Generative difference training method and prediction method for wind power low output event

Abstract

The invention relates to a wind power low output event generation type difference training method and prediction method. The training method comprises the following steps: obtaining operation data of at least one wind power plant and numerical weather forecast data of at least one wind power plant in the same time period; constructing a first training set according to the operation data and the numerical weather forecast data; the first training set refers to a real data set related to the wind power low output event; training a pre-constructed event sample generation model according to the first training set and the obtained noise data so as to generate a second training set according to the trained event sample generation model; the second training set refers to a generated data set related to the wind power low output event; and training a pre-constructed event prediction model based on the first training set and the second training set to obtain a trained event prediction model. According to the method provided by the invention, the model capable of accurately predicting the wind power low output event in the future is obtained through training, and the safety and stability of the power grid and the power supply guaranteeing capability are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power generation, and in particular to a wind power low output event generation-type differential training method and a prediction method. Background Art

[0002] As a clean and renewable energy source, wind power plays a vital role in ensuring electricity supply. However, the high randomness and volatility of wind power pose significant challenges, especially as the proportion of wind power connected to the grid continues to increase, putting even greater pressure on the power system's ability to maintain power supply. A wind power low output event refers to a significant drop in wind power output or shutdown during a specific period of time, severely impacting the grid's supply and demand balance and potentially leading to safety incidents such as a drop in grid frequency and power outages. Therefore, accurately predicting wind power low output events is key to improving the grid's security and stability and its ability to maintain power supply. Summary of the Invention

[0003] In order to solve the above technical problems, the embodiments of the present disclosure provide a wind power low output event generation differential training method and prediction method.

[0004] In a first aspect, an embodiment of the present disclosure provides a method for generating differential training for wind power low output events, comprising:

[0005] Obtaining operating data of at least one wind farm and numerical weather forecast data of at least one wind farm during the same period;

[0006] Constructing a first training set based on the operational data and the numerical weather forecast data; wherein the first training set refers to a real data set related to wind power low output events;

[0007] A pre-built event sample generation model is trained based on the first training set and the acquired noise data, so as to generate a second training set based on the trained event sample generation model; wherein the second training set is a generated data set related to the wind power low output event;

[0008] A pre-built event prediction model is trained based on the first training set and the second training set to obtain a trained event prediction model; wherein the event prediction model is used to predict wind power low output events.

[0009] In a second aspect, an embodiment of the present disclosure provides a method for predicting a low wind power output event, comprising:

[0010] Get weather forecast data for a preset period;

[0011] Weather forecast data is used as input of an event prediction model to predict low wind power output events within a preset time period and obtain event prediction results; wherein the event prediction model is trained using the wind power low output event generative differential training method of the first aspect.

[0012] In a third aspect, an embodiment of the present disclosure provides a wind power low output event generation type differential training device, comprising:

[0013] A first acquisition unit is configured to acquire operating data of at least one wind farm and numerical weather forecast data of at least one wind farm in the same period;

[0014] A construction unit is used to construct a first training set based on the operation data and the numerical weather forecast data; wherein the first training set is a real data set related to the wind power low output event;

[0015] A first training unit is configured to train a pre-built event sample generation model based on the first training set and the acquired noise data, so as to generate a second training set based on the trained event sample generation model; wherein the second training set is a generated data set related to the wind power low output event;

[0016] The second training unit is used to train a pre-built event prediction model based on the first training set and the second training set to obtain a trained event prediction model; wherein the event prediction model is used to predict wind power low output events.

[0017] In a fourth aspect, an embodiment of the present disclosure provides a device for predicting low wind power output events, comprising:

[0018] A second acquisition unit, configured to acquire weather forecast data for a preset period of time;

[0019] The event prediction unit is used to use weather forecast data as input of the event prediction model to predict low wind power output events within a preset time period and obtain event prediction results; wherein the event prediction model is trained using the method of the first aspect.

[0020] In a fifth aspect, an embodiment of the present disclosure provides an electronic device, including:

[0021] Memory;

[0022] processor; and

[0023] computer programs;

[0024] The computer program is stored in the memory and is configured to be executed by the processor to implement the methods of the first and second aspects described above.

[0025] In a sixth aspect, an embodiment of the present disclosure provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the methods of the first and second aspects described above are implemented.

[0026] The present disclosure provides a method for training a wind power low output event generation formula with a difference, including: obtaining operating data of at least one wind farm and numerical weather forecast data of at least one wind farm in the same period; constructing a first training set based on the operating data and the numerical weather forecast data; wherein the first training set refers to a real data set related to the wind power low output event; training a pre-constructed event sample generation model based on the first training set and the obtained noise data, so as to generate a second training set based on the trained event sample generation model; wherein the second training set refers to a generated data set related to the wind power low output event; training a pre-constructed event prediction model based on the first training set and the second training set to obtain a trained event prediction model; wherein the event prediction model is used to predict wind power low output events. The method provided in the present application trains a model that can accurately predict future wind power low output events, effectively improving the security and stability of the power grid and the ability to "guarantee power supply." BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0028] In order to more clearly illustrate the embodiments of the present disclosure 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, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0029] Figure 1 A flow chart of a method for generating differential training for wind power low output events provided by an embodiment of the present disclosure;

[0030] Figure 2 A schematic diagram of the structure of an event sample generation model provided by an embodiment of the present disclosure;

[0031] Figure 3 A schematic diagram of the structure of an event prediction model provided by an embodiment of the present disclosure;

[0032] Figure 4 A schematic flow chart of a method for predicting low wind power output events provided by an embodiment of the present disclosure;

[0033] Figure 5 A flow chart of another method for generating differential training for wind power low output events provided by an embodiment of the present disclosure;

[0034] Figure 6 A schematic structural diagram of a wind power low output event generation differential training device provided by an embodiment of the present disclosure;

[0035] Figure 7 A schematic diagram of the structure of a device for predicting low wind power output events provided by an embodiment of the present disclosure;

[0036] Figure 8 A schematic structural diagram of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0037] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments of the present disclosure and the features therein can be combined with each other in the absence of conflict.

