Short-term wind power prediction method, system and device and storage medium

Abstract

The invention discloses a short-term wind power prediction method, system and device and a storage medium, and the method comprises the steps: collecting weather feature data and wind power data, and carrying out the preprocessing of the collected weather feature data and wind power data; performing feature extraction on the preprocessed data by using a convolutional neural network, a stack denoising auto-encoder and a deep belief network to form a high-order feature data pool; performing training modeling on the high-order feature data in the high-order feature data pool by adopting an optimized XGBoost model to obtain a nonlinear relationship between the high-order feature data and the power of the wind power plant so as to form a short-term wind power prediction model; and performing short-term wind power prediction by using the short-term wind power prediction model. According to the method, the system, the equipment and the storage medium, the accuracy of short-term wind power prediction can be improved.

Description

Technical Field

[0001] The present invention belongs to the field of wind power prediction, and relates to a short-term wind power prediction method, system, equipment and storage medium. Background Art

[0002] At present, the main methods for wind power prediction are physical methods, statistical methods and artificial intelligence methods, among which artificial intelligence methods are the hotspot of research. The widely used artificial intelligence models mainly include back propagation neural network, support vector machine, convolutional neural network (CNN), long short-term memory network (LSTM), stacked denoising autoencoder network (SDAE) and deep belief network (DBN).

[0003] As an emerging research direction in the field of artificial intelligence, deep learning has been applied in the field of wind power forecasting. However, there is still room for improvement in the prediction effect, that is, the prediction accuracy. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the above-mentioned prior art and provide a short-term wind power prediction method, system, device and storage medium, which can improve the accuracy of short-term wind power prediction.

[0005] In order to achieve the above object, the present invention adopts the following technical scheme:

[0006] In one aspect, the present invention provides a short-term wind power prediction method, comprising:

[0007] Collect weather characteristic data and wind power data, and pre-process the collected weather characteristic data and wind power data;

[0008] Use convolutional neural networks, stacked denoising autoencoders and deep belief networks to extract features from preprocessed data to form a high-order feature data pool;

[0009] The optimized XGBoost model is used to train and model the high-order feature data in the high-order feature data pool, so as to obtain a nonlinear relationship between the high-order feature data and the wind farm power, so as to form a short-term wind power prediction model;

[0010] The short-term wind power prediction model is used to perform short-term wind power prediction.

[0011] The short-term wind power prediction method of the present invention is further improved in that:

[0012] The process of preprocessing the collected weather characteristic data and wind power data is as follows:

[0013] The collected weather characteristic data and wind power data are subjected to outlier elimination and normalization processing.

[0014] The process of using convolutional neural networks, stacked denoising autoencoders and deep belief networks to extract features from preprocessed data to form a high-order feature data pool is as follows:

[0015] Convolutional neural networks, stacked denoising autoencoders and deep belief networks are used to extract features from the preprocessed data. The features of the last hidden layer of the stacked denoising autoencoder and deep belief network and the output features of the last Flatten layer in the convolutional neural network are extracted to form a high-order feature data pool.

[0016] The differential evolution-balanced optimizer hybrid algorithm is used to optimize the hyperparameters of the XGBoost model to obtain the optimized XGBoost model.

[0017] The process of optimizing the various hyperparameters of XGBoost using the differential evolution-balanced optimizer hybrid algorithm is as follows:

[0018] 31) Initialize the value range of each parameter of the XGBoost model and set the number of iterations of the differential evolution-balanced optimizer hybrid algorithm to T*;

[0019] 32) Randomly generate N populations, and the individuals in each population are equivalent to a set of XGBoost parameters;

[0020] 33) Using the prediction accuracy as the individual fitness value, the individual fitness values ​​of all particles in the N populations in step 32) are initialized;

[0021] 34) performing a DE optimization on the N populations to obtain N optimal particles;

[0022] 35) Calculating the fitness values ​​of the N optimal particles, and updating the optimal individual representing the current number of iterations according to the fitness values;

[0023] 36) Return to step 34) and continue to complete the classic DE optimization until the upper limit T1 of the number of iterations is reached. Each population will have T1 historical optimal particles. The fitness of the T1 historical optimal particles is compared, and the particle with the highest fitness is taken as the optimal individual of the population, and N optimal individuals from N populations are obtained;

