0087] Update the parameters of the generator to minimize L G .

[0088] Then, the generator is fixed and the discriminator is trained.

[0089] Global discriminator training: input real data and generated data into the global discriminator and calculate the loss of the global discriminator Using cross entropy loss:

[0090]

[0091] Update the parameters of the global discriminator to minimize

[0092] Local discriminator training (similar to global discriminator, but for each segment):

[0093]

[0094] Update the parameters of the local discriminator to minimize

[0095] Statistical indicator discriminators are usually not trained directly. They are mainly used to calculate the statistical characteristics of generated data and use statistical loss The generator is constrained to ensure that the generated predictions are not only visually indistinguishable from the real data (through adversarial training of global and local discriminators), but also statistically consistent with the real data.

[0096] Step 4: Input the real-time collected data into the trained model for prediction to obtain the wind power predicted by the model.

[0097] It should be noted that through the above steps, the architecture design of the multi-scale discriminator can pay more attention to the local characteristics of the time series, and introduce a statistical indicator discriminator to focus on the statistical characteristics of the generated data. By calculating the L2 distance between the key statistical indicators and the real data, the generator is additionally constrained to make the predicted data closer to the real data at the statistical level. This method of combining local and global consistency and statistical difference constraints helps to improve the accuracy and reliability of wind power forecasting.

[0098] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of each embodiment of the present application.

[0099] In this embodiment, a device for determining wind power is also provided, which is used to implement the above embodiments and preferred implementations, and will not be repeated here. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0100] Figure 5 is a structural block diagram of a device for determining wind power according to an embodiment of the present application, such as Figure 5 As shown, the device comprises:

[0101] An acquisition module 52, used for acquiring wind power data when the wind turbine generator set is in operation, wherein the wind power data at least includes unit data of the wind turbine generator set and environmental data when the wind turbine generator set is in operation;

[0102] Obtaining module 54, used for inputting the data set generated based on the unit data and the environmental data into the output power prediction model set for the wind turbine generator set, to obtain the wind power output by the output power prediction model, wherein the data set includes feature vectors corresponding to the unit data and the environmental data, and the output power prediction model is trained using historical feature vectors in a historical data set as input samples and using historical wind power corresponding to the historical data set as output samples.

[0103] Through the above-mentioned device, the wind power data of the wind turbine generator set when it is working is obtained, wherein the wind power data at least includes the unit data of the wind turbine generator set and the environmental data of the wind turbine generator set when it is working; the data set generated based on the unit data and the environmental data is input into the output power prediction model set for the wind turbine generator set to obtain the wind power output by the output power prediction model, wherein the data set includes the feature vectors corresponding to the unit data and the environmental data, and the output power prediction model is trained with the historical feature vectors in the historical data set as input samples and the historical wind power corresponding to the historical data set as output samples. The present application takes into account local and global consistency, and uses the output power prediction model to accurately predict the wind power of the wind turbine generator set. The prediction of wind power is of great significance, and can solve the technical problem of low accuracy in wind power prediction, thereby improving the accuracy in wind power prediction.

[0104] In an exemplary embodiment, the acquisition module 52 is also used to generate a data set based on the unit data and the environmental data before inputting the data set generated based on the unit data and the environmental data into the output power prediction model set for the wind turbine generator set, including: performing data cleaning operations on the unit data and the environmental data to obtain cleaned data; normalizing the cleaned data to obtain normalized data; extracting a first eigenvector corresponding to the unit data from the normalized data, and extracting a second eigenvector corresponding to the environmental data; and determining the data set based on the first eigenvector and the second eigenvector.

[0105] In an exemplary embodiment, the above-mentioned wind power determination device also includes a determination module for determining a model structure constructed for the output power prediction model, wherein the model structure includes a generator and a discriminator, the generator adopts a network structure of a convolutional neural network, and the discriminator includes a global discriminator, a local discriminator and a statistical indicator discriminator.

[0106] In an exemplary embodiment, the above-mentioned wind power determination device also includes a training module for initializing the parameters of the generator, the global discriminator, the local discriminator and the statistical indicator discriminator, and performing batch training using historical feature vectors; for each training batch, the generator is trained while the discriminator is fixed, and the discriminator is trained while the generator is fixed.

[0107] In an exemplary embodiment, the training module is further used to obtain, during the process of training the generator, a global discriminator feedback value generated by the global discriminator when training the generator, a local discriminator feedback value generated by the local discriminator when training the generator, and a statistical indicator discriminator feedback value generated by the statistical indicator discriminator when training the generator, and determine the total loss value of the generator according to the global discriminator feedback value, the local discriminator feedback value, and the statistical indicator discriminator feedback value, and minimize the total loss value of the generator by updating the parameters of the generator; wherein the total loss value of the generator is expressed as follows: L G is the total loss value of the generator, is the global discriminator feedback value, is the local discriminator feedback value, is the feedback value of the statistical indicator discriminator, λ1 is the weight coefficient corresponding to the feedback value of the local discriminator, and λ2 is the weight coefficient corresponding to the feedback value of the statistical indicator discriminator.

[0108] In an exemplary embodiment, the training module is also used to, during the process of training the discriminator, obtain, for the global discriminator, a first loss value of the global discriminator during training, and minimize the first loss value by updating the training parameters of the global discriminator; and obtain, for the local discriminator, a second loss value of the local discriminator during training, and minimize the second loss value by updating the training parameters of the global discriminator.

[0109] An embodiment of the present application further provides a computer-readable storage medium, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above method embodiments when run.