[0038] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.

[0039] Currently, much research focuses on predicting future wind power output sequences, but there is limited research specifically on predicting low wind power output events. Due to the low probability of wind power low output events and their significant small sample size, relevant prediction models are difficult to effectively train using only representative samples, resulting in suboptimal prediction results. Therefore, there is an urgent need to develop small-sample prediction methods suitable for low wind power output events to improve the wind power dispatching capabilities of power systems and ensure reliable grid operation. Addressing this issue is crucial for achieving a low-carbon, sustainable power system.

[0040] To address the above technical issues, embodiments of the present disclosure provide a method for generating differential training for wind power low output events, which will be described in detail through one or more of the following embodiments.

[0041] The wind power low output event generation differential training method provided by the embodiment of the present disclosure can be applied to model training scenarios. The method can be executed by a wind power low output event generation differential training device, which can be implemented by software and / or hardware, and the device can be integrated into an electronic device. Among them, the electronic device can include but is not limited to mobile terminals such as smart phones, laptops, digital broadcast receivers, personal digital assistants (PDAs), tablet personal computers (Tablet PCs), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), wearable devices, etc., and fixed terminals such as digital televisions, desktop computers, smart home devices, etc.

[0042] Figure 1 A flow chart of a method for generating differential training of wind power low output events provided by the embodiment of the present disclosure, specifically including the following steps: Figure 1 Steps shown:

[0043] S101: Acquire operating data of at least one wind farm and numerical weather forecast data of at least one wind farm in the same period.

[0044] As can be understood, at least one wind farm is selected as the research subject from among the multiple wind farms within the study area. A wind farm is defined as an area where multiple wind turbines are concentrated and possesses a certain scale of power generation facilities. Subsequently, actual operating data for at least one wind farm is obtained. At least one wind farm can constitute a wind farm cluster within the area. This operating data typically includes wind turbine power, output, wind speed, wind direction, temperature, humidity, and other data, directly reflecting the wind farm's power generation performance and output variations. Simultaneously, numerical weather forecast data is obtained for at least one wind farm during the same time period. Numerical weather forecast data is meteorological data calculated and simulated based on atmospheric physics and dynamic equations, including predicted wind speed, air pressure, humidity, and other data. Data obtained during the same time period reflects conditions within the same time period, ensuring correlation and consistency between the data and facilitating analysis of the impact of weather forecasts on the wind farm's power generation capacity. Furthermore, since wind speed and direction directly affect a wind farm's power generation capacity, accurate weather forecast data is crucial for predicting low wind power output events.

[0045] It is understandable that after obtaining the operating data, the operating data can be pre-processed by data cleaning, etc. The pre-processing includes inspection and screening, elimination of erroneous data, and reasonable interpolation and correction.

[0046] In one possible embodiment, a wind farm cluster consisting of multiple wind farms in a certain province is selected as the research object. The operational data primarily includes power data, and the numerical weather forecast data for multiple wind farm sites primarily includes wind speed data. The power data is sampled over a one-year period with a 15-minute time resolution. This means that every 15 minutes is used as a sampling point, and power data for each wind farm is collected at each sampling point. This means that each sampling point can collect multiple power data points. The total power data collected from the multiple power data points can then be calculated. Furthermore, all total power data collected every 10 days can be used as a single sample. The numerical weather forecast data is also sampled over a one-year period. The weather forecast data and operational data are collected during the same period, with a 15-minute time resolution. The forecast period is 10 days in the future. On the 1st, the weather conditions for the next 10 days are predicted, and all wind speed data for the next 10 days is used as a single sample. The sampling interval for each sample in the operational data is 10 days, and the corresponding wind speed data predicted in the weather forecast data is also 10 days, meaning that the sampling parameters for the two are one-to-one corresponding.

[0047] S102: Construct a first training set based on the operational data and the numerical weather forecast data.

[0048] The first training set refers to a real data set related to wind power low output events; the first training set includes a first sample set and a first label set of the first sample set.

[0049] It is understood that, based on the above S101, a first training set is constructed based on the operating data of at least one wind farm and the weather forecast data of at least one wind farm. The first training set includes a first sample set and a first label set. The first sample set includes multiple real samples, each of which is a combination of operating data and weather forecast data. For example, a sample is "Wind farm output: 50MW, weather forecast wind speed: 10m / s." The first label set includes a label for each real sample. For example, the label can be the type of power fluctuation amplitude corresponding to a low wind power output event.

[0050] Optionally, a first training set is constructed based on the operational data and numerical weather forecast data, which can be specifically achieved through the following steps:

[0051] Calculate a first sequence of operating data; wherein the first sequence reflects the sum of operating data collected by at least one wind farm within a set time period; perform dimensionality reduction processing on the numerical weather forecast data to obtain a second sequence with the same dimension as the first sequence; fuse the first sequence and the second sequence to obtain a first sample set; identify at least one wind power low output event corresponding to the first sample set, determine a label for the at least one wind power low output event, and construct a first label set.

