Hydraulic power plant AGC frequency modulation method and device based on multi-mode control strategy

Abstract

The invention provides a hydraulic power plant AGC frequency modulation method and device based on a multi-mode control strategy. The method comprises the following steps: collecting unit parameters of a hydraulic power plant unit at a first moment in real time; determining a current operation condition mode of the hydraulic power plant unit based on the current water head height value and the current water flow value, and matching a target frequency modulation control model based on the current operation condition mode; inputting the historical power grid frequency into the power grid frequency prediction model to obtain a future power grid frequency, output by the power grid frequency prediction model, of the hydraulic power plant unit at a second moment; inputting the future power grid frequency into the target frequency modulation control model to obtain a power adjustment amount output by the target frequency modulation control model; and adjusting the current unit output of the hydraulic power plant unit based on the power adjustment amount to obtain the target unit output of the hydraulic power plant unit at the second moment. According to the invention, the frequency modulation performance and frequency modulation response of AGC frequency modulation of the hydraulic power plant are improved, and stable operation of a power system is ensured.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method and apparatus for frequency regulation of hydropower plants based on a multimodal control strategy (AGC). Background Technology

[0002] In the power system, hydropower plants are an important source of power supply, and their Automatic Generation Control (AGC) frequency regulation function plays a key role in maintaining the stability of the power grid frequency.

[0003] Traditional AGC frequency regulation methods for hydropower plants have several drawbacks: First, they employ a single control mode, making it difficult to adapt to the complex and variable operating conditions of hydropower plants. For example, under different head, flow, and load conditions, the characteristics of the turbine will change significantly, and a single control mode cannot flexibly adjust control parameters, leading to a decline in frequency regulation performance. Second, traditional AGC frequency regulation methods for hydropower plants are poorly adaptable to grid frequency fluctuations. When faced with rapid or frequent changes in grid frequency, due to the limitations of the control algorithm, the frequency regulation response of the hydropower plant often exhibits lag, failing to track grid frequency changes in a timely and effective manner, which may cause the grid frequency to deviate from its rated value for an extended period. Summary of the Invention

[0004] This invention provides a hydropower plant AGC frequency regulation method and apparatus based on a multimodal control strategy, which improves the frequency regulation performance and frequency regulation response of hydropower plant AGC frequency regulation and ensures the stable operation of the power system.

[0005] In a first aspect, the present invention provides a hydropower plant AGC frequency regulation method based on a multimodal control strategy, comprising: Real-time acquisition of hydropower plant unit parameters at the first moment; the unit parameters include the current water head, current water flow, current unit output, and historical grid frequency; Based on the current head height and the current flow rate, the current operating mode of the hydropower plant unit is determined, and a target frequency regulation control model is matched based on the current operating mode. The historical power grid frequency is input into the power grid frequency prediction model to obtain the future power grid frequency of the hydropower plant unit at the second moment, which is output by the power grid frequency prediction model; the power grid frequency prediction model is trained based on the sample power grid frequency and its corresponding labeled data; The future grid frequency is input into the target frequency regulation control model to obtain the power adjustment amount output by the target frequency regulation control model; the target frequency regulation control model is trained based on the first power result corresponding to the rated grid frequency under the current operating condition mode, and the sample grid frequency and its corresponding second power result; The current unit output of the hydropower plant is adjusted based on the power adjustment amount to obtain the target unit output of the hydropower plant at the second time moment; the first time moment represents the current operating time of the hydropower plant unit, and the second time moment represents the next time moment after the first time moment.

[0006] According to the embodiment of the present invention, the training process of the target frequency control model for the hydropower plant AGC frequency regulation method based on a multimodal control strategy is as follows: The sample power grid frequency is divided into training sample data and test sample data. The first grid frequency difference for each training sample data is obtained by calculating the difference between each training sample data and the rated grid frequency. The difference between the second power result and the first power result corresponding to each training sample data is calculated to obtain the first power difference of each training sample data. Based on the first grid frequency difference and the first power difference of each training sample data, the first model parameters to be optimized are determined; The target frequency regulation control model is obtained by training the model based on the first model parameters to be optimized, the second grid frequency difference between each test sample data and the rated grid frequency, and the second power difference between the second power result and the first power result corresponding to each test sample data.

