Power prediction method and device for wind power plant, equipment and storage medium

Abstract

The embodiment of the invention discloses a power prediction method and device for a wind power plant, equipment and a storage medium, and relates to the technical field of wind power generation. The method comprises the following steps: acquiring an operation monitoring data sequence of a wind power plant in a historical time period; predicting a power fluctuation period of the wind power plant based on the operation monitoring data sequence to obtain a prediction period; dividing a historical time period based on the prediction period to obtain a plurality of historical sub-time periods, and performing power prediction on the generated power of the wind power plant at the future moment based on the operation monitoring data sequence in each historical sub-time period by using a power prediction model to obtain a plurality of candidate generated power at the future moment; the multiple candidate generation powers are fused, the target generation power of the wind power plant at the future moment is obtained, the method can adapt to the dynamic characteristics of the power change of the wind power plant, the power prediction model can timely capture the power fluctuation and accurately reflect the real power change rule, and then the power prediction accuracy is improved.

Description

Technical Field

[0001] The embodiments of the present application relate to the technical field of wind power generation, and in particular, to a method, device, equipment and storage medium for power prediction of a wind farm. Background Technique

[0002] Currently, the power prediction method of a wind farm usually collects and analyzes wind speed data at a fixed time period (such as every hour or every day, etc.), and combines a simple linear regression model for power prediction. For example, the average wind speed data within a period of time is obtained, and according to the pre-established wind speed-power correspondence curve, the target power generation power at a future moment is calculated.

[0003] However, the fixed-period data collection and analysis method cannot adapt to the dynamic characteristics of the power change of the wind farm, resulting in the power prediction model being unable to capture the power fluctuation in time, and only relying on wind speed data for prediction, ignoring the influence of other factors on the power generation power, resulting in the power prediction model being unable to accurately reflect the real power change law, and further making the prediction accuracy of the power low. Summary of the Invention

[0004] The embodiments of the present application provide a method, device, equipment and storage medium for power prediction of a wind farm, realizing the power prediction function of the wind farm to solve the problem of low prediction accuracy of power in the prior art.

[0005] In a first aspect, the embodiments of the present application provide a method for power prediction of a wind farm, and the method includes:

[0006] Obtain the operation monitoring data sequence of the wind farm in a historical time period;

[0007] Predict the power fluctuation period of the wind farm based on the operation monitoring data sequence to obtain a predicted period;

[0008] Divide the historical time period based on the predicted period to obtain a plurality of historical sub-time periods, and use the power prediction model to perform power prediction on the power generation power of the wind farm at a future moment based on the operation monitoring data sequence within each historical sub-time period to obtain a plurality of candidate power generation powers at the future moment, and one historical sub-time period corresponds to one candidate power generation power;

[0009] Fuse the plurality of candidate power generation powers to obtain the target power generation power of the wind farm at the future moment.

[0010] In the embodiments of the present application, first, a power fluctuation period is predicted based on the operation monitoring data sequence within a historical time period. The prediction period can be dynamically adjusted according to the operation monitoring data of the wind farm. For example, when the wind speed changes violently, a shorter prediction period is determined; when the wind speed is relatively stable, a longer prediction period is determined, so as to adapt to the dynamic characteristics of the power change in the wind farm. Then, based on the prediction period, the historical time period is divided into multiple historical sub-time periods, and the power prediction model is respectively used to perform power prediction based on the operation monitoring data sequence within each historical sub-time period, which can avoid the lag of fixed-period prediction, enable the power prediction model to capture power fluctuations in a timely manner, and thus improve the timeliness and reliability of the power prediction model. Moreover, in addition to wind speed data, the operation monitoring data sequence also includes other meteorological data, enabling the power prediction model to comprehensively consider the influence of multi-dimensional factors on the generated power, and further enabling the power prediction model to accurately reflect the real power change law and improve the prediction accuracy of the power prediction model. In addition, by fusing multiple candidate generated powers, the prediction results of each historical sub-time period can be comprehensively considered, further improving the prediction accuracy of the power.

[0011] In a second aspect, an embodiment of the present application provides a power prediction device for a wind farm. The device includes:

[0012] An acquisition module, configured to acquire the operation monitoring data sequence of the wind farm within a historical time period;

[0013] A period prediction module, configured to predict the power fluctuation period of the wind farm based on the operation monitoring data sequence to obtain a prediction period;

[0014] A power prediction module, configured to divide the historical time period based on the prediction period to obtain multiple historical sub-time periods, and use the power prediction model to perform power prediction on the generated power of the wind farm at a future moment based on the operation monitoring data sequence within each historical sub-time period to obtain multiple candidate generated powers at the future moment, and one historical sub-time period corresponds to one candidate generated power;

[0015] A fusion module, configured to fuse multiple candidate generated powers to obtain the target generated power of the wind farm at a future moment.

[0016] In a third aspect, an embodiment of the present application provides an electronic device. The electronic device includes:

[0017] At least one processor; and a memory communicatively connected to the at least one processor;

[0018] Wherein, the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the power prediction method for the wind farm according to any embodiment of the present application.

[0019] Fourthly, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the power prediction method for a wind farm according to any embodiment of the present application.

[0020] For the descriptions of the second, third, and fourth aspects in the present application, reference may be made to the detailed description of the first aspect; and for the beneficial effects described in the second, third, and fourth aspects, reference may be made to the analysis of the beneficial effects of the first aspect, which will not be elaborated here.

[0021] In the present application, the name of the above-mentioned power prediction device for a wind farm does not limit the device or functional module itself. In actual implementation, these devices or functional modules may appear under other names. As long as the functions of each device or functional module are similar to those of the present application and fall within the scope of the claims of the present application and their equivalent technologies.

[0022] These aspects or other aspects of the present application will be more clearly understood in the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0024] Figure 1 is a flowchart of the power prediction method for a wind farm provided by an embodiment of the present application;

[0025] Figure 2 is another flowchart of the power prediction method for a wind farm provided by an embodiment of the present application;

[0026] Figure 3 is a structural diagram of the power prediction device for a wind farm provided by an embodiment of the present application;

[0027] Figure 4 is a structural diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0028] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the following will clearly and completely describe the technical solutions in the embodiments of this application with reference to the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part rather than all of the embodiments of this application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without creative efforts shall fall within the scope of protection of this application.