[0052] It is understandable that, taking 10 days of power data as a real sample as an example, the total power data of multiple power data collected from multiple wind farms in each sampling point for 10 days is calculated, and the total power data of multiple real samples constitutes the first sequence. The weather forecast data is subjected to dimensionality reduction processing to obtain a second sequence with the same dimension as the first sequence, and the first sequence and the second sequence are both one-dimensional. Specifically, the weather forecast data can be reduced in dimension by principal component analysis, and the data dimension after dimensionality reduction is 1-dimensional. Subsequently, the first sequence and the second sequence corresponding to the same time period are fused, such as splicing, to obtain the first sample set. The multiple real samples included in the first sample set are subjected to wind power low output event identification, and the label of the sample is determined according to the specific category identified to obtain the first label set. Specifically, with a sampling period of 10 days, the total power data of the power data of multiple wind farms collected every 15 minutes is calculated, that is, 15 minutes corresponds to one total power data. For example, 3 power data of 3 wind farms are collected (denoted as a, b, and c), and the sum of a, b, and c is calculated to obtain the total power data (denoted as d). d is the sampling data corresponding to one sampling point. There are 2 total power data in 30 minutes, and a large amount of total power data in 10 days. A large amount of total power data is counted to determine whether it meets the identification criteria of a duration of more than one day and an average power of less than 10% of the rated power. If this identification criteria is met, this sample is determined to be a true sample of a wind power low output event. The identification method of other wind power low output events is not limited. Subsequently, a wind power low output event sample set is constructed based on the first label set and the first sample set.

[0053] The at least one wind power low output event includes small fluctuations, medium fluctuations and large fluctuations classified by power fluctuation amplitude.

[0054] Understandably, based on the first sample set, using the power fluctuation amplitude of real samples as a screening criterion, the identified samples containing wind power low output events are divided into small fluctuations (often occurring in summer), medium fluctuations (often occurring in autumn), and large fluctuations (often occurring in winter or spring). Understandably, different power fluctuation amplitudes generally correspond to different seasons.

[0055] Optionally, identifying at least one wind power low output event corresponding to the first sample set, determining a label for at least one wind power low output event, and constructing a first label set can be specifically achieved through the following steps:

[0056] Wind power low output events are identified based on the total power data included in each sample in the first sample set; the wind power low output event corresponding to the first sample whose total power data is less than the first set threshold is determined as a small fluctuation; the wind power low output event corresponding to the second sample whose total power data is greater than the first set threshold and less than the second set threshold is determined as a medium fluctuation; the wind power low output event corresponding to the third sample whose total power data is greater than the second set threshold is determined as a large fluctuation; the small fluctuation, medium fluctuation or large fluctuation corresponding to each sample is encoded to construct a first label set.

[0057] It can be understood that based on the first sample set, the power fluctuation amplitude is divided according to the first set threshold and the second set threshold. Specifically, the first set threshold is 30%, the second set threshold is 60%, and the power fluctuation amplitude less than 30% is divided into small fluctuations, the power fluctuation amplitude greater than 30% and less than 60% is divided into medium fluctuations, and the power fluctuation amplitude greater than 60% is divided into large fluctuations. Subsequently, based on the division results, a wind power low output event condition label set is constructed through OneHot encoding. For example, small fluctuations are encoded as 001, medium fluctuations are encoded as 002, and large fluctuations are encoded as 003. Among them, the number of set thresholds can be set according to user needs and is not limited here. In addition, after the power fluctuation amplitude has not yet been divided into types according to the set thresholds, it can also be verified by referring to the seasons in which each power fluctuation amplitude usually occurs to ensure the accuracy of the power fluctuation amplitude division of wind power low output events.

[0058] S103: Train a pre-built event sample generation model according to the first training set and the acquired noise data, so as to generate a second training set according to the trained event sample generation model.

[0059] The second training set refers to a generated data set related to wind power low output events.

[0060] As can be understood, based on the above S102, the first training set (including the first sample set and the first label set) and the noise data are used to train the event sample generation model. The event sample generation model learns how to generate new training samples based on the first training set, thereby generating a second training set containing a large number of samples based on the noise data and the wind power low output event label set. The event sample generation model enables the expansion of wind power low output event samples, addresses the small sample size characteristic, and subsequently provides key input for the accurate prediction of wind power low output events.

[0061] Optionally, a pre-built event sample generation model is trained based on the first training set and the acquired noise data, so as to generate a second training set based on the trained event sample generation model. This can be specifically achieved through the following steps:

[0062] An event sample generation model for generating wind power low output event samples is constructed; wherein the event sample generation model includes a generator and a discriminator; the acquired noise data and the first label set included in the first training set are used as inputs of the generator to obtain a generated sample set output by the generator; the generated sample set and the first sample set included in the first training set are used as inputs of the discriminator to obtain a true or false judgment result of the sample output by the discriminator; when the generator and the discriminator reach a balanced state after adversarial training, a trained event sample generation model is obtained to generate a second training set based on the trained event sample generation model.

[0063] It is understandable that an event sample generation model for wind power low output events is constructed. The event sample generation model can be constructed based on a conditional generative adversarial network. The conditional generative adversarial network consists of two parts: a generator and a discriminator. Its essence is the game learning of the generator and the discriminator. The training goal of the generator is to make the probability distribution p of the generated sample G(z) G (z) is as close as possible to the probability distribution p of the true sample x data (z) is the same. The discriminator's training goal is to accurately distinguish whether its input is a real sample or a generated sample. After adversarial training, the generator and discriminator can reach a balanced state, making the distribution of artificially generated data gradually approach that of real data, making it indistinguishable from real data.

