Ultra-short-term wind power prediction method and device based on improved Hilbert-Huang transform
By improving the combination of Hilbert-yellow transformation and the time series lightweight adaptive network, the problem of inaccurate wind power prediction is solved, accurate prediction of wind speed and efficient prediction of wind power are achieved, and the accuracy and safety of power grid scheduling are improved.
Patent Information
- Application Number
- CN202510439732.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-09
- Publication Date
- 2025-07-11
AI Technical Summary
The prior art is difficult to achieve accurate prediction of wind speed, resulting in inaccurate prediction of wind power, affecting power grid scheduling and safety.
The improved Hilbert-yellow transformation method is used to decompose and correct the wind speed data. Combined with the time series lightweight adaptive network, an ultra-short-term wind power power prediction model is built, and the residual components and modal components are converted through the Hilbert transformation algorithm, and the linear network is used for processing to output more accurate wind power power prediction.
It improves the accuracy of wind power prediction, can track the real wind speed trend, improves prediction accuracy, reflects the dynamic characteristics of the system, is simple to calculate and high prediction performance, and the results are effective and practical.
Smart Images

Figure CN120300783A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of wind power prediction, and particularly to an ultra-short-term wind power prediction method and device based on improved Hilbert-Huang transform. Background Art
[0002] As the proportion of renewable energy such as wind energy in the energy structure continues to increase, high-penetration wind power becomes crucial for the normal operation of the power system. Wind energy, as a clean and environmentally friendly renewable energy, provides an important way to reduce dependence on fossil fuels and lower greenhouse gas emissions.
[0003] The rapid development of the wind power industry is restricted by various factors. Wind speed and wind direction are two key physical parameters that determine the potential of wind power generation. The research on wind energy resources mainly focuses on these attributes, and wind speed is regarded as the main indicator. For example, the power grid requires stable and smooth power input, but natural wind, due to its inherent randomness and intermittency, may cause fluctuations after being integrated into the power grid, which poses challenges to the scheduling and control of wind farms and may trigger potential safety problems. Therefore, accurate and effective wind speed prediction is crucial for solving the wind power prediction problem. To address these challenges, it is particularly important to develop more accurate wind speed prediction and wind power prediction models. These models can predict the power output of wind farms within a specific future time period, thus assisting grid operators to make more accurate scheduling decisions.
[0004] Therefore, how to invent an ultra-short-term wind power prediction method to achieve accurate prediction of wind speed and improve the prediction accuracy of wind power has become an urgent problem to be solved. Summary of the Invention
[0005] For this purpose, the present invention provides an ultra-short-term wind power prediction method and device based on improved Hilbert-Huang transform. By decomposing and correcting the wind speed in numerical weather prediction, the predicted wind speed is made closer to the real wind speed, and thus the power can be predicted more accurately. At the same time, the dynamic relationship between wind power time series data is also considered, and an adaptive module with feature enhancement and a noise interference reduction module are used for modeling to achieve more accurate prediction of wind power within the next 4 hours.
[0006] To achieve the above object, the present invention provides the following technical solutions: An ultra-short-term wind power prediction method based on improved Hilbert-Huang transform, comprising:
[0007] Detect and process outliers and missing values in the collected wind power data and wind speed data to obtain initial wind power data and initial wind speed data; Add white noise to the initial wind speed data to generate new wind speed data; Perform iterative empirical mode decomposition on the fresh wind speed data to obtain several modal components; subtract the several modal components from the initial wind speed data in sequence to obtain a residual component; Convert the residual component and the modal components through the Hilbert transform algorithm to obtain new sequence data; input the new sequence data into a linear network for processing to obtain corrected wind speed data; Construct a very short-term wind power prediction model based on a time series lightweight adaptive network; Input the corrected wind speed data and the initial wind power data into the very short-term wind power prediction model for prediction, and output very short-term wind power prediction data.