[0024] 37) The N optimal individuals obtained in step 36) are used as the initial population of the EO algorithm to perform the improved EO optimization, and the number of iterations of the EO algorithm is set to T2;

[0025] 38) Initialize the parameters of the EO algorithm, including the generation rate GP, the number of iterations T2, the number of particles N, the upper and lower limits Cmax and Cmin of the particle search space, use the prediction accuracy as the calculation formula of the fitness value, calculate the fitness value of the particle, and then perform an improved EO optimization;

[0026] 39) Determine whether the current number of iterations is ≤ T2. If so, return to step 38) to continue the current EO optimization. Otherwise, jump to step 310);

[0027] 310) Determine whether the current total number of iterations is ≤T*. If so, randomly select K historical optimal particles in EO to replace the K individuals in each DE population in step 32), and return to step 32) to continue optimization; if not, output the optimal solution.

[0028] The process of improving EO optimization in step 38) is as follows:

[0029] Update the balancing pool and exponential term coefficients, select the guide particles and update the generation rate, and update the concentration. The constant a1, constant a2 and iteration function I in the standard EO are:

[0030]

[0031]

[0032]

[0033] In a second aspect of the present invention, the present invention provides a short-term wind power prediction system, comprising:

[0034] A collection module is used to collect weather characteristic data and wind power data, and pre-process the collected weather characteristic data and wind power data;

[0035] A feature extraction module is used to extract features from the preprocessed data using a convolutional neural network, a stacked denoising autoencoder, and a deep belief network to form a high-order feature data pool;

[0036] A training modeling module is used to use the optimized XGBoost model to train and model the high-order feature data in the high-order feature data pool, so as to obtain a nonlinear relationship between the high-order feature data and the wind farm power, so as to form a short-term wind power prediction model;

[0037] The prediction module is used to perform short-term wind power prediction using the short-term wind power prediction model.

[0038] The process of using convolutional neural networks, stacked denoising autoencoders and deep belief networks to extract features from preprocessed data to form a high-order feature data pool is as follows:

[0039] Convolutional neural networks, stacked denoising autoencoders and deep belief networks are used to extract features from the preprocessed data. The features of the last hidden layer of the stacked denoising autoencoder and deep belief network and the output features of the last Flatten layer in the convolutional neural network are extracted to form a high-order feature data pool.

[0040] In a third aspect of the present invention, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the short-term wind power forecasting method when executing the computer program.

[0041] In a fourth aspect of the present invention, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the short-term wind power prediction method are implemented.

[0042] The present invention has the following beneficial effects:

[0043] During specific operation, the short-term wind power prediction method, system, device and storage medium described in the present invention use convolutional neural networks, stacked denoising autoencoders and deep belief networks to extract features from input data, effectively utilizing the feature mining capabilities of three deep learning models. In addition, the XGBoost (DEEO-XGBoost) optimized by the differential evolution-balanced optimizer hybrid algorithm used in the present invention can effectively model the nonlinear relationship between the feature data mined by the deep learning algorithm and the wind farm power, thereby effectively improving the prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings:

[0045] Figure 1 is a flow chart of the method of the present invention;

[0046] Figure 2 A flowchart for optimizing the hyperparameters of XGBoost using the differential evolution-balanced optimizer hybrid algorithm DEEO. DETAILED DESCRIPTION

[0047] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0048] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0049] The present invention is further described in detail below in conjunction with the accompanying drawings:

[0050] Embodiment 1

[0051] refer to Figure 1 The short-term wind power prediction method of the present invention comprises the following steps:

[0052] 1) Collect weather characteristic data and wind power data, and pre-process the collected weather characteristic data and wind power data. Specifically, remove abnormal values ​​and normalize the weather characteristic data and wind power data;

[0053] 2) Use convolutional neural network (CNN), stacked denoising autoencoder (SDAE) and deep belief network (DBN) to extract features from the preprocessed data, extract the features of the last hidden layer of SDAE and DBN and the output features of the last Flatten layer in CNN to form a high-order feature data pool;

[0054] Convolutional neural networks can respond to surrounding units within a part of the coverage area and have excellent performance in large-scale image processing; convolutional neural networks are composed of multiple alternating convolutional layers and pooling layers. Convolutional neural networks use convolution kernels (filters) to extract local features of input data, realize weight sharing and reduce the number of parameters. The present invention uses a fully convolutional neural network to realize feature mining of wind power data, and injects the output of the last Flatten layer of the network into the feature pool as part of the input features of DEEO-XGBoost.