[0029] It should be noted that the terms "first", "second", "target", and "original" in the specification, claims, and the above-mentioned drawings of this application are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of this application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "include", "have", and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily limit to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0030] Figure 1 is a flowchart of a power prediction method for a wind farm provided by an embodiment of this application. This embodiment can be applied to scenarios where the power generation of a wind farm needs to be predicted. A power prediction method for a wind farm provided by this embodiment can be executed by a power prediction device for a wind farm provided by an embodiment of this application, and this device can be implemented in software and / or hardware. In a specific embodiment, this power prediction device for a wind farm can be integrated in an electronic device. For example, this electronic device can be a computer or the like. The execution subject of this method can be an electronic device. Refer to Figure 1 , the power prediction method for a wind farm in this embodiment includes but is not limited to the following steps:

[0031] S110. Obtain the operation monitoring data sequence of the wind farm within a historical time period.

[0032] Among them, a wind farm is a place where multiple wind turbines are centrally arranged in a certain geographical area in a certain arrangement, and corresponding power transmission and transformation equipment, control equipment, and other auxiliary facilities are built to convert wind energy into electrical energy and achieve large-scale power generation.

[0033] The historical time period is a certain continuous time range in the past, which can be preset according to the actual business needs of the wind farm.

[0034] The operation monitoring data sequence is a sequentially ordered data set formed by recording the operation status of a wind farm within a historical time period. Exemplarily, the operation monitoring data sequence may include a meteorological data sequence of the area where the wind farm is located within the historical time period, a power generation power sequence of the wind farm within the historical time period, etc.

[0035] Specifically, when it is necessary to predict the power generation power of a wind farm, the operation monitoring data sequence of the wind farm within the historical time period can be obtained. That is, by installing sensors at key parts of each wind turbine generator and using these sensors to monitor the operation monitoring data of the wind farm in real time, such as meteorological data such as wind speed, direction, air density, temperature, air pressure, and humidity, as well as the power generation power of the wind farm, and then recording these data and the corresponding timestamps.

[0036] Among them, the determination process of the historical time period is as follows: In one implementation, the time period within a preset time range before the current moment can be determined as the historical time period, where the preset time range is a preset duration data, such as 15 minutes, 30 minutes, or 1 hour, etc.; in another implementation, the historical time period can be determined according to the preset reporting duration in the actual business requirements, where the preset reporting duration is the preset time in advance for outputting the power generation power at a future moment, such as 1 minute, 5 minutes, 10 minutes, or 30 minutes, etc. Here, the power generation power is the predicted power. For example, if the current moment is 14:00 and the preset reporting duration is 30 minutes, then the historical time period is from 13:30 to 14:00. At this time, the future moment is 14:30, that is, the operation monitoring data sequence from 13:30 to 14:00 is used to predict the power generation power of the wind farm at 14:30, and the predicted power generation power is output at 14:00, that is, the power generation power at 14:30 is output at 14:00.

[0037] Optionally, obtaining the operation monitoring data sequence of the wind farm within the historical time period includes: the initial operation monitoring data sequence of the wind farm within the historical time period can be obtained; outlier detection is performed on the initial operation monitoring data sequence, and the outliers in the initial operation monitoring data sequence are removed to obtain an intermediate operation monitoring data sequence; the missing values in the intermediate operation monitoring data sequence are filled to obtain the operation monitoring data sequence.

[0038] Specifically, an autoencoder can be used for outlier identification, that is, the encoder and decoder of the autoencoder can be trained using the normal operation monitoring data sequence to construct the feature space of the normal operation monitoring data sequence. At this time, if the operation monitoring data sequence includes multiple types of data, one type of data corresponds to one feature space. Then, each data in the initial operation monitoring data sequence is input into the encoder to obtain the feature vector corresponding to each data, and the deviation degree between the feature vector and the corresponding feature space is calculated. If the deviation degree is greater than the set deviation threshold, it is determined that the data is an outlier; then the outliers in the initial operation monitoring data sequence are removed to obtain the intermediate operation monitoring data sequence. The autoencoder among them is a neural network structure designed to encode and compress the input data through the encoder and then restore the original input data as much as possible through the decoder.

[0039] After that, a method combining time series difference and machine learning can be used for missing value filling, that is, according to the trend and seasonal characteristics of the time series, it can be determined which time points in the intermediate operation monitoring data sequence have missing values. Secondly, the time series difference method (such as linear interpolation, spline interpolation, and moving average difference, etc.) is used to preliminarily fill the missing values to obtain the initial filled data sequence, so as to utilize the continuity and local trend characteristics of the time series itself to quickly fill some missing situations that are easier to handle, and at the same time provide a relatively complete data basis for the subsequent machine learning model; then the initial filled data sequence is input into the trained machine learning model (such as random forest, etc.) to predict the missing values, and then the missing values predicted by the machine learning model are filled into the corresponding missing positions in the initial filled data sequence to obtain the operation monitoring data sequence, so as to restore the true value of the data as much as possible.

[0040] In the embodiment of the present application, through outlier identification and missing value filling, the data quality of the operation monitoring data sequence can be guaranteed, the influence of data error on the power prediction result is reduced, and thus the prediction accuracy of the power is improved.

[0041] S120. Predict the power fluctuation period of the wind farm based on the operation monitoring data sequence to obtain the prediction period.

[0042] Among them, the power fluctuation period is used to characterize the law of the power generation power of the wind farm changing with time. The prediction period is used to characterize the scale of the time window of the operation monitoring data sequence that should be used for power prediction.