[0008] As a preferred solution of the very short-term wind power prediction method based on the improved Hilbert-Huang transform, add white noise to the initial wind speed data, and the expression for generating the fresh wind speed data is: ; In the formula, is the fresh wind speed data; is the initial wind speed data; is the white noise; is the noise amplitude.
[0009] As a preferred solution of the very short-term wind power prediction method based on the improved Hilbert-Huang transform, the calculation formula for the residual component is: ; In the formula, is the residual component; K is the number of iterations; are the modal components obtained at different iteration times.
[0010] As a preferred solution of the very short-term wind power prediction method based on the improved Hilbert-Huang transform, the time series lightweight adaptive network includes several adaptive spectral blocks and several interactive convolution blocks; map the time series data from the time domain to the frequency domain through the adaptive spectral blocks, and cut high-frequency noise through an adaptive threshold to highlight the frequency domain features; improve the time domain features through the interactive convolution blocks to improve the adaptability of the time domain features to time dynamics.
[0011] As a preferred solution of the very short-term wind power prediction method based on the improved Hilbert-Huang transform, evaluate the prediction performance of the very short-term wind power prediction model through the root mean square error and mean absolute error of the prediction results; The expression for the root mean square error is: ; In the formula, RMSE is the root mean square error; For predicting wind power is the actual wind power is the installed capacity of the wind farm; N is the number of prediction points The expression of the mean absolute error is as follows ; In the formula, MAE is the mean absolute error
[0012] The present invention also provides an ultra-short-term wind power prediction device based on the improved Hilbert-Huang transform. Based on the above ultra-short-term wind power prediction method based on the improved Hilbert-Huang transform, it includes The original data acquisition and processing module is used to detect and process outliers and missing values in the collected wind power data and wind speed data to obtain initial wind power data and initial wind speed data The new wind speed data generation module is used to add white noise to the initial wind speed data to generate new wind speed data The new wind speed data mode decomposition module is used to perform iterative empirical mode decomposition on the new wind speed data to obtain several mode components; subtract the several mode components from the initial wind speed data in turn to obtain a residual component The corrected wind speed data acquisition module is used to convert the residual component and the mode component through the Hilbert transform algorithm to obtain new sequence data; input the new sequence data into a linear network for processing to obtain corrected wind speed data The ultra-short-term wind power prediction model construction module is used to construct an ultra-short-term wind power prediction model based on a time series lightweight adaptive network The ultra-short-term wind power prediction model prediction module is used to input the corrected wind speed data and the initial wind power data into the ultra-short-term wind power prediction model for prediction, and output ultra-short-term wind power prediction data
[0013] As a preferred scheme of the ultra-short-term wind power prediction device based on the improved Hilbert-Huang transform, in the new wind speed data generation module, the expression for adding white noise to the initial wind speed data to generate the new wind speed data is ; In the formula is the new wind speed data is the initial wind speed data is the white noise is the noise amplitude
[0014] As a preferred scheme of the ultra-short-term wind power prediction device based on the improved Hilbert-Huang transform, in the new wind speed data mode decomposition module, the calculation formula for the residual component is ; In the formula, is the residual component; K is the number of iterations; are the modal components obtained at different numbers of iterations.
[0015] As an optimized solution of the ultra-short-term wind power prediction device based on the improved Hilbert-Huang transform, in the ultra-short-term wind power prediction model construction module, the time series lightweight adaptive network includes a number of adaptive spectrum blocks and a number of interactive convolution blocks; the time series data is mapped from the time domain to the frequency domain through the adaptive spectrum blocks, and the high-frequency noise is reduced by an adaptive threshold to highlight the frequency domain features; the time domain features are improved through the interactive convolution blocks to improve the adaptability of the time domain features to time dynamics.