[0055] The stacked denoising autoencoder is composed of multiple denoising autoencoders (DAE) stacked together, and then a regression output layer is added to the top layer. The denoising autoencoder is an autoencoder (AE) that can learn high-level feature representations of data by adding random noise to the input data and then trying to reconstruct the original data. The training process of SDAE for regression problems is divided into two stages: pre-training and fine-tuning. In the pre-training stage, each DAE is trained layer by layer to minimize the error between the input and the reconstruction; in the fine-tuning stage, a regression output layer is added to the top layer of SDAE, and then the entire network is fine-tuned in a supervised manner using labeled data. SDAE for regression problems can be used for unsupervised feature extraction or supervised regression tasks. The present invention injects the output of the last hidden layer of the SDAE network into the feature pool as another part of the input features of DEEO-XGBoost.

[0056] The deep belief network DBN is a probabilistic generative model composed of multiple layers of random latent variables. DBN can be seen as a stack of a series of restricted Boltzmann machines RBM. DBN can be used for regression problems, that is, predicting a continuous value or discrete value based on the input data. The training process of DBN for regression problems is divided into the following steps: layer-by-layer sampling, calculation of positive gradient, Gibbs sampling, updating weights and biases, and supervised back propagation. The present invention injects the output of the last hidden layer of the DBN network into the feature pool as the last part of the input features of DEEO-XGBoost.

[0057] 3) XGBoost (DEEO-XGBo ost) optimized by the differential evolution-balanced optimizer hybrid algorithm is used to train and model the high-order feature data in the high-order feature data pool in step 2), and a nonlinear relationship between the high-order feature data and the wind farm power is established by continuously optimizing the parameters to form a short-term wind power prediction model;

[0058] XGBoost can be used for various machine learning tasks such as regression, classification, and sorting. Figure 2As shown in the figure, DEEO-XGBoost uses the differential evolution-balanced optimizer hybrid algorithm DEEO to optimize the various hyperparameters of XGBoo st. Figure 2 , the specific steps are:

[0059] 31) Initialize the value range of each parameter of the XGBoost model and set the number of iterations of the DEEO overall optimization algorithm to T*;

[0060] 32) Randomly generate N populations, and the individuals in each population are equivalent to a set of XGBoost parameters (parameter 1, parameter 2, ..., parameter d);

[0061] 33) Using the prediction accuracy as the individual fitness value, the individual fitness values ​​of all particles in the N populations in step 32) are initialized;

[0062] 34) Performing a classical DE optimization on the N populations to obtain N optimal particles, the DE optimization process is: each population contains m individuals, the number of iterations of each population is set to T1, and the m individuals are subjected to mutation, crossover and selection operations to further update the population;

[0063] 35) Calculate the fitness value of each updated particle, and update the optimal individual representing the current number of iterations according to the size of the fitness value;

[0064] 36) Return to step 34) and continue to complete the classic DE optimization until the upper limit of the number of iterations T1 is reached. Each population will have T1 historical optimal particles. The fitness of these particles will be compared, and the particle with the highest fitness value will be taken as the optimal individual of the population. Finally, N optimal individuals from N populations will be obtained;

[0065] 37) The N optimal individuals obtained in step 36) are used as the initial population of the EO algorithm to perform the improved EO optimization, and the number of iterations of the EO algorithm is set to T2;

[0066] 38) Initialize the parameters of the EO algorithm, including the generation rate GP, the number of iterations T2, the number of particles N, the upper and lower limits Cmax and Cmin of the particle search space, use the prediction accuracy as the calculation formula of the fitness value, calculate the fitness value of the particle, and then perform an improved EO optimization: update the balance pool and the exponential term coefficient, select the guide particle and update the generation rate, update the concentration, wherein the present invention improves the constant a1, constant a2 and iteration function I in the standard EO to:

[0067]

[0068]

[0069]

[0070] 39) Determine whether the current number of iterations is ≤ T2. If so, return to step 38) to continue the current EO optimization. Otherwise, jump to step 10);

[0071] 310) Determine whether the current total number of iterations is ≤ T*. If so, randomly select K historical optimal particles in EO to replace the K individuals in each DE population in step 32), and return to step 32) to continue optimization; if not, output the optimal solution.