[0043] Specifically, after obtaining the operation monitoring data sequence within the historical time period, the power fluctuation period of the wind farm can be predicted based on the operation monitoring data sequence to obtain the prediction period. For example, different power fluctuation periods can be set in advance according to the wind speed variation range to obtain a wind speed period relationship table. At this time, the wind speed period relationship table is used to save the relationship between the wind speed variation range and the power fluctuation period. That is, if the wind speed changes drastically, it will cause the power generation of the wind farm to fluctuate greatly. At this time, the corresponding power fluctuation period is set to be shorter to track the power change in time; if the wind speed is relatively stable, it will cause the power generation of the wind farm to fluctuate less. At this time, the corresponding power fluctuation period is set to be longer to improve the power prediction efficiency; then based on the wind speed variation in the operation monitoring data sequence, the wind speed period relationship table is queried to obtain the corresponding power fluctuation period, that is, the prediction period, so as to determine the time window that should be used for subsequent power prediction.

[0044] S130. Divide the historical time period based on the prediction cycle to obtain multiple historical sub-time periods, and use the power prediction model to predict the power generation of the wind farm at a future moment based on the operation monitoring data sequence in each historical sub-time period to obtain multiple candidate power generation powers at the future moment.

[0045] The historical sub-time period is a relatively small time interval obtained by further dividing the historical time period based on the forecast period.

[0046] The power prediction model is a neural network model pre-trained using the historical operation monitoring data sequence of the wind farm. It can capture the relationship between the operation monitoring data sequence of the wind farm and the power generation, and is used to predict the power generation of the wind farm at future times. Exemplarily, the power prediction model can be a multivariate linear regression model or an extreme learning machine model, etc.

[0047] The candidate power generation power is a power value predicted by the power prediction model based on the operation monitoring data sequence in the historical sub-time period, and one historical sub-time period corresponds to one candidate power generation power.

[0048] Specifically, after obtaining the prediction cycle, the historical time period can be divided based on the prediction cycle to obtain multiple historical sub-time periods. For example, if the historical time period is known to be 13:30 to 14:00 and the prediction cycle is 5 minutes, 6 historical sub-time periods can be obtained, namely 13:30 to 13:35, 13:35 to 13:40, 13:40 to 13:45, 13:45 to 13:50, 13:50 to 13:55, and 13:55 to 14:00.

[0049] Next, a pre-trained power prediction model can be obtained, and for the current historical sub-period among multiple historical sub-periods, the operation monitoring data sequence within the current historical sub-period is input into the power prediction model. At this time, the power prediction model uses the relationship between the learned operation monitoring data sequence and the generated power to predict the generated power of the wind farm at a future moment, that is, the candidate generated power. Then, each historical sub-period is traversed to obtain the candidate generated power predicted based on the operation monitoring data sequence within each historical sub-period, that is, one historical sub-period corresponds to one candidate generated power.

[0050] S140. Fuse multiple candidate generated powers to obtain the target generated power of the wind farm at a future moment.

[0051] Among them, the target generated power is the predicted value of the generated power of the wind farm at a future moment.

[0052] Specifically, after obtaining multiple candidate generated powers, multiple candidate generated powers can be fused to obtain the target generated power of the wind farm at a future moment. For example, the average value of multiple candidate generated powers can be calculated to fuse multiple candidate generated powers to obtain the target generated power.

[0053] The technical solution of the embodiment of the present application first predicts the power fluctuation period based on the operation monitoring data sequence within the historical period, and can dynamically adjust the prediction period according to the operation monitoring data of the wind farm. For example, when the wind speed changes violently, the prediction period is determined to be shorter, and when the wind speed is relatively stable, the prediction period is determined to be longer, so as to adapt to the dynamic characteristics of the power change of the wind farm. Then, based on the prediction period, the historical period is divided into multiple historical sub-periods, and the power prediction model is respectively used to perform power prediction based on the operation monitoring data sequence within each historical sub-period, which can avoid the lag of fixed-period prediction, so that the power prediction model can capture power fluctuations in a timely manner, thereby improving the timeliness and reliability of the power prediction model. Moreover, the operation monitoring data sequence includes other meteorological data in addition to the wind speed data, so that the power prediction model can comprehensively consider the influence of multi-dimensional factors on the generated power, and further enables the power prediction model to accurately reflect the real power change law, improving the prediction accuracy of the power prediction model. In addition, fusing multiple candidate generated powers can comprehensively consider the prediction results of each historical sub-period, further improving the prediction accuracy of the power.

[0054] Next, a power prediction method for a wind farm provided by an embodiment of the present application is further described. Figure 2 It is another process schematic diagram of the power prediction method for a wind farm provided by an embodiment of the present application. The embodiment of the present application is optimized on the basis of the above embodiments. See Figure 2, the method of this embodiment includes but is not limited to the following steps:

[0055] S210. Obtain the operation monitoring data sequence of the wind farm within the historical time period.

[0056] Optionally, the operation monitoring data sequence includes a meteorological data sequence, a wind turbine operation data sequence, and a power generation sequence. Among them, the meteorological data sequence may include a wind speed data sequence, a wind direction data sequence, an air density data sequence, a temperature data sequence, a pressure data sequence, and a temperature data sequence, etc. in the area where the wind farm is located within the historical time period; the wind turbine operation data sequence may include a wind turbine blade angle data sequence, a wind turbine rotation speed data sequence, and an equipment failure status data sequence, etc. of the wind turbines in the wind farm within the historical time period; the power generation sequence includes the power generation at each moment of the wind farm within the historical time period.

[0057] Specifically, the meteorological data sequence, the wind turbine operation data sequence, and the power generation sequence of the wind farm within the historical time period can be obtained.

[0058] S220. Calculate the power fluctuation index of the power generation sequence according to time windows of different scales, and obtain at least two power fluctuation indexes.

[0059] Among them, the power fluctuation index is used to measure the fluctuation of the power generation of the wind farm within a certain time range, and can intuitively reflect the stability, fluctuation degree, and change characteristics of the power generation, etc.

[0060] Specifically, after obtaining the operation monitoring data sequence, the power fluctuation index of the power generation sequence can be calculated according to time windows of different scales, that is, calculate the standard deviation or change rate, etc. of the power generation within time windows of different scales, which are used to characterize the change amplitude, and no specific limitation is made here. In this way, the power fluctuation indexes of the wind farm within time windows of different scales are obtained. The time windows of different scales here may include 5 minutes, 10 minutes, 15 minutes, 20 minutes, and 30 minutes, etc., and can be adjusted according to the actual situation.