[0016] As an optimized solution of the ultra-short-term wind power prediction device based on the improved Hilbert-Huang transform, in the ultra-short-term wind power prediction model prediction module, the prediction performance of the ultra-short-term wind power prediction model is evaluated through the root mean square error and the mean absolute error of the prediction results; The expression of the root mean square error is: ; In the formula, RMSE is the root mean square error; is the predicted wind power; is the actual wind power; is the installed capacity of the wind farm; N is the number of prediction points; The expression of the mean absolute error is: ; In the formula, MAE is the mean absolute error.
[0017] The present invention has the following advantages: The present invention detects and processes outliers and missing values in the collected wind power data and wind speed data to obtain initial wind power data and initial wind speed data; adds white noise to the initial wind speed data to generate new wind speed data; performs iterative empirical mode decomposition on the new wind speed data to obtain a number of modal components; subtracts the number of modal components from the initial wind speed data in sequence to obtain a residual component; converts the residual component and the modal components through the Hilbert transform algorithm to obtain new sequence data; inputs the new sequence data into a linear network for processing to obtain corrected wind speed data; constructs a very short-term wind power prediction model based on a time series lightweight adaptive network; inputs the corrected wind speed data and the initial wind power data into the very short-term wind power prediction model for prediction, and outputs very short-term wind power prediction data. The present invention corrects the predicted wind speed of numerical weather prediction by improving the Hilbert-Huang transform, can track the real wind speed trend fluctuation, and captures the dynamic relationship between time series data through a time series lightweight adaptive network, thereby improving the prediction accuracy of very short-term wind power. It can reflect the dynamic characteristics of the system and track the future power trend. The present invention has simple calculation, high prediction performance, clear meaning, effective prediction results, and strong practicability. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings described below are only exemplary, and for those of ordinary skill in the art, without creative efforts, other implementation drawings can be obtained based on the provided drawings.
[0019] The structures, ratios, sizes, etc. illustrated in this specification are only used to cooperate with the content disclosed in the specification for those who are familiar with this technology to understand and read, and are not used to limit the limited conditions under which the present invention can be implemented. Therefore, they do not have technical essential meanings. Any modification of the structure, change of the proportional relationship, or adjustment of the size, without affecting the effects that the present invention can produce and the purposes that can be achieved, should still fall within the scope covered by the technical content disclosed in the present invention.
[0020] Figure 1 It is a schematic flowchart of a very short-term wind power prediction method based on an improved Hilbert-Huang transform provided in Embodiment 1 of the present invention; Figure 2 It is a schematic flowchart of the specific implementation of a very short-term wind power prediction method based on an improved Hilbert-Huang transform provided in Embodiment 1 of the present invention; Figure 3Schematic diagram of the wind speed correction process in the ultra-short-term wind power prediction method based on the improved Hilbert-Huang transform provided in Embodiment 1 of the present invention; Figure 4 Schematic diagram of the time series lightweight adaptive network structure in the ultra-short-term wind power prediction method based on the improved Hilbert-Huang transform provided in Embodiment 1 of the present invention; Figure 5 Schematic diagram of the application of the Hilbert transform to the EMD decomposition result in a possible embodiment provided in Embodiment 1 of the present invention; Figure 6 Schematic diagram of the result of introducing CEEMDAN to improve the Hilbert transform in a possible embodiment provided in Embodiment 1 of the present invention; Figure 7 Schematic diagram of the result curve for comparing different wind speed correction methods in a possible embodiment provided in Embodiment 1 of the present invention; Figure 8 Schematic diagram of the result curve for comparing the prediction results of different prediction methods at the 1st hour in a possible embodiment provided in Embodiment 1 of the present invention; Figure 9 Schematic diagram of the result curve for comparing the prediction results of different prediction methods at the 2nd hour in a possible embodiment provided in Embodiment 1 of the present invention; Figure 10 Schematic diagram of the result curve for comparing the prediction results of different prediction methods at the 3rd hour in a possible embodiment provided in Embodiment 1 of the present invention; Figure 11 Schematic diagram of the result curve for comparing the prediction results of different prediction methods at the 4th hour in a possible embodiment provided in Embodiment 1 of the present invention; Figure 12 Schematic diagram for comparing the RMSE values of different prediction methods at different prediction time scales in a possible embodiment provided in Embodiment 1 of the present invention; Figure 13 Schematic diagram for comparing the MAE values of different prediction methods at different prediction time scales in a possible embodiment provided in Embodiment 1 of the present invention; Figure 14 Schematic diagram of the architecture of the ultra-short-term wind power prediction device based on the improved Hilbert-Huang transform provided in Embodiment 2 of the present invention. Detailed implementation manners
[0021] The following specific embodiments illustrate the implementation manners of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.