[0064] It is understandable that the noise data can be measurement noise. When the sensor measurement is inaccurate, the measured wind speed, temperature, humidity and other data may contain errors. Other possible noise data are not limited and can be determined according to user needs. Subsequently, the noise data and the first label set as the true label are used as inputs of the generator, and multiple first generated samples are output to construct a generated sample set. Subsequently, the generated sample set and the first sample set including the true sample are used as inputs of the discriminator, and the judgment result of each generated sample is output, and the judgment result includes whether the generated sample is true or false. When the discriminator and the generator reach a balanced state after at least one adversarial training, a trained discriminator and a trained generator are obtained, and then a trained event sample generation model is obtained. Subsequently, the noise data and the first label set are continued to be used as inputs of the trained generator, and multiple second generated samples are output to constitute a second training set, wherein the noise data can be the same as that during the generator training or different, and is not limited here.

[0065] In a possible embodiment, during the training of the event sample generation model, the learning rate of the generator can be set to 0.0001, the learning rate of the discriminator can be set to 0.0005, the batch size can be set to 16, and a total of 1000 epochs can be trained.

[0066] Among them, the generator includes at least one first multi-layer perceptron, a long short-term memory network and at least one first attention layer, at least one first multi-layer perceptron is connected sequentially, at least one first attention layer is connected through a residual connection, the long short-term memory network connects the last first multi-layer perceptron and the first first attention layer, the first first multi-layer perceptron receives noise data and a first label set as input, and the last residual connection outputs a generated sample set; the discriminator includes at least one second multi-layer perceptron and at least one second attention layer, at least one second multi-layer perceptron is connected sequentially, at least one second attention layer is connected through a residual connection, the last second attention layer is connected to the first second multi-layer perceptron, the first residual connection receives the first sample set and the generated sample set as input, and the last second multi-layer perceptron outputs the sample true or false judgment result.

[0067] For example, see Figure 2 , Figure 2 This is a structural diagram of an event sample generation model provided by an embodiment of the present disclosure. The generator is composed of at least one first multi-layer perceptron, a long short-term memory network, and at least one first attention layer (Self-Attention). Figure 2 As shown, the generator includes two first multi-layer perceptrons, a long short-term memory network and two first attention layers, wherein the two first multi-layer perceptrons are connected in sequence, the first first multi-layer perceptron receives the noise data and the first label set, and a long short-term memory network (LSTM) connects the second first multi-layer perceptron and the first first attention layer. The two first attention layers are connected through residual connections, which can be understood as convolutional neural networks, that is, each first attention layer is connected to a convolutional neural network, and the last convolutional neural network outputs the first generated sample. It is understandable that the number of first multi-layer perceptrons, long short-term memory networks and first attention layers is not limited. The discriminator consists of at least one second multi-layer perceptron and at least one second attention layer (Self-Attention), as shown in Figure 2 As shown, the discriminator includes two second multi-layer perceptrons and two second attention layers, wherein the two second attention layers are connected via a residual connection. The residual connection can also be regarded as a convolutional neural network, that is, the input of each second attention layer is the output of the convolutional neural network. The first convolutional neural network receives the first sample set and the first generated sample. The two second multi-layer perceptrons are connected sequentially, and the last second attention layer is connected to the first second multi-layer perceptron. The last second multi-layer perceptron outputs the true or false judgment result of the sample. It is understandable that the network structure of the first multi-layer perceptron and the second multi-layer perceptron can be the same or different.

[0068] S104: Train the pre-built event prediction model based on the first training set and the second training set to obtain a trained event prediction model.

[0069] Among them, the event prediction model is used to predict wind power low output events.

[0070] It is understandable that, based on the above S103, an event prediction model for predicting low wind power output events is established based on a meta-learning architecture, wherein the meta-learning architecture includes two modules: a base learner and a meta-learner. The base learner uses the CNN-ProbSparse and Attention-GRU models, and the meta-learner uses the XGBoost model. Subsequently, the event prediction model is trained based on a first training set as real samples and a second training set as generated samples. The event prediction model is used to predict whether a low wind power output event will occur in the future period and the power curve based on weather forecast data that predicts the weather conditions in the future period.

[0071] Optionally, a pre-built event prediction model is trained based on the first training set and the second training set to obtain a trained event prediction model, which can be specifically achieved through the following steps:

[0072] An event prediction model for predicting low wind power output events is constructed; wherein the event prediction model includes a base learner and a meta learner; the base learner is trained according to the second training set to obtain a trained base learner; the first sample set included in the first training set is used as the input of the trained base learner to obtain an event prediction result output by the base learner; and the meta learner is trained according to the event prediction result and the first sample set.

[0073] The training process of the event prediction model is as follows: First, the base learner is trained based on the second training set (generated samples) to obtain a trained base learner. Then, the first sample set (real samples) is input into the trained base learner to obtain preliminary event prediction results for each real sample output by the base learner. Finally, the event prediction results and the first sample set are combined to train the meta-learner to obtain a trained meta-learner.

[0074] Optionally, the base learner is trained according to the second training set to obtain a trained base learner, which can be specifically achieved through the following steps:

[0075] Identify at least one wind power low output event corresponding to the second sample set included in the second training set, and determine at least one generated sample corresponding to each wind power low output event; set a corresponding submodule for each wind power low output event; wherein the base learner includes at least one submodule; and train each submodule according to the at least one generated sample corresponding to each submodule.