[0061] S230. Input at least two power fluctuation indexes, the meteorological data sequence, and the wind turbine operation data sequence into the periodic prediction model to obtain the prediction period.

[0062] Among them, the periodic prediction model is a pre-trained neural network model, which can capture the relationship between the power fluctuation and the prediction period, so as to realize the adaptive period adjustment, and is used to predict what scale of time window of the operation monitoring data sequence should be used for power prediction. Exemplarily, the periodic prediction model can be a gradient boosting decision tree.

[0063] Optionally, the historical power fluctuation index of the wind farm, as well as the corresponding historical meteorological data sequence and historical wind turbine operation data sequence, can be used to train the cycle prediction model until the maximum number of iterations is reached, so as to obtain the optimal cycle prediction model. At this time, the optimal cycle prediction model can capture the relationship between power fluctuation and prediction cycle. For example, when the wind speed changes violently and the power fluctuation index is large, the cycle prediction model outputs a shorter prediction cycle (such as 5 minutes) to more timely track the power change; when the wind speed is relatively stable and the power fluctuation index is small, the cycle prediction model outputs a longer prediction cycle (such as 1 hour) to improve the prediction efficiency of power.

[0064] Specifically, after obtaining at least two power fluctuation indexes, a pre-trained cycle prediction model can be obtained. Then, the at least two power fluctuation indexes, meteorological data sequence and wind turbine operation data sequence are input into the cycle prediction model. At this time, the cycle prediction model can utilize the relationship between power fluctuation and prediction cycle that has been learned to output the prediction cycle.

[0065] S240. Divide the historical time period based on the prediction cycle to obtain multiple historical sub-time periods.

[0066] S250. Determine the indirect feature data sequence that affects the generated power according to the meteorological data sequence and the wind turbine operation data sequence.

[0067] Among them, the indirect feature is a new feature that affects the generated power obtained by performing feature engineering processing on the meteorological data and the wind turbine operation data; the indirect feature data sequence is a set formed by arranging the indirect feature data within the historical time period in chronological order.

[0068] Specifically, after obtaining multiple historical sub-time periods, feature engineering processing can be performed on the meteorological data sequence and the wind turbine operation data sequence. Specifically, the indirect factors that affect the generated power within the area where the wind farm is located, that is, the indirect features, such as the air quality passing through the wind turbine blades per unit time, can be determined according to the historical operation monitoring data sequence of the wind farm; then the meteorological data sequence and the wind turbine operation data sequence are processed to obtain the data sequence of this indirect feature within the historical time period, that is, the indirect feature data sequence. For example, the product of the wind speed data and the air density data at the corresponding moment is calculated to obtain the air quality passing through the wind turbine blades per unit time at the corresponding moment.

[0069] S260. Input the meteorological data sequence, wind turbine operation data sequence and indirect feature data sequence within each historical sub-time period into the power prediction model respectively to obtain multiple candidate generated powers.

[0070] Specifically, after obtaining the indirect feature data sequence, for each current historical sub-period within each historical sub-period, the meteorological data sequence, the wind turbine operation data sequence, and the indirect feature data sequence within the current historical sub-period are input into the power prediction model. At this time, the power prediction model uses the relationship between the learned operation monitoring data sequence and the generated power to predict the generated power of the wind farm at a future moment, that is, the candidate generated power. Then, each historical sub-period is traversed to obtain the candidate generated power predicted based on the meteorological data sequence, the wind turbine operation data sequence, and the indirect feature data sequence within each historical sub-period, that is, multiple candidate generated powers are obtained.

[0071] Optionally, the training process of the power prediction model is as follows, including Sa1 - Sa4:

[0072] Sa1. Obtain sample data.

[0073] Among them, the sample data includes a historical meteorological data sequence, a historical wind turbine operation data sequence, a historical generated power sequence, a historical indirect feature data sequence, and a generated power label; the generated power label is the generated power of the wind farm at the to-be-predicted moment corresponding to the sample data.

[0074] Sa2. Calculate the power fluctuation index of the historical generated power sequence according to time windows of different scales to obtain at least two historical power fluctuation indexes, and input the at least two historical power fluctuation indexes, the historical meteorological data sequence, and the historical wind turbine operation data sequence into the period prediction model to obtain a period label.

[0075] Among them, the historical power fluctuation index is used to measure the fluctuation of the historical generated power sequence; the period label is used to characterize which historical meteorological data sequence, historical wind turbine operation data sequence, and historical indirect feature data sequence within a time window of what scale should be used to predict the generated power at the to-be-predicted moment.

[0076] Sa3. Divide the sample data based on the period label to obtain multiple sample sub-data.

[0077] Among them, the time period corresponding to each sample sub-data is within the time period corresponding to the sample data, and the generated power label in each sample sub-data is the same as the generated power label in the sample data.

[0078] Sa4. Input the historical meteorological data sequence, the historical wind turbine operation data sequence, and the historical indirect feature data sequence in the sample sub-data into the initial power prediction model, and use the generated power label in the corresponding sample sub-data to guide the training output of the initial power prediction model, and train and optimize the initial power prediction model to obtain the power prediction model.

[0079] Among them, the initial power prediction model is a model framework that has not been trained yet and is used to learn the relationship between the operation monitoring data sequence of the wind farm and the generated power to achieve the power prediction function. Exemplarily, the initial power prediction model is an extreme learning machine model.

[0080] Specifically, in one implementation, the historical meteorological data sequence, the historical wind turbine operation data sequence, and the historical indirect feature data sequence in the sample sub-data can be input into the initial power prediction model to obtain the power training value; calculate the loss value between the power training value and the generated power label in the corresponding sample sub-data; use the backpropagation algorithm to train and optimize the initial power prediction model to minimize the loss value and obtain the power prediction model. That is, using the backpropagation algorithm, according to the gradient information of the loss function, update and adjust the input weights and hidden layer biases of the initial power prediction model to minimize the loss value, so that the power prediction model can gradually learn the relationship between the operation monitoring data sequence of the wind farm and the generated power.