[0022] Embodiment 1
[0023] Refer to Figure 1 and Figure 2 Embodiment 1 of the present invention provides a very short-term wind power prediction method based on an improved Hilbert-Huang transform, including the following steps: S1. Detect and process outliers and missing values in the collected wind power data and wind speed data to obtain initial wind power data and initial wind speed data; S2. Add white noise to the initial wind speed data to generate new wind speed data; S3. Perform iterative empirical mode decomposition on the new wind speed data to obtain several modal components; subtract the several modal components from the initial wind speed data in sequence to obtain a residual component; S4. Convert the residual component and the modal components through the Hilbert transform algorithm to obtain new sequence data; input the new sequence data into a linear network for processing to obtain corrected wind speed data; S5. Construct a very short-term wind power prediction model based on a time series lightweight adaptive network; S6. Input the corrected wind speed data and the initial wind power data into the very short-term wind power prediction model for prediction, and output very short-term wind power prediction data.
[0024] In this embodiment, in step S1, detect and process outliers and missing values in the collected wind power data and wind speed data to obtain initial wind power data and initial wind speed data; Specifically, detect and process outliers and missing values in the wind power data obtained from the data acquisition and monitoring control system and the wind speed data in the numerical weather forecast provided by the meteorological center to obtain initial wind power data and initial wind speed data.
[0025] In this embodiment, in step S2, add white noise to the initial wind speed data to generate new wind speed data; Specifically, considering the error between the initial wind speed in the numerical weather forecast and the actual wind speed, therefore, first add white noise to the initial wind speed data in the numerical weather forecast to generate new wind speed data.
[0026] Among them, the expression for generating the new wind speed data is: ; In the formula, is the new wind speed data; is the initial wind speed data; is white noise; is the noise amplitude.
[0027] In this embodiment, in step S3, the new wind speed data is subjected to iterative empirical mode decomposition to obtain a number of modal components; the initial wind speed data is successively subtracted from the number of modal components to obtain a residual component; Specifically, as Figure 3 shown, the wind speed data after adding white noise is subjected to empirical mode decomposition to obtain the first group of modal components (IMF) and residuals. Then, the initial wind speed is used to subtract the modal components of the first group to obtain a new residual, and the iteration is performed successively to obtain a number of modal components; the initial wind speed data is successively subtracted from the number of modal components to obtain the final residual component.
[0028] Among them, the calculation formula for the residual component is: ; In the formula, is the residual component; K is the number of iterations; are the modal components obtained from different numbers of iterations.
[0029] In this embodiment, in step S4, the residual component and the modal components are converted through the Hilbert transform algorithm to obtain new sequence data; the new sequence data is input into a linear network for processing to obtain the corrected wind speed data; Specifically, as Figure 3 shown, the obtained components are input into the Hilbert transform algorithm for conversion to obtain new sequence data. Then, the converted components are input into the linear network for separate training. After separately modeling using the linear neural network, the weights of each component and the residual are obtained using the dynamic adaptive weight layer and accumulated to obtain the corrected wind speed data.