[0076] It is understandable that at least one wind power low output event corresponding to the generated samples included in the second training set is identified using the average power and event duration as identification criteria. Subsequently, the power fluctuation amplitude is continued to be used as a screening condition to determine the second sample set corresponding to each power fluctuation amplitude, that is, the second training set is further divided into multiple second sample sets, wherein the power fluctuation amplitude includes the above-mentioned small fluctuations, medium fluctuations and large fluctuations. The specific identification process and screening process refer to the above embodiment and will not be repeated here. A corresponding submodule is set for the second sample set corresponding to each power fluctuation amplitude, wherein the base learner includes multiple submodules, or multiple base learners are directly set, and a corresponding base learner is set for the second sample set corresponding to each power fluctuation amplitude. Subsequently, a detailed description is given by taking the example of setting a submodule for each power fluctuation amplitude. For each submodule, the corresponding second sample set is used as input to train each submodule.

[0077] Optionally, the submodule includes a convolutional neural network model, an attention layer, multiple recurrent neural network models and multiple linear layers corresponding to the multiple recurrent neural network models. The attention layer connects the convolutional neural network model and the multiple recurrent neural network models. The convolutional neural network model receives at least one corresponding generated sample, and the multiple linear layers output event prediction results, wherein the event prediction results include occurrence prediction results and power prediction results. The occurrence prediction results are used to record whether a wind power low output event occurs, and the power prediction results are used to record the power changes of the wind power low output event.

[0078] For example, see Figure 3 , Figure 3A structural diagram of an event prediction model provided in an embodiment of the present disclosure is provided. The event prediction model includes a base model (base learner) and a meta-model (original learner). The base model includes CNN, Attention, 2 GRUs and 2 Linears. The input of CNN is TASK, which refers to the second sample set corresponding to a certain power fluctuation amplitude. For example, TASK1 refers to the second sample set corresponding to a small fluctuation, TASK2 refers to the second sample set corresponding to a medium fluctuation, and TASK3 refers to the second sample set corresponding to a large fluctuation. The output of CNN is the input of Attention, and the output of Attention is the input of 2 GRUs. Each GRU has a corresponding Linear, and the output of GRU is the input of Linear. A multi-task framework is used to construct an occurrence prediction decoder based on GRU, Linear and Focalloss and a power prediction decoder based on GRU, Linear and MSEloss. The occurrence prediction decoder is used to predict whether a wind power low output event occurs, and the power prediction decoder is used to generate a power curve when a wind power low output event occurs. After completing the training of the three base models corresponding to the power fluctuation amplitudes, real samples are used as the input of at least one base model, and the occurrence prediction result and power prediction result output by at least one base model are obtained. Subsequently, the occurrence prediction result and power prediction result output by each base model can be fused to obtain a fused prediction result. The input of the meta-learner is the fused prediction result and the first sample set as real samples. The output is an occurrence prediction result for recording whether the wind power low output event has occurred and a power prediction result for recording the power curve, among other event prediction results.

[0079] Optionally, the meta-learner is trained according to the event prediction result and the first sample set, which can be specifically implemented by the following steps:

[0080] At least one event prediction result output by at least one submodule included in the base learner is fused to obtain a fused prediction result; and the meta learner is trained according to the fused prediction result and the first sample set.

[0081] It is understandable that the event prediction results output by each base model are fused to obtain a fused prediction result. For example, taking three base models as an example, the three occurrence prediction results can be fused / joined together, the three power prediction results can be joined together, and finally the two parts can be joined together to obtain a fused prediction result. Alternatively, the occurrence prediction results and power prediction results output by each base model can be joined together first, and finally the three parts can be joined together to obtain a fused prediction result. The specific fusion order is not limited and can be determined according to user needs. Subsequently, the fused prediction results and the meta-model are used as inputs to the meta-model, and the meta-model outputs a final occurrence prediction result and a power prediction result.

[0082] In one possible embodiment, during event prediction model training, the second training set is conditionally divided into three sample sets: small fluctuations, medium fluctuations, and large fluctuations, as described above based on the power fluctuation amplitude. A base learner is trained for each sample set, with a learning rate of 0.0001, a batch size of 16, and 100 epochs. In the Focalloss loss function, alpha is 0.7 and beta is 2. XGBoost is used as the meta-learner model, with a learning rate of 0.001, n_estimators = 200, and max_depth = 6.

[0083] The embodiment of the present disclosure provides a method for training wind power low-output events with differentials. First, an event sample generation model is built based on a generative adversarial network, which realizes the effective generation of wind power low-output event samples and completes the expansion of wind power low-output event samples, thus coping with the small sample size of low-output events. Then, considering the "quantity-value" difference between the generated samples and the real samples, a wind power low-output event prediction model is established based on a meta-learning architecture to resolve the difference between the generated samples and the real samples. Through multi-task learning, a joint and accurate prediction of whether a wind power low-output event will occur and the power of the wind power low-output event is achieved. On the one hand, it provides an early warning for the power system and ensures the safe and stable operation of the power system. On the other hand, it also improves the wind power penetration rate. Secondly, the base model is pre-trained using generated samples, and the model is fine-tuned using real samples, which improves the generalization ability of the model and realizes the accurate prediction of wind power low-output events. In addition, the generative differential training prediction of wind power low-output events can be used in the optimization and scheduling of power systems and is applicable to any time scale and spatial scale.

[0084] Based on the above embodiments, Figure 4 A flow chart of a method for predicting low wind power output events provided by an embodiment of the present disclosure, specifically including the following steps: Figure 4 The following steps are shown:

[0085] S401: Obtain weather forecast data for a preset time period.

[0086] It is understandable that the preset period may be the next ten days, that is, after the training of the event prediction model is completed, weather forecast data for future events may be obtained, wherein the length of the preset period is not limited.

[0087] S402: Using weather forecast data as input to an event prediction model, predicting low wind power output events within a preset time period, and obtaining event prediction results.