[0081] In another implementation, the historical meteorological data sequence, the historical wind turbine operation data sequence, and the historical indirect feature data sequence in the sample sub-data can be input into the initial power prediction model to obtain the power training value; calculate the loss value between the power training value and the generated power label in the corresponding sample sub-data; take the loss value as the fitness value of the particle, and use the particle swarm optimization algorithm to train and optimize the initial power prediction model to obtain the power prediction model. Among them, the particle swarm optimization algorithm is an optimization algorithm based on swarm intelligence, which simulates the group behaviors such as foraging and migration of bird flocks or fish schools to find the optimal solution. In the particle swarm optimization algorithm, each possible solution to the optimization problem is regarded as a particle. In the embodiments of the present application, the particle represents the input weights and hidden layer biases of the initial power prediction model, that is, each particle represents a set of input weights and hidden layer biases; the fitness function is used to measure the quality of the solution represented by the particle. In the embodiments of the present application, the loss value is taken as the fitness value of the particle, that is, the fitness function value.

[0082] Specifically, randomly initialize the position and velocity of each particle, and based on the fitness value of each particle, determine the individual optimum of each particle and the global optimum of the entire particle swarm. Then, use the update formula of the particle swarm optimization algorithm to continuously adjust the position and velocity of the particle based on the individual optimum of each particle and the global optimum of the entire particle swarm to find the optimal input weights and hidden layer thresholds. When the preset termination condition is met (such as the fitness value converges or the maximum number of iterations is reached), the optimal power prediction model is obtained.

[0083] In the embodiments of the present application, the initial power prediction model is trained and optimized by the particle swarm optimization algorithm, which can quickly converge to the optimal solution, effectively reducing the time cost consumed in the optimization process, and can avoid complex gradient calculations, simplifying the optimization process, thereby improving the optimization efficiency of the power prediction model. At the same time, it can search for the true optimal solution more comprehensively in the entire solution space, avoiding being trapped in local optima, thereby enhancing the learning and prediction capabilities of the power prediction model for complex data and improving the prediction accuracy of the power prediction model.

[0084] In the embodiments of the present application, the initial power prediction model is trained through a large number of historical operation monitoring data sequences of the wind farm, enabling the power prediction model to more accurately discover the pattern between the input and output, thereby improving the prediction accuracy of the power prediction model. The sample data is divided into multiple sample sub-data through the periodic labels predicted by the periodic prediction model, and model training is performed based on the multiple sample sub-data, which can comprehensively consider the dynamic characteristics of the power generation of the wind farm during model training, further improving the prediction accuracy of the power prediction model.

[0085] S270. Perform a rationality check on each candidate generated power to obtain the check result corresponding to the candidate generated power.

[0086] Among them, the check result is the result obtained by performing a rationality check on the candidate generated power, which may include a successful check and a failed check.

[0087] Specifically, after obtaining the candidate generated power, a rationality check can be performed on each candidate generated power based on a preset power range. That is, if the candidate generated power is within the preset power range, it indicates that the candidate generated power is reasonable, and at this time, the check result of the candidate generated power is determined to be a successful check; if the candidate generated power is not within the preset power range, it indicates that the candidate generated power is unreasonable, and at this time, the check result of the candidate generated power is determined to be a failed check. The preset power range can be determined in advance according to the rated power of the wind turbines in the wind farm and the historical operation monitoring data sequence of the wind farm, and is used to represent the reasonable power generation range of the wind farm.

[0088] S280. Perform weighted fusion on the candidate generated powers with a successful check result to obtain an intermediate generated power.

[0089] Among them, the intermediate generated power is the power obtained by performing weighted fusion on the candidate generated powers.

[0090] Specifically, after obtaining the verification results of each candidate power generation, the weights corresponding to each candidate power generation with successful verification can be determined. That is, the weights can be determined according to the time distance between the historical sub-time period corresponding to the candidate power generation and the future moment. For example, if the historical sub-time period corresponding to the candidate power generation is closer to the future moment, the weight corresponding to the candidate power generation is set larger; if the historical sub-time period corresponding to the candidate power generation is farther from the future moment, the weight corresponding to the candidate power generation is set smaller, and the sum of the weights corresponding to each candidate power generation with successful verification is 1. Then, based on the weights corresponding to each candidate power generation with successful verification, the weighted sum of these candidate power generations with successful verification is calculated to obtain the intermediate power generation.

[0091] S290. Correct the intermediate power generation to obtain the target power generation of the wind farm at the future moment.

[0092] Specifically, in one implementation, the intermediate power generation can be corrected based on the historical power generation of the wind farm. That is, the average value of the power generation within a preset duration before the future moment of the wind farm can be calculated. The preset duration is the preset duration data and can be adjusted according to the actual situation. Then, the power deviation value between the intermediate power generation and the average value is calculated. Here, the power deviation value is the difference between the intermediate power generation and the average value, which is used to represent the deviation degree between the intermediate power generation and the average value. Then, the intermediate power generation is corrected based on the power deviation value to obtain the target power generation. That is, the relationship between the power deviation value and the power adjustment value can be preset to obtain a set relationship table. Here, the power adjustment value can be a positive number or a negative number. Thus, based on the power deviation value, the set relationship table is queried to obtain the corresponding power adjustment value, and the sum of the intermediate power generation and the power adjustment value is calculated to obtain the target power generation.

[0093] In another implementation, the intermediate power generation can be corrected based on the meteorological forecast data and grid load demand of the wind farm at the future moment. That is, the meteorological forecast data and grid load demand of the wind farm at the future moment are obtained, and the intermediate power generation is corrected based on the meteorological forecast data and grid load demand to obtain the target power generation. That is, if the wind speed data in the meteorological forecast data is large, or the grid load demand is high, the intermediate power generation can be increased to obtain the target power generation. At this time, the increased value can be preset according to the actual situation to meet the actual needs of power dispatching.