[0030] In this embodiment, in step S5, a very short-term wind power prediction model is constructed based on the time series lightweight adaptive network; Among them, the time-series lightweight adaptive network includes a number of adaptive spectrum blocks (ASBs) and a number of interactive convolution blocks (ICBs); the time-series data is mapped from the time domain to the frequency domain through the adaptive spectrum blocks, and high-frequency noise is reduced through an adaptive threshold to highlight the frequency-domain features; after processing, IFFT is used to reconstruct the time-domain features, aiming to reduce noise and enhance representation. The time-domain features are improved interactively using different kernel sizes through the interactive convolution blocks to improve the adaptability of the time-domain features to time dynamics.
[0031] Specifically, as Figure 4 shown, in order to balance the time feature extraction of local and global time-series analysis, the neural network layer composed of ASBs and ICBs forms a cohesive structure. Finally, a very short-term wind power prediction model is obtained through multiple rounds of training and learning.
[0032] In this embodiment, in step S6, the corrected wind speed data and the initial wind power data are input into the very short-term wind power prediction model for prediction, and very short-term wind power prediction data is output.
[0033] Specifically, the corrected wind speed data and the initial wind power data are input into the very short-term wind power prediction model for prediction, and very short-term wind power prediction data is output.
[0034] In a possible embodiment, a specific wind power prediction example is provided as follows: In this embodiment, multiple prediction models are selected to perform very short-term prediction on the wind power of the test set. The power data of a wind farm in Inner Mongolia and the NWP wind speed are used for experimental analysis, and the time resolution is 15 min. The data time is two years from 2021 to 2022. Among them, the wind speed correction model is established using the data in 2021. Then, this model is applied to correct the wind speed data in the NWP data in 2022. Finally, a wind power prediction model is constructed based on the power data in 2022 and the corrected wind speed data.
[0035] As Figure 5 and Figure 6 shown, the decomposition results of EMD and CEEMDAN for the same wind speed sequence are respectively presented. From Figure 5 and Figure 6 it can be observed that more IMFs are obtained through CEEMDAN decomposition, effectively alleviating the mode mixing of the sequence, and the data processed by the Hilbert transform is presented.
[0036] As Figure 7As shown, it presents the comparison results between the wind speed correction method combining the improved Hilbert-Huang transform and neural network proposed by the present invention and other methods. Among them, the "Corrected Wind Speed with NWP" method directly establishes a mapping relationship between NWP wind speed and measured wind speed through a linear network layer for wind speed correction. The "HHT" method, after applying the Hilbert-Huang transform, uses a linear network layer to establish mapping relationships for different components respectively, and sums up each component to obtain the wind speed correction result. The "iHHT" method is the wind speed correction method proposed by the present invention. From Figure 7 it can be observed that the present invention is closer to the actual wind speed value in the correction result, showing higher correction accuracy. Therefore, applying the wind speed corrected by the present invention to wind power prediction will effectively improve the accuracy of the power prediction result.
[0037] As Figure 8 , Figure 9 , Figure 10 and Figure 11 shown, it presents the prediction results of the present invention compared with different prediction methods for 1 hour, 2 hours, 3 hours, and 4 hours respectively. It can be seen from this that the prediction results of the present invention can better follow the curve change trend and fluctuations compared with other methods.
[0038] As Figure 12 and Figure 13 shown, it presents the RMSE and MAE evaluation index values of the present invention and other methods respectively. It can be obtained from Figure 12 and Figure 13 that the RMSE of the present invention has increased by more than 1% compared with other methods. However, in terms of the MAE evaluation index, only the prediction indexes for the 1st hour, 2nd hour, and 3rd hour are better than other methods, and the MAE index value for the 4th hour is higher than other methods. The MAE index is not sensitive to outliers in the prediction results, and the RMSE value can amplify the errors of outliers. The present invention has the lowest RMSE index, indicating that the prediction results of the present invention are closer to the true value and there are few predicted outliers. The main reason for such a result is that after the wind speed correction by the present invention, the strong learning ability of TSLANet itself makes the prediction results closer to the true value.