[0110] Optionally, in this embodiment, the storage medium may be configured to store a computer program for performing the following steps:

[0111] S1, acquiring wind power data when a wind turbine generator set is in operation, wherein the wind power data at least includes unit data of the wind turbine generator set and environmental data when the wind turbine generator set is in operation;

[0112] S2. Input a data set generated based on the unit data and the environmental data into an output power prediction model set for the wind turbine generator set to obtain the wind power output by the output power prediction model, wherein the data set includes feature vectors corresponding to the unit data and the environmental data, and the output power prediction model is trained using historical feature vectors in a historical data set as input samples and using historical wind power corresponding to the historical data set as output samples.

[0113] In an exemplary embodiment, the above-mentioned computer-readable storage medium may include, but is not limited to: a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and other media that can store computer programs.

[0114] For specific examples in this embodiment, reference may be made to the examples described in the above embodiments and exemplary implementation modes, and this embodiment will not be described in detail herein.

[0115] An embodiment of the present application further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0116] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0117] S1, acquiring wind power data when a wind turbine generator set is in operation, wherein the wind power data at least includes unit data of the wind turbine generator set and environmental data when the wind turbine generator set is in operation;

[0118] S2. Input a data set generated based on the unit data and the environmental data into an output power prediction model set for the wind turbine generator set to obtain the wind power output by the output power prediction model, wherein the data set includes feature vectors corresponding to the unit data and the environmental data, and the output power prediction model is trained using historical feature vectors in a historical data set as input samples and using historical wind power corresponding to the historical data set as output samples.

[0119] In an exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0120] An embodiment of the present application further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the steps in any one of the above method embodiments are implemented.

[0121] An embodiment of the present application further provides another computer program product, including a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in any of the above method embodiments are implemented.

[0122] An embodiment of the present application also provides a computer program, which includes computer instructions stored in a computer-readable storage medium; a processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the steps in any one of the above method embodiments.

[0123] For specific examples in this embodiment, reference may be made to the examples described in the above embodiments and exemplary implementation modes, and this embodiment will not be described in detail herein.

[0124] Obviously, those skilled in the art should understand that the above modules or steps of the present application can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order from that herein, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0125] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for determining wind power, characterized in that: include: Acquire wind power data when the wind turbine generator set is in operation, wherein the wind power data at least includes unit data of the wind turbine generator set and environmental data when the wind turbine generator set is in operation; A data set generated based on the unit data and the environmental data is input into an output power prediction model set for the wind turbine generator set to obtain the wind power output by the output power prediction model, wherein the data set includes feature vectors corresponding to the unit data and the environmental data, and the output power prediction model is trained using historical feature vectors in a historical data set as input samples and using historical wind power corresponding to the historical data set as output samples.

2. The method according to claim 1, characterized in that Before inputting the data set generated based on the unit data and the environmental data into an output power prediction model set for the wind turbine generator set, the method further includes: Generating a data set based on the unit data and the environmental data includes: Performing data cleaning operations on the unit data and the environmental data to obtain cleaned data; performing normalization processing on the cleaned data to obtain normalized data; Extracting a first feature vector corresponding to the unit data and a second feature vector corresponding to the environment data from the normalized data; The data set is determined based on the first feature vector and the second feature vector.

3. The method according to claim 1, characterized in that: The method further comprises: Determine a model structure constructed for the output power prediction model, wherein the model structure includes a generator and a discriminator, the generator adopts a network structure of a convolutional neural network, and the discriminator includes a global discriminator, a local discriminator and a statistical indicator discriminator.

4. The method according to claim 3, characterized in that The method further comprises: Initializing the parameters of the generator, the global discriminator, the local discriminator and the statistical indicator discriminator, and performing batch training using historical feature vectors; For each training batch, the generator is trained with the discriminator fixed, and vice versa.

5. The method according to claim 4, characterized in that During the training of the generator, the method further comprises: Obtaining a global discriminator feedback value generated by the global discriminator when training the generator, a local discriminator feedback value generated by the local discriminator when training the generator, and a statistical indicator discriminator feedback value generated by the statistical indicator discriminator when training the generator, and determining a total loss value of the generator according to the global discriminator feedback value, the local discriminator feedback value, and the statistical indicator discriminator feedback value, and minimizing the total loss value of the generator by updating the parameters of the generator; Among them, the total loss value of the generator is expressed as follows: L G is the total loss value of the generator, is the global discriminator feedback value, is the local discriminator feedback value, is the feedback value of the statistical indicator discriminator, λ1 is the weight coefficient corresponding to the feedback value of the local discriminator, and λ2 is the weight coefficient corresponding to the feedback value of the statistical indicator discriminator.

6. The method according to claim 5, characterized in that In the process of training the discriminator, the method further comprises: For the global discriminator, obtaining a first loss value during training of the global discriminator, and minimizing the first loss value by updating training parameters of the global discriminator; For the local discriminator, a second loss value during training of the local discriminator is obtained, and the second loss value is minimized by updating the training parameters of the global discriminator.

7. A device for determining wind power, characterized in that: include: An acquisition module, used for acquiring wind power data when the wind turbine generator set is in operation, wherein the wind power data at least includes unit data of the wind turbine generator set and environmental data when the wind turbine generator set is in operation; An obtaining module is used to input a data set generated based on the unit data and the environmental data into an output power prediction model set for the wind turbine generator set to obtain the wind power output by the output power prediction model, wherein the data set includes feature vectors corresponding to the unit data and the environmental data, and the output power prediction model is trained using historical feature vectors in a historical data set as input samples and using historical wind power corresponding to the historical data set as output samples.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the program executes the method according to any one of claims 1 to 6 when executed.

9. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 6 through the computer program.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.