[0088] Among them, the event prediction model is trained using the above-mentioned wind power low output event generation differential training method.

[0089] As can be understood, based on the above-described S401, weather forecast data is used as input to the event prediction model trained using the above-described wind power low output event generative differential training method. The event prediction model predicts wind power low output events for the next ten days, thereby obtaining event prediction results for the next ten days. The event prediction results include whether a wind power low output event will occur in the next ten days and the power at which the wind power low output event will occur. As can be understood, the specific prediction process of the event prediction model is described in the above-described embodiment and is not further described here.

[0090] The method for predicting low wind power output events provided by the present disclosure achieves accurate prediction of low wind power output events.

[0091] Based on the above embodiments, Figure 5 A flow chart of another wind power low output event generation differential training method provided by the embodiment of the present disclosure, specifically including the following steps: Figure 5 The following steps are shown:

[0092] 1) Obtain the actual operating data of each wind farm in the region, clean and preprocess the actual operating data, and obtain the numerical weather forecast data of each wind farm point in the same period; 2) Perform dimensionality reduction processing on the numerical weather forecast data; 3) Identify wind power low output events and construct a wind power low output event sample set; 4) Construct a wind power low output event condition label set; 5) Construct a wind power low output event sample generation model; 6) Generate wind power low output event samples based on the event sample generation model; 7) Establish a wind power low output event prediction model based on the meta-learning architecture; 8) Train the wind power low output event prediction model.

[0093] It is understandable that the specific implementation steps of the above 1)-8) can be found in the above embodiments and will not be repeated here.

[0094] Based on the above embodiments, Figure 6The structure diagram of a wind power low output event generation type differential training device provided by the embodiment of the present disclosure. The wind power low output event generation type differential training device provided by the embodiment of the present disclosure can execute the processing flow provided by the wind power low output event generation type differential training method embodiment, such as Figure 6 As shown, the apparatus 600 includes:

[0095] A first acquisition unit 601 is configured to acquire operating data of at least one wind farm and numerical weather forecast data of at least one wind farm in the same period;

[0096] A construction unit 602 is configured to construct a first training set based on the operating data and the numerical weather forecast data; wherein the first training set is a real data set related to wind power low output events;

[0097] A first training unit 603 is configured to train a pre-built event sample generation model based on the first training set and the acquired noise data, so as to generate a second training set based on the trained event sample generation model; wherein the second training set is a generated data set related to wind power low output events;

[0098] The second training unit 604 is used to train the pre-built event prediction model based on the first training set and the second training set to obtain a trained event prediction model; wherein the event prediction model is used to predict wind power low output events.

[0099] The first training set includes a first sample set and a first label set of the first sample set.

[0100] Optionally, the construction unit 602 is used to:

[0101] Calculating a first sequence of operating data; wherein the first sequence reflects the sum of operating data collected from at least one wind farm within a set period of time;

[0102] Perform dimensionality reduction on the numerical weather forecast data to obtain a second sequence with the same dimension as the first sequence;

[0103] Fusing the first sequence with the second sequence to obtain a first sample set;

[0104] At least one wind power low output event corresponding to the first sample set is identified, and a label of the at least one wind power low output event is determined to construct a first label set.

[0105] The at least one wind power low output event includes small fluctuations, medium fluctuations and large fluctuations classified by power fluctuation amplitude.

[0106] Optionally, the construction unit 602 is used to:

[0107] Identifying wind power low output events based on total power data included in each sample in the first sample set;

[0108] Determine the wind power low output event corresponding to the first sample whose total power data is less than the first set threshold as a small fluctuation;

[0109] The wind power low output event corresponding to the second sample where the total power data is greater than the first set threshold and less than the second set threshold is determined as a medium fluctuation;

[0110] Determine the wind power low output event corresponding to the third sample whose total power data is greater than the second set threshold as a large fluctuation;

[0111] The small fluctuation, medium fluctuation or large fluctuation corresponding to each sample is encoded to construct the first label set.

[0112] Optionally, the first training unit 603 is configured to:

[0113] Constructing an event sample generation model for generating wind power low output event samples; wherein the event sample generation model includes a generator and a discriminator;

[0114] Using the acquired noise data and the first label set included in the first training set as inputs of the generator, and obtaining a generated sample set output by the generator;

[0115] Using the generated sample set and the first sample set included in the first training set as inputs to the discriminator, and obtaining a true or false judgment result of the sample output by the discriminator;

[0116] When the generator and the discriminator reach a balanced state after adversarial training, a trained event sample generation model is obtained, and a second training set is generated according to the trained event sample generation model.

[0117] Optionally, the generator in the device 600 includes at least one first multilayer perceptron, a long short-term memory network, and at least one first attention layer, the at least one first multilayer perceptron is connected sequentially, the at least one first attention layer is connected via a residual connection, the long short-term memory network connects the last first multilayer perceptron and the first first attention layer, the first first multilayer perceptron receives the noise data and the first label set as input, and the last residual connection outputs the generated sample set;

[0118] The discriminator includes at least one second multi-layer perceptron and at least one second attention layer. The at least one second multi-layer perceptron is connected sequentially. The at least one second attention layer is connected via a residual connection. The last second attention layer is connected to the first second multi-layer perceptron. The first residual connection receives the first sample set and the generated sample set as input, and the last second multi-layer perceptron outputs the sample true or false judgment result.

[0119] Optionally, the second training unit 604 is configured to:

[0120] Construct an event prediction model for predicting wind power low output events; wherein the event prediction model includes a base learner and a meta learner;

[0121] Training the base learner according to the second training set to obtain a trained base learner;

[0122] Using the first sample set included in the first training set as the input of the trained base learner, and obtaining the event prediction result output by the base learner;

[0123] The meta-learner is trained based on the event prediction results and the first sample set.