[0039] In summary, the present invention detects and processes outliers and missing values in the collected wind power data and wind speed data to obtain initial wind power data and initial wind speed data; adds white noise to the initial wind speed data to generate new wind speed data; performs iterative empirical mode decomposition on the new wind speed data to obtain several modal components; subtracts the several modal components from the initial wind speed data in sequence to obtain a residual component; converts the residual component and the modal components through the Hilbert transform algorithm to obtain new sequence data; inputs the new sequence data into a linear network for processing to obtain corrected wind speed data; constructs a very short-term wind power prediction model based on a time series lightweight adaptive network; inputs the corrected wind speed data and the initial wind power data into the very short-term wind power prediction model for prediction, and outputs very short-term wind power prediction data. The present invention corrects the predicted wind speed of numerical weather prediction by improving the Hilbert-Huang transform, can track the trend fluctuation of the real wind speed, and captures the dynamic relationship between time series data through a time series lightweight adaptive network, thereby improving the prediction accuracy of very short-term wind power. It can reflect the dynamic characteristics of the system and track the future power trend. The present invention has simple calculation, high prediction performance, clear meaning, effective prediction results, and strong practicability.
[0040] It should be noted that the method of the embodiments of the present disclosure can be executed by a single device, such as a computer or a server. The method of this embodiment can also be applied to a distributed scenario and completed by the cooperation of multiple devices. In such a distributed scenario, one of the multiple devices can only execute one or more steps of the method of the embodiments of the present disclosure, and these multiple devices will interact with each other to complete the described method.
[0041] It should be noted that some embodiments of the present disclosure have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be executed in a different order than in the above embodiments and still achieve the desired results. Additionally, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0042] Embodiment 2
[0043] See Figure 14 , Embodiment 2 of the present invention further provides a very short-term wind power prediction device based on an improved Hilbert-Huang transform, including: An original data acquisition and processing module 001, configured to detect and process outliers and missing values in the collected wind power data and wind speed data to obtain initial wind power data and initial wind speed data; A new wind speed data generation module 002 for adding white noise to the initial wind speed data to generate new wind speed data; A new wind speed data modal decomposition module 003 for performing iterative empirical modal decomposition on the new wind speed data to obtain several modal components; successively subtracting several modal components from the initial wind speed data to obtain a residual component; A corrected wind speed data acquisition module 004 for converting the residual component and the modal component through a Hilbert transform algorithm to obtain new sequence data; inputting the new sequence data into a linear network for processing to obtain corrected wind speed data; A very short-term wind power prediction model construction module 005 for constructing a very short-term wind power prediction model based on a time series lightweight adaptive network; A very short-term wind power prediction model prediction module 006 for inputting the corrected wind speed data and the initial wind power data into the very short-term wind power prediction model for prediction and outputting very short-term wind power prediction data.
[0044] In this embodiment, in the new wind speed data generation module 002, the expression for adding white noise to the initial wind speed data to generate the new wind speed data is: ; In the formula, is the new wind speed data; is the initial wind speed data; is the white noise; is the noise amplitude.
[0045] In this embodiment, in the new wind speed data modal decomposition module 003, the calculation formula for the residual component is: ; In the formula, is the residual component; K is the number of iterations; are the modal components obtained at different iteration times.
[0046] In this embodiment, in the very short-term wind power prediction model construction module 005, the time series lightweight adaptive network includes several adaptive spectrum blocks and several interactive convolution blocks; mapping time series data from the time domain to the frequency domain through the adaptive spectrum blocks, and reducing high-frequency noise through an adaptive threshold to highlight frequency domain features; improving the time domain features through the interactive convolution blocks to improve the adaptability of the time domain features to time dynamics.
[0047] In this embodiment, in the ultra-short-term wind power prediction model prediction module 006, the prediction performance of the ultra-short-term wind power prediction model is evaluated by the root mean square error and the mean absolute error of the prediction results; The expression of the root mean square error is: ; In the formula, RMSE is the root mean square error; is the predicted wind power; is the actual wind power; is the installed capacity of the wind farm; N is the number of prediction points; The expression of the mean absolute error is: ; In the formula, MAE is the mean absolute error.