[0094] Optionally, the intermediate power generation can be corrected based on the historical power generation of the wind farm first to obtain the initial corrected power, and then the initial corrected power can be corrected based on the meteorological forecast data and grid load demand of the wind farm at the future moment to obtain the target power generation.

[0095] In the embodiments of the present application, a method for correcting the historical power generation power of a wind farm and a method for correcting based on the meteorological forecast data and grid load demand of the wind farm at a future moment can adaptively adjust the power prediction value according to the actual situation, making the target power generation power closer to the actual power value of the wind farm at the future moment, thereby improving the accuracy of power prediction.

[0096] Optionally, after obtaining the candidate power generation powers, rationality verification can be performed on each candidate power generation power to obtain the verification result corresponding to the candidate power generation power; the candidate power generation powers with verification failure in the verification result are corrected to obtain the corrected candidate power generation powers; and all candidate power generation powers are weighted and fused to obtain the target power generation power.

[0097] The technical solution of the embodiments of the present application can obtain the operation monitoring data sequence of the wind farm in the historical time period. Secondly, the power fluctuation index of the power generation power sequence is calculated according to time windows of different scales to obtain at least two power fluctuation indexes, and the at least two power fluctuation indexes, the meteorological data sequence, and the fan operation data sequence are input into the periodic prediction model to obtain the prediction period. By analyzing the power fluctuation index to adaptively adjust the prediction period, dynamic tracking of the power change of the wind farm can be realized, thereby avoiding the lag of fixed-period prediction, enabling the power prediction model to capture power fluctuations in a timely manner, and thus improving the timeliness and reliability of the power prediction model. Then, based on the prediction period, the historical time period is divided into multiple historical sub-time periods, and according to the meteorological data sequence and the fan operation data sequence, the indirect characteristic data sequence affecting the power generation power is determined, which can fully explore various factors affecting the power generation power of the wind farm and their correlation relationships. Then, the meteorological data sequence, the fan operation data sequence, and the indirect characteristic data sequence in each historical sub-time period are respectively input into the power prediction model to obtain multiple candidate power generation powers, which can comprehensively consider the synergistic effect of multiple parameters. Compared with the power prediction method that only relies on a single or a few parameters, the power prediction model of the embodiments of the present application can more truly reflect the change law of the power generation power of the wind farm, making the prediction result closer to the actual power value, thereby improving the determination accuracy of the candidate power generation power. After that, rationality verification is performed on each candidate power generation power to obtain the verification result corresponding to the candidate power generation power, and the candidate power generation powers with successful verification in the verification result are weighted and fused to obtain the intermediate power generation power. Only the candidate power generation powers with successful verification are weighted and fused, which can reduce the influence of incorrect power data and provide an accurate data basis for determining the target power generation power in the future. Finally, the intermediate power generation power is corrected to obtain the target power generation power of the wind farm at the future moment, which can further optimize the power prediction result and further improve the determination accuracy of the target power generation power, thereby improving the accuracy of power prediction.

[0098] Figure 3 is a schematic structural diagram of a power prediction device for a wind farm provided by an embodiment of the present application. Referring to Figure 3 , the power prediction device for the wind farm may include:

[0099] An acquisition module 310, configured to acquire an operation monitoring data sequence of the wind farm within a historical time period;

[0100] A period prediction module 320, configured to predict a power fluctuation period of the wind farm based on the operation monitoring data sequence to obtain a predicted period;

[0101] A power prediction module 330, configured to divide the historical time period based on the predicted period to obtain a plurality of historical sub-time periods, and use a power prediction model to perform power prediction on the power generation power of the wind farm at a future moment based on the operation monitoring data sequence within each historical sub-time period, to obtain a plurality of candidate power generation powers at the future moment, where one historical sub-time period corresponds to one candidate power generation power;

[0102] A fusion module 340, configured to fuse a plurality of candidate power generation powers to obtain a target power generation power of the wind farm at a future moment.

[0103] In one embodiment, the operation monitoring data sequence includes a meteorological data sequence, a fan operation data sequence, and a power generation power sequence. The period prediction module 320 is specifically configured to:

[0104] Calculate power fluctuation indexes of the power generation power sequence according to time windows of different scales to obtain at least two power fluctuation indexes;

[0105] Input the at least two power fluctuation indexes, the meteorological data sequence, and the fan operation data sequence into a period prediction model to obtain a predicted period.

[0106] In one embodiment, the power prediction module 330 uses a power prediction model to perform power prediction on the power generation power of the wind farm at a future moment based on the operation monitoring data sequence within each historical sub-time period, to obtain a plurality of candidate power generation powers at the future moment, including:

[0107] Determine an indirect feature data sequence affecting the power generation power according to the meteorological data sequence and the fan operation data sequence;

[0108] Input the meteorological data sequence, the fan operation data sequence, and the indirect feature data sequence within each historical sub-time period into the power prediction model respectively to obtain a plurality of candidate power generation powers.

[0109] In one embodiment, the training process of the power prediction model in the power prediction module 330 is as follows:

[0110] Obtain sample data, where the sample data includes a historical meteorological data sequence, a historical wind turbine operation data sequence, a historical power generation sequence, a historical indirect feature data sequence, and a power generation label, and the power generation label is the power generation of the wind farm at the to-be-predicted moment corresponding to the sample data;

[0111] Calculate the power fluctuation index of the historical power generation sequence according to time windows of different scales to obtain at least two historical power fluctuation indexes, and input the at least two historical power fluctuation indexes, the historical meteorological data sequence, and the historical wind turbine operation data sequence into the periodic prediction model to obtain a periodic label;

[0112] Based on the periodic label, divide the sample data to obtain multiple sample sub-data;

[0113] Input the historical meteorological data sequence, the historical wind turbine operation data sequence, and the historical indirect feature data sequence in the sample sub-data into the initial power prediction model, and use the power generation label in the corresponding sample sub-data to guide the training output of the initial power prediction model, and train and optimize the initial power prediction model to obtain a power prediction model.