[0072] 4) Using the short-term wind power prediction model obtained in step 3) to predict short-term wind power.

[0073] Confirmatory trials

[0074] To prove the effectiveness of the present invention, the data from a wind farm in Jiangsu from July 1, 2009 to July 12, 2012 are used to train and test the model, with a time resolution of 1 hour. The data from July 1, 2009 to February 21, 2012 are used for training, while the data after February 21, 2012 are used for testing. The standard mean absolute error and standard root mean square error of the five prediction models are shown in Table 1.

[0075] Table 1

[0076] method Standard mean absolute error Standardized root mean square error SDAE+CNN+DBN+DEEO-XGBoost 0.0212 0.0411 SDAE+CNN+DBN+EO-XGBoost 0.0224 0.0417 SDAE 0.0353 0.0567 CNN 0.0364 0.0569 DBN 0.0368 0.0571

[0077] Embodiment 2

[0078] Accordingly, the present invention provides a short-term wind power prediction system, comprising:

[0079] A collection module is used to collect weather characteristic data and wind power data, and pre-process the collected weather characteristic data and wind power data;

[0080] A feature extraction module is used to extract features from the preprocessed data using a convolutional neural network, a stacked denoising autoencoder, and a deep belief network to form a high-order feature data pool;

[0081] A training modeling module is used to use the optimized XGBoost model to train and model the high-order feature data in the high-order feature data pool, so as to obtain a nonlinear relationship between the high-order feature data and the wind farm power, so as to form a short-term wind power prediction model;

[0082] The prediction module is used to perform short-term wind power prediction using the short-term wind power prediction model.

[0083] The process of using convolutional neural networks, stacked denoising autoencoders and deep belief networks to extract features from preprocessed data to form a high-order feature data pool is as follows:

[0084] Convolutional neural networks, stacked denoising autoencoders and deep belief networks are used to extract features from the preprocessed data. The features of the last hidden layer of the stacked denoising autoencoder and deep belief network and the output features of the last Flatten layer in the convolutional neural network are extracted to form a high-order feature data pool.

[0085] Embodiment 3

[0086] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the short-term wind power prediction method are implemented. The memory may include a memory, such as a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk memory, etc. The processor, the network interface, and the memory are interconnected through an internal bus. The internal bus may be an industrial standard architecture bus, a peripheral component interconnection standard bus, an extended industrial standard structure bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. The memory is used to store programs. Specifically, the program may include a program code, and the program code includes computer operation instructions. The memory may include a memory and a non-volatile memory, and provide instructions and data to the processor.

[0087] Embodiment 4

[0088] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the short-term wind power prediction method are implemented. Specifically, the computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. The volatile memory may include random access memory (RAM) and / or cache memory (cache), etc. The non-volatile memory may include read-only memory (ROM), hard disk, flash memory, optical disk, magnetic disk, etc.

[0089] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0090] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0091] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0092] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0093] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A short-term wind power prediction method, characterized in that: include: Collect weather characteristic data and wind power data, and pre-process the collected weather characteristic data and wind power data; Use convolutional neural networks, stacked denoising autoencoders and deep belief networks to extract features from preprocessed data to form a high-order feature data pool; The optimized XGBoost model is used to train and model the high-order feature data in the high-order feature data pool, so as to obtain a nonlinear relationship between the high-order feature data and the wind farm power, so as to form a short-term wind power prediction model; The short-term wind power prediction model is used to perform short-term wind power prediction.

2. The short-term wind power prediction method according to claim 1, characterized in that: The process of preprocessing the collected weather characteristic data and wind power data is as follows: The collected weather characteristic data and wind power data are subjected to outlier elimination and normalization processing.