[0048] It should be noted that the information interaction, execution process, etc. between the above system modules, due to being based on the same concept as the method embodiment in Embodiment 1 of the present application, have the same technical effects as the method embodiment of the present application. For specific content, reference can be made to the description in the method embodiment shown above in the present application, and details will not be repeated here.
[0049] Embodiment 3
[0050] Embodiment 3 of the present invention provides a non-transitory computer-readable storage medium, in which program codes of an ultra-short-term wind power prediction method based on an improved Hilbert-Huang transform are stored, and the program codes include instructions for executing the ultra-short-term wind power prediction method based on the improved Hilbert-Huang transform in Embodiment 1 or any possible implementation thereof.
[0051] The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center integrating one or more available media. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid state disk (SSD)), etc.
[0052] Embodiment 4
[0053] Embodiment 4 of the present invention provides an electronic device, including: a memory and a processor; The processor and the memory complete communication with each other through a bus; the memory stores program instructions executable by the processor, and the processor can execute the ultra-short-term wind power prediction method based on the improved Hilbert-Huang transform in Embodiment 1 or any possible implementation thereof by calling the program instructions.
[0054] Specifically, the processor can be implemented by hardware or software. When implemented by hardware, the processor can be a logic circuit, an integrated circuit, etc.; when implemented by software, the processor can be a general-purpose processor that realizes its functions by reading software code stored in a memory. The memory can be integrated in the processor or exist independently outside the processor.
[0055] In the above embodiments, they can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, they can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable systems. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.).
[0056] Obviously, those skilled in the art should understand that the above-mentioned modules or steps of the present invention can be implemented by a general computing system. They can be concentrated on a single computing system or distributed on a network composed of multiple computing systems. Optionally, they can be implemented by program code executable by the computing system. Thus, they can be stored in a storage system and executed by the computing system. And in some cases, the steps shown or described can be executed in a different order than here, or they can be separately fabricated into individual integrated circuit modules, or multiple modules or steps of them can be fabricated into a single integrated circuit module for implementation. In this way, the present invention is not limited to any specific combination of hardware and software.
[0057] Although the present invention has been described in detail above with general descriptions and specific embodiments, based on the present invention, some modifications or improvements can be made, which are obvious to those skilled in the art. Therefore, these modifications or improvements made without departing from the spirit of the present invention fall within the scope of protection of the present invention.
Claims
1. A very short-term wind power prediction method based on an improved Hilbert-Huang transform, characterized in that, Including: Detect and process outliers and missing values in the collected wind power data and wind speed data to obtain initial wind power data and initial wind speed data; Add white noise to the initial wind speed data to generate new wind speed data; Perform iterative empirical mode decomposition on the new wind speed data to obtain several modal components; subtract the several modal components from the initial wind speed data in sequence to obtain a residual component; Convert the residual component and the modal components through the Hilbert transform algorithm to obtain new sequence data; Input the new sequence data into a linear network for processing to obtain corrected wind speed data; Construct a very short-term wind power prediction model based on a time series lightweight adaptive network; Input the corrected wind speed data and the initial wind power data into the very short-term wind power prediction model for prediction, and output very short-term wind power prediction data.
2. The ultra-short-term wind power prediction method based on the improved Hilbert-Huang transform according to claim 1, characterized in that The expression for adding white noise to the initial wind speed data to generate the new wind speed data is: ; In the formula, is the fresh air velocity data; is the initial velocity data; is white noise; is the noise amplitude.
3. The ultra-short-term wind power prediction method based on the improved Hilbert-Huang transform according to claim 2, characterized in that The calculation formula for the residual component is: ; wherein, is the residual component; K is the number of iterations; is the modal component obtained at different numbers of iterations.