[0114] In one embodiment, the power prediction module 330 inputs the historical meteorological data sequence, the historical wind turbine operation data sequence, and the historical indirect feature data sequence in the sample sub-data into the initial power prediction model, and uses the power generation label in the corresponding sample sub-data to guide the training output of the initial power prediction model, and trains and optimizes the initial power prediction model to obtain a power prediction model, including:

[0115] Input the historical meteorological data sequence, the historical wind turbine operation data sequence, and the historical indirect feature data sequence in the sample sub-data into the initial power prediction model to obtain a power training value;

[0116] Calculate the loss value between the power training value and the power generation label in the corresponding sample sub-data;

[0117] Use the loss value as the fitness value of the particle, and use the particle swarm optimization algorithm to train and optimize the initial power prediction model to obtain a power prediction model, where the particle represents the input weight and the hidden layer bias of the initial power prediction model.

[0118] In one embodiment, the fusion module 340 is specifically used for:

[0119] Perform a rationality check on each candidate power generation to obtain a check result corresponding to the candidate power generation;

[0120] Perform weighted fusion on the candidate power generations with a successful check result to obtain an intermediate power generation;

[0121] Correct the intermediate power generation to obtain the target power generation of the wind farm at a future time.

[0122] In one embodiment, the fusion module 340 corrects the intermediate power generation to obtain the target power generation of the wind farm at a future time, including:

[0123] Calculate the average value of the power generation of the wind farm within a preset time period before the future time, calculate the power deviation value between the intermediate power generation and the average value, and correct the intermediate power generation based on the power deviation value to obtain the target power generation;

[0124] Alternatively, obtain the meteorological forecast data and grid load demand of the wind farm at the future time, and correct the intermediate power generation based on the meteorological forecast data and grid load demand to obtain the target power generation.

[0125] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the above division of each functional module is used as an example. In practical applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. The specific working process of the above-described functional modules can refer to the corresponding process in the foregoing method embodiments and will not be elaborated herein.

[0126] The power prediction device of the wind farm provided in this embodiment can be applied to the power prediction method of the wind farm provided in any of the above embodiments, and has corresponding functions and beneficial effects.

[0127] Figure 4 It is a schematic structural diagram of an electronic device provided in an embodiment of the present application. Figure 4 It shows a block diagram of an exemplary electronic device 11 suitable for implementing the embodiments of the present application. Figure 4 The displayed electronic device 11 is only an example and should not bring any limitations to the functions and usage scope of this embodiment.

[0128] As Figure 4 shown, the electronic device 11 is presented in the form of a general-purpose computing electronic device. The components of the electronic device 11 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 connecting different system components (including the system memory 28 and the processing unit 16).

[0129] Bus 18 represents one or more of several types of bus architectures, including a memory bus or memory controller, a peripheral bus, an Accelerated Graphics Port, a processor, or a local bus using any of the various bus architectures. By way of example, these architectures include, but are not limited to, Industry Standard Architecture (ISA) bus, Micro Channel Architecture (MAC) bus, Enhanced ISA bus, Video Electronics Standards Association (VESA) local bus, and Peripheral Component Interconnect (PCI) bus.

[0130] Electronic device 11 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by electronic device 11, including volatile and nonvolatile media, removable and non-removable media.

[0131] System memory 28 can include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. Electronic device 11 can further include other removable / non-removable, volatile / nonvolatile computer system storage media. By way of example only, storage system 34 can be used for reading and writing on non-removable, nonvolatile magnetic media ( Figure 4 not shown, typically referred to as a "hard disk drive"). Although Figure 4 not shown in, a disk drive for reading and writing on removable nonvolatile disks (such as a "floppy disk"), and an optical disk drive for reading and writing on removable nonvolatile optical disks (such as a CD-ROM, DVD-ROM, or other optical media) can be provided. In these cases, each drive can be connected to bus 18 via one or more data media interfaces. System memory 28 can include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the embodiments of the present application.

[0132] A program / utility 40 having a set (at least one) of program modules 42 can be stored, for example, in system memory 28. Such program modules 42 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data, and an implementation of a network environment may be included in each or some combination of these examples. Program modules 42 typically execute the functions and / or methods in the embodiments described in the present application.

[0133] The electronic device 11 can also communicate with one or more external devices 14 (such as a keyboard, a pointing device, a display 24, etc.), and can also communicate with one or more devices that enable a user to interact with the electronic device 11, and / or communicate with any device that enables the electronic device 11 to communicate with one or more other computing devices (such as a network card, a modem, etc.). Such communication can be carried out through an input / output (I / O) interface 22. Moreover, the electronic device 11 can also communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through a network adapter 20.

[0134] As Figure 4 shown, the network adapter 20 communicates with other modules of the electronic device 11 through a bus 18. It should be understood that although Figure 4 not shown in the figure, other hardware and / or software modules can be used in combination with the electronic device 11, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.

[0135] The processing unit 16 executes various functional applications and page displays by running programs stored in the system memory 28, for example, implementing a power prediction method for a wind farm provided by any embodiment of the present application.

[0136] The embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements, for example, a power prediction method for a wind farm provided by any embodiment of the present application.

[0137] The computer storage medium of this embodiment can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to: an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the computer-readable storage medium include: an electrical connection having 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 this document, the computer-readable storage medium can be any tangible medium that contains or stores a program, and the program can be used by or in combination with an instruction execution system, apparatus, or device.

[0138] A computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which computer-readable program code is carried. Such a propagated data signal may take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device.

[0139] The program code contained on a computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0140] The computer program code for performing the operations of this application may be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's 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).

[0141] Those of ordinary skill in the art should understand that the various modules or steps of the present application described above may be implemented using a general-purpose computing device. They may be concentrated on a single computing device or distributed across a network composed of multiple computing devices. Optionally, they may be implemented using program code executable by a computer device, so that they can be stored in a storage device and executed by the computing device, or they may be separately fabricated into individual integrated circuit modules, or multiple modules or steps among them may be fabricated into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0142] In addition, the acquisition, storage, use, processing, etc. of data in the technical solution of the present application all comply with the relevant provisions of national laws and regulations.