[0124] Optionally, the second training unit 604 is configured to:

[0125] Identifying at least one wind power low output event corresponding to a second sample set included in the second training set, and determining at least one generated sample corresponding to each wind power low output event;

[0126] A corresponding submodule is set for each wind power low output event; wherein the base learner includes at least one submodule;

[0127] Each submodule is trained according to at least one generated sample corresponding to each submodule.

[0128] Optionally, the submodule of device 600 includes a convolutional neural network model, an attention layer, multiple recurrent neural network models and multiple linear layers corresponding to the multiple recurrent neural network models. The attention layer connects the convolutional neural network model and the multiple recurrent neural network models. The convolutional neural network model receives at least one corresponding generated sample, and the multiple linear layers output event prediction results, wherein the event prediction results include occurrence prediction results and power prediction results. The occurrence prediction results are used to record whether a wind power low output event occurs, and the power prediction results are used to record the power changes of the wind power low output event.

[0129] Optionally, the second training unit 604 is configured to:

[0130] performing fusion processing on at least one event prediction result output by at least one submodule included in the base learner to obtain a fusion prediction result;

[0131] The meta-learner is trained based on the fusion prediction results and the first sample set.

[0132] Figure 6 The model training device of the illustrated embodiment can be used to execute the technical solution of the above-mentioned method embodiment. Its implementation principle and technical effects are similar and will not be repeated here.

[0133] Based on the above embodiments, Figure 7 The schematic diagram of the structure of a wind power low output event prediction device provided by the embodiment of the present disclosure. The wind power low output event prediction device provided by the embodiment of the present disclosure can execute the processing flow provided by the wind power low output event prediction method embodiment, such as Figure 7 As shown, the prediction device 700 for wind power low output events includes:

[0134] The second acquisition unit 701 is used to acquire weather forecast data for a preset period of time;

[0135] The event prediction unit 702 is used to use weather forecast data as input to the event prediction model to predict the wind power low output event within a preset time period and obtain the event prediction result; wherein, the event prediction model is trained by the above-mentioned wind power low output event generation differential training method.

[0136] Figure 7 The device for predicting low wind power output events in the illustrated embodiment can be used to implement the technical solution of the above-mentioned method embodiment. Its implementation principle and technical effects are similar and will not be described in detail here.

[0137] Figure 8 This is a schematic diagram of the structure of the electronic device provided by the embodiment of the present disclosure. Figure 8 , which shows a schematic structural diagram of an electronic device 100 suitable for implementing the embodiments of the present disclosure. The electronic device 100 in the embodiments of the present disclosure may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), wearable electronic devices, and the like, as well as fixed terminals such as digital TVs, desktop computers, smart home devices, and the like. Figure 7 The electronic device shown is only an example and should not limit the functions and scope of use of the embodiments of the present disclosure.

[0138] like Figure 8 As shown, the electronic device 100 may include a processing device 101 (such as a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 102 or the program loaded from the storage device 108 to the random access memory (RAM) 103 to implement the wind power low output event generation formula differential training method and prediction method as an embodiment of the present disclosure. Various programs and data required for the operation of the electronic device 100 are also stored in the RAM 103. The processing device 101, ROM 102 and RAM 103 are connected to each other through a bus 104. An input / output (I / O) interface 105 is also connected to the bus 104.

[0139] Typically, the following devices may be connected to the I / O interface 105: an input device 106 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 107 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 108 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 109. The communication device 109 may allow the electronic device 100 to communicate with other devices wirelessly or by wire to exchange data. Figure 7 The electronic device 100 is shown with various devices, but it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed instead.

[0140] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart, thereby realizing the above-mentioned wind power low output event generation formula differential training method and prediction method. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 109, or installed from the storage device 108, or installed from the ROM 102. When the computer program is executed by the processing device 101, the above-mentioned functions defined in the method of the embodiment of the present disclosure are performed.

[0141] It should be noted that the computer-readable medium mentioned above in the present disclosure may be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or component, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, device, or component. In the present disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium may be transmitted using any suitable medium, including but not limited to wires, optical cables, RF (radio frequency), etc., or any suitable combination thereof.

[0142] In some embodiments, the client and server can communicate using any currently known or future developed network protocol, such as HTTP (HyperText Transfer Protocol), and can be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0143] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0144] Optionally, when the above one or more programs are executed by the electronic device, the electronic device may also execute other steps described in the above embodiments.

[0145] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages, or a combination thereof, including, but not limited to, object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the device computer, partially on the device computer, as a stand-alone software package, partially on the device computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the device computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0146] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the module, program segment, or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of the boxes in the block diagram and / or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0147] The units involved in the embodiments described in this disclosure may be implemented in software or hardware, wherein the name of a unit does not necessarily limit the unit itself.

[0148] The functions described above herein may be performed, at least in part, by one or more hardware logic components. For example, and without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chip (SOCs), complex programmable logic devices (CPLDs), and the like.

[0149] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0150] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or gateway that includes a series of elements includes not only those elements, but also other elements that are not explicitly listed, or also includes elements that are inherent to such process, method, article or gateway. In the absence of further restrictions, the elements defined by the statement "including a model training" do not exclude the presence of other identical elements in the process, method, article or gateway that includes the elements.

[0151] The foregoing description is intended only to provide specific embodiments of the present disclosure, intended to enable those skilled in the art to understand and implement the present disclosure. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure is not intended to be limited to the embodiments described herein, but rather to be construed in the broadest manner consistent with the principles and novel features disclosed herein.