4. The ultra-short-term wind power prediction method based on the improved Hilbert-Huang transform according to claim 3, wherein The time series lightweight adaptive network includes several adaptive spectral blocks and several interactive convolution blocks; map time series data from the time domain to the frequency domain through the adaptive spectral blocks, and reduce high-frequency noise through an adaptive threshold to highlight frequency domain features; improve the time domain features through the interactive convolution blocks to enhance the adaptability of the time domain features to time dynamics.
5. The ultra-short-term wind power prediction method based on the improved Hilbert-Huang transform according to claim 4, wherein Evaluate the prediction performance of the very short-term wind power prediction model through the root mean square error and mean absolute error of the prediction results; The expression for the root mean square error is: ; Wherein, RMSE is the root mean square error; is the predicted wind power; is the actual wind power; is the installed capacity of the wind farm; N is the number of prediction points; The expression for the mean absolute error is: ; In the formula, MAE is the mean absolute error.
6. The ultra-short-term wind power prediction device based on the improved Hilbert-Huang transform adopts the ultra-short-term wind power prediction method based on the improved Hilbert-Huang transform described in any one of claims 1-5, and is characterized in that Including: An original data acquisition and processing module, which is used to detect and process outliers and missing values in the collected wind power data and wind speed data to obtain initial wind power data and initial wind speed data; A new wind speed data generation module, which is used to add white noise to the initial wind speed data to generate new wind speed data; A new wind speed data modal decomposition module, which is used to perform iterative empirical mode decomposition on the new wind speed data to obtain several modal components; subtract the several modal components from the initial wind speed data in sequence to obtain a residual component; A corrected wind speed data acquisition module, which is used to convert the residual component and the modal components through the Hilbert transform algorithm to obtain new sequence data; Input the new sequence data into a linear network for processing to obtain corrected wind speed data; A very short-term wind power prediction model construction module, which is used to construct a very short-term wind power prediction model based on a time series lightweight adaptive network; A very short-term wind power prediction model prediction module, which is used to input the corrected wind speed data and the initial wind power data into the very short-term wind power prediction model for prediction, and output very short-term wind power prediction data.
7. The ultra-short-term wind power prediction device based on the improved Hilbert-Huang transform according to claim 6, characterized in that In the new wind speed data generation module, the expression for adding white noise to the initial wind speed data to generate the new wind speed data is: ; In the formula, is the fresh air velocity data; is the initial velocity data; is white noise; is the noise amplitude.
8. The ultra-short-term wind power prediction device based on the improved Hilbert-Huang transform according to claim 7, wherein In the new wind speed data modal decomposition module, the calculation formula for the residual component is: ; wherein, is the residual component; K is the number of iterations; are the modal components obtained at different numbers of iterations.
9. The ultra-short-term wind power prediction device based on the improved Hilbert-Huang transform according to claim 8, characterized in that In the construction module of the ultra-short-term wind power prediction model, the time series lightweight adaptive network includes a number of adaptive spectral blocks and a number of interactive convolution blocks; the time series data is mapped from the time domain to the frequency domain through the adaptive spectral blocks, and high-frequency noise is reduced through an adaptive threshold to highlight the frequency domain features; the time domain features are improved through the interactive convolution blocks to enhance the adaptability of the time domain features to time dynamics.
10. The ultra-short-term wind power prediction device based on the improved Hilbert-Huang transform according to claim 9, characterized in that In the prediction module of the ultra-short-term wind power prediction model, the prediction performance of the ultra-short-term wind power prediction model is evaluated through the root mean square error and the mean absolute error of the prediction results; The expression of the root mean square error is: ; Wherein, RMSE is the root mean square error; is the predicted wind power; is the actual wind power; is the installed capacity of the wind farm; N is the number of prediction points; The expression of the mean absolute error is: ; Wherein, MAE is the mean absolute error.
Citation Information
Cited By
Cyclic error improved wind speed prediction super-capacity energy storage grid connection method and related product
CN120511759A