3. The short-term wind power prediction method according to claim 1, characterized in that: The process of using convolutional neural networks, stacked denoising autoencoders and deep belief networks to extract features from preprocessed data to form a high-order feature data pool is as follows: Convolutional neural networks, stacked denoising autoencoders and deep belief networks are used to extract features from the preprocessed data. The features of the last hidden layer of the stacked denoising autoencoder and deep belief network and the output features of the last Flatten layer in the convolutional neural network are extracted to form a high-order feature data pool.

4. The short-term wind power prediction method according to claim 1, characterized in that: The differential evolution-balanced optimizer hybrid algorithm is used to optimize the hyperparameters of the XGBoost model to obtain the optimized XGBoost model.

5. The short-term wind power prediction method according to claim 4, characterized in that: The process of optimizing the various hyperparameters of XGBoost using the differential evolution-balanced optimizer hybrid algorithm is as follows: 31) Initialize the value range of each parameter of the XGBoost model and set the number of iterations of the differential evolution-balanced optimizer hybrid algorithm to T*; 32) Randomly generate N populations, and the individuals in each population are equivalent to a set of XGBoost parameters; 33) Using the prediction accuracy as the individual fitness value, the individual fitness values ​​of all particles in the N populations in step 32) are initialized; 34) performing a DE optimization on the N populations to obtain N optimal particles; 35) Calculating the fitness values ​​of the N optimal particles, and updating the optimal individual representing the current number of iterations according to the fitness values; 36) Return to step 34) and continue to complete the classic DE optimization until the upper limit T1 of the number of iterations is reached. Each population will have T1 historical optimal particles. The fitness of the T1 historical optimal particles is compared, and the particle with the highest fitness is taken as the optimal individual of the population, and N optimal individuals from N populations are obtained; 37) The N optimal individuals obtained in step 36) are used as the initial population of the EO algorithm to perform the improved EO optimization, and the number of iterations of the EO algorithm is set to T2; 38) Initialize the parameters of the EO algorithm, including the generation rate GP, the number of iterations T2, the number of particles N, the upper and lower limits Cmax and Cmin of the particle search space, use the prediction accuracy as the calculation formula of the fitness value, calculate the fitness value of the particle, and then perform an improved EO optimization; 39) Determine whether the current number of iterations is ≤ T2. If so, return to step 38) to continue the current EO optimization. Otherwise, jump to step 310); 310) Determine whether the current total number of iterations is ≤T*. If so, randomly select K historical optimal particles in EO to replace the K individuals in each DE population in step 32), and return to step 32) to continue optimization; if not, output the optimal solution.

6. The short-term wind power prediction method according to claim 5, characterized in that: The process of improving EO optimization in step 38) is as follows: Update the balancing pool and exponential term coefficients, select the guide particles and update the generation rate, and update the concentration. The constant a1, constant a2 and iteration function I in the standard EO are:

7. A short-term wind power prediction system, characterized in that: include: A collection module is used to collect weather characteristic data and wind power data, and pre-process the collected weather characteristic data and wind power data; A feature extraction module is used to extract features from the preprocessed data using a convolutional neural network, a stacked denoising autoencoder, and a deep belief network to form a high-order feature data pool; A training modeling module is used to use the optimized XGBoost model to train and model the high-order feature data in the high-order feature data pool, so as to obtain a nonlinear relationship between the high-order feature data and the wind farm power, so as to form a short-term wind power prediction model; The prediction module is used to perform short-term wind power prediction using the short-term wind power prediction model.

8. The short-term wind power prediction system according to claim 7, characterized in that: The process of using convolutional neural networks, stacked denoising autoencoders and deep belief networks to extract features from preprocessed data to form a high-order feature data pool is as follows: Convolutional neural networks, stacked denoising autoencoders and deep belief networks are used to extract features from the preprocessed data. The features of the last hidden layer of the stacked denoising autoencoder and deep belief network and the output features of the last Flatten layer in the convolutional neural network are extracted to form a high-order feature data pool.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the short-term wind power prediction method according to any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the short-term wind power prediction method according to any one of claims 1 to 6 are implemented.

Images