Claims

1. A wind power low output event generation type differential training method, characterized in that: include: Obtaining operating data of at least one wind farm and numerical weather forecast data of the at least one wind farm during the same period; Constructing a first training set based on the operating data and the numerical weather forecast data; wherein the first training set is a real data set related to wind power low output events; Training a pre-built event sample generation model based on the first training set and the acquired noise data to generate a second training set based on the trained event sample generation model; wherein the second training set is a generated data set related to wind power low output events; The pre-built event prediction model is trained based on the first training set and the second training set to obtain the trained event prediction model; wherein the event prediction model is used to predict wind power low output events.

2. The method according to claim 1, characterized in that The first training set includes a first sample set and a first label set for the first sample set, and constructing the first training set according to the operation data and the numerical weather forecast data includes: Calculating a first sequence of the operating data; wherein the first sequence reflects the sum of the operating data collected from the at least one wind farm within a set period; Performing dimensionality reduction processing on the numerical weather forecast data to obtain a second sequence having the same dimension as the first sequence; fusing the first sequence and the second sequence to obtain the first sample set; At least one wind power low output event corresponding to the first sample set is identified, and a label of the at least one wind power low output event is determined to construct the first label set.

3. The method according to claim 2, characterized in that The at least one wind power low output event includes small fluctuations, medium fluctuations, and large fluctuations classified by power fluctuation amplitude. Identifying the at least one wind power low output event corresponding to the first sample set, determining a label for the at least one wind power low output event, and constructing the first label set include: Identifying wind power low output events based on total power data included in each sample in the first sample set; Determining the wind power low output event corresponding to the first sample whose total power data is less than the first set threshold as the small fluctuation; Determine the wind power low output event corresponding to the second sample, wherein the total power data is greater than the first set threshold and less than the second set threshold, as the medium amplitude fluctuation; Determining the low wind power output event corresponding to the third sample in which the total power data is greater than the second set threshold as the large fluctuation; The small fluctuation, the medium fluctuation, or the large fluctuation corresponding to each sample is encoded to construct the first label set.

4. The method according to claim 1, wherein The step of training a pre-built event sample generation model according to the first training set and the acquired noise data to generate a second training set according to the trained event sample generation model includes: Constructing an event sample generation model for generating wind power low output event samples; wherein the event sample generation model includes a generator and a discriminator; Using the acquired noise data and the first label set included in the first training set as inputs of the generator, to obtain a generated sample set output by the generator; Using the generated sample set and the first sample set included in the first training set as inputs to the discriminator, and obtaining a true or false sample judgment result output by the discriminator; When the generator and the discriminator reach a balanced state after adversarial training, a trained event sample generation model is obtained, so as to generate a second training set according to the trained event sample generation model.

5. The method according to claim 4, characterized in that The generator includes at least one first multilayer perceptron, a long short-term memory network, and at least one first attention layer. The at least one first multilayer perceptron is connected sequentially, and the at least one first attention layer is connected via a residual connection. The long short-term memory network connects the last first multilayer perceptron and the first first attention layer. The first first multilayer perceptron receives the noise data and the first label set as input, and the last residual connection outputs the generated sample set. The discriminator includes at least one second multi-layer perceptron and at least one second attention layer, the at least one second multi-layer perceptron is connected sequentially, the at least one second attention layer is connected via a residual connection, the last second attention layer is connected to the first second multi-layer perceptron, the first residual connection receives the first sample set and the generated sample set as input, and the last second multi-layer perceptron outputs the true or false judgment result of the sample.

6. The method according to claim 1, characterized in that The pre-built event prediction model is trained based on the first training set and the second training set to obtain the trained event prediction model, including: Constructing an event prediction model for predicting wind power low output events; wherein the event prediction model includes a base learner and a meta learner; Training the base learner according to the second training set to obtain a trained base learner; Using the first sample set included in the first training set as input to the trained base learner, and obtaining an event prediction result output by the base learner; The meta-learner is trained according to the event prediction result and the first sample set.

7. The method according to claim 6, characterized in that The step of training the base learner according to the second training set to obtain a trained base learner includes: Identifying at least one wind power low output event corresponding to the second sample set included in the second training set, and determining at least one generated sample corresponding to each wind power low output event; A corresponding submodule is set for each wind power low output event; wherein the base learner includes at least one submodule; Each submodule is trained according to at least one generated sample corresponding to each submodule.

8. The method according to claim 7, characterized in that The submodule includes a convolutional neural network model, an attention layer, multiple recurrent neural network models and multiple linear layers corresponding to the multiple recurrent neural network models. The attention layer connects the convolutional neural network model and the multiple recurrent neural network models. The convolutional neural network model receives at least one corresponding generated sample, and the multiple linear layers output the event prediction results. The event prediction results include occurrence prediction results and power prediction results. The occurrence prediction results are used to record whether a wind power low output event occurs, and the power prediction results are used to record the power changes of the wind power low output event.

9. The method according to claim 6, characterized in that The training of the meta-learner according to the event prediction result and the first sample set includes: performing fusion processing on at least one event prediction result output by at least one submodule included in the base learner to obtain a fusion prediction result; The meta-learner is trained according to the fusion prediction result and the first sample set.

10. A method for predicting low wind power output events, characterized in that: include: Get weather forecast data for a preset period; The weather forecast data is used as the input of the event prediction model to predict the wind power low output event within the preset time period to obtain an event prediction result; wherein, the event prediction model is trained by the wind power low output event generation differential training method described in any one of claims 1-9.

Images