[0143] Note that the above is only the preferred embodiment of the present application and the technical principles applied. Those skilled in the art will understand that the present application is not limited to the specific embodiments here, and various obvious changes, re-adjustments and substitutions can be made by those skilled in the art without departing from the protection scope of the present application. Therefore, although the present application has been described in more detail through the above embodiments, the present application is not limited to the above embodiments. Without departing from the inventive concept of the present application, more other equivalent embodiments can be included, and the scope of the present application is determined by the scope of the appended claims.

Claims

1. A power prediction method for a wind farm, characterized in that, The method includes: Obtaining an operation monitoring data sequence of a wind farm within a historical time period; Predicting the power fluctuation period of the wind farm based on the operation monitoring data sequence to obtain a predicted period; Dividing the historical time period based on the predicted period to obtain a plurality of historical sub-time periods, and using a power prediction model to perform power prediction on the generated power of the wind farm at a future moment based on the operation monitoring data sequence within each historical sub-time period, to obtain a plurality of candidate generated powers at the future moment, where one historical sub-time period corresponds to one candidate generated power; Fusing the plurality of candidate generated powers to obtain the target generated power of the wind farm at the future moment.

2. The power prediction method for a wind farm according to claim 1, characterized in that, The operation monitoring data sequence includes a meteorological data sequence, a fan operation data sequence, and a generated power sequence. The predicting the power fluctuation period of the wind farm based on the operation monitoring data sequence to obtain a predicted period includes: Calculating power fluctuation indexes of the generated power sequence according to time windows of different scales to obtain at least two power fluctuation indexes; Inputting the at least two power fluctuation indexes, the meteorological data sequence, and the fan operation data sequence into a period prediction model to obtain the predicted period.

3. The power prediction method for a wind farm according to claim 2, characterized in that, The using a power prediction model to perform power prediction on the generated power of the wind farm at a future moment based on the operation monitoring data sequence within each historical sub-time period to obtain a plurality of candidate generated powers at the future moment includes: Determining an indirect feature data sequence affecting the generated power according to the meteorological data sequence and the fan operation data sequence; Respectively inputting the meteorological data sequence, the fan operation data sequence, and the indirect feature data sequence within each historical sub-time period into the power prediction model to obtain the plurality of candidate generated powers.

4. The power prediction method for a wind farm according to claim 1, characterized in that, The training process of the power prediction model is as follows: Obtaining sample data, where the sample data includes a historical meteorological data sequence, a historical fan operation data sequence, a historical generated power sequence, a historical indirect feature data sequence, and a generated power label, and the generated power label is the generated power of the wind farm at a to-be-predicted moment corresponding to the sample data; Calculating power fluctuation indexes of the historical generated power sequence according to time windows of different scales to obtain at least two historical power fluctuation indexes, and inputting the at least two historical power fluctuation indexes, the historical meteorological data sequence, and the historical fan operation data sequence into a period prediction model to obtain a period label; Dividing the sample data based on the period label to obtain a plurality of sample sub-data; Inputting the historical meteorological data sequence, the historical fan operation data sequence, and the historical indirect feature data sequence in the sample sub-data into an initial power prediction model, and guiding the training output of the initial power prediction model with the generated power label in the corresponding sample sub-data, and training and optimizing the initial power prediction model to obtain the power prediction model.

5. The power prediction method for a wind farm according to claim 4, wherein inputting the historical meteorological data sequence, historical wind turbine operation data sequence, and historical indirect feature data sequence in the sample sub-data into the initial power prediction model, and guiding the training output of the initial power prediction model with the generated power label corresponding to the sample sub-data to train and optimize the initial power prediction model to obtain the power prediction model, includes: Inputting the historical meteorological data sequence, historical wind turbine operation data sequence, and historical indirect feature data sequence in the sample sub-data into the initial power prediction model to obtain a power training value; Calculating the loss value between the power training value and the generated power label corresponding to the sample sub-data; Taking the loss value as the fitness value of the particle, and using the particle swarm optimization algorithm to train and optimize the initial power prediction model to obtain the power prediction model, where the particle represents the input weight and hidden layer bias of the initial power prediction model.

6. The power prediction method for a wind farm according to claim 1, wherein The step of fusing the multiple candidate generated powers to obtain the target generated power of the wind farm at the future moment includes: Performing a rationality check on each candidate generated power to obtain the check result corresponding to the candidate generated power; Performing weighted fusion on the candidate generated powers with a successful check result to obtain an intermediate generated power; Modifying the intermediate generated power to obtain the target generated power of the wind farm at the future moment.

7. The power prediction method for a wind farm according to claim 6, characterized in that, The step of modifying the intermediate generated power to obtain the target generated power of the wind farm at the future moment includes: Calculating the average value of the generated power within a preset time period before the future moment of the wind farm, calculating the power deviation value between the intermediate generated power and the average value, and modifying the intermediate generated power based on the power deviation value to obtain the target generated power; Alternatively, obtaining the meteorological forecast data and grid load demand at the future moment of the wind farm, and modifying the intermediate generated power based on the meteorological forecast data and the grid load demand to obtain the target generated power.

8. A power prediction device for a wind farm, characterized in that, The device includes: An acquisition module, configured to acquire the operation monitoring data sequence of the wind farm within a historical time period; A period prediction module, configured to predict the power fluctuation period of the wind farm based on the operation monitoring data sequence to obtain a predicted period; A power prediction module, configured to divide the historical time period based on the predicted period to obtain a plurality of historical sub-time periods, and use the power prediction model to perform power prediction on the generated power of the wind farm at a future moment based on the operation monitoring data sequence within each historical sub-time period to obtain a plurality of candidate generated powers at the future moment, where one historical sub-time period corresponds to one candidate generated power; A fusion module, configured to fuse the plurality of candidate generated powers to obtain the target generated power of the wind farm at the future moment.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and when the computer program is executed by the at least one processor, the at least one processor is enabled to execute the power prediction method of the wind farm according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the power prediction method of the wind farm according to any one of claims 1 to 7.

Images