[0007] The frequency regulation method for hydropower plants based on a multimodal control strategy according to an embodiment of the present invention includes a model training process based on first model parameters to be optimized, a second grid frequency difference between each test sample data and the rated grid frequency, and a second power difference between the second power result and the first power result corresponding to each test sample data, to obtain the target frequency regulation control model, comprising: Based on the parameters of the first model to be optimized and the first test sample data and the rated grid frequency, the predicted power difference of the first test sample data is determined, and based on the parameters of the first model to be optimized and the second test sample data and the rated grid frequency, the predicted power difference of the second test sample data is determined. Based on the first preset loss function and the predicted power difference and the second power difference of the first test sample data, the loss function value of the first test sample data is determined, and based on the second preset loss function and the predicted power difference and the second power difference of the second test sample data, the loss function value of the second test sample data is determined. The parameters of the first model to be optimized are adjusted based on the difference between the loss function values ​​of the first test sample data and the loss function values ​​of the second test sample data. The difference between the loss function values ​​of two adjacent test sample data is obtained based on the adjusted parameters of the first model to be optimized until the first consecutive number of differences are less than or equal to the first preset threshold. The first optimal model parameters are obtained, and the training of the target frequency modulation control model is completed. The first preset loss function is expressed as: ; in, This represents the loss function value for the i-th test sample data. This represents the difference in predicted power for the i-th test sample data. This represents the second power difference of the i-th test sample data.

[0008] According to the multimodal control strategy-based AGC frequency regulation method for hydropower plants provided by embodiments of the present invention, the training process of the power grid frequency prediction model is as follows: The sample power grid frequency is divided according to a sliding window with a preset step size to obtain multiple training sample sequences and multiple test sample sequences; Obtain the first target grid frequency for each training sample sequence; the first target grid frequency for each training sample sequence is the next grid frequency for each training sample sequence. Based on the first target grid frequency of each training sample sequence and the mean grid frequency of each training sample sequence, the third grid frequency difference of each training sample sequence is determined. The mean value of the grid frequency difference is obtained based on the third grid frequency difference of each training sample sequence; The power grid frequency prediction model is obtained by training the model based on the mean difference of the power grid frequency, each test sample sequence and its second target power grid frequency; the second target power grid frequency of each test sample sequence is the next power grid frequency of each test sample sequence.

[0009] According to an embodiment of the present invention, a hydropower plant AGC frequency regulation method based on a multimodal control strategy is provided, wherein the power grid frequency prediction model is obtained by training a model based on the mean of the power grid frequency difference, each test sample sequence and its second target power grid frequency, including: Based on the average grid frequency difference, the parameters of the second model to be optimized, and the average grid frequency of the first test sample sequence, the final grid frequency of the first test sample sequence is determined. Based on the second preset loss function and the final grid frequency and the second target grid frequency of the first test sample sequence, the loss function value of the first test sample sequence is determined; The second model parameters to be optimized are adjusted based on the loss function value of the first test sample sequence. The loss function value of the second test sample sequence is obtained based on the adjusted second model parameters to be optimized. This process continues until the loss function value of the second consecutive number of test sample sequences is less than or equal to the second preset threshold. The second optimal model parameters are then obtained, and the training of the power grid frequency prediction model is completed. The second preset loss function is expressed as: ; in, This represents the loss function value of the i-th test sample sequence. This represents the final grid frequency of the i-th test sample sequence. This represents the second target grid frequency of the i-th test sample sequence. This represents the mean frequency of the i-th test sample sequence.

[0010] According to an embodiment of the present invention, the AGC frequency regulation method for hydropower plants based on a multimodal control strategy includes matching a target frequency regulation control model based on the current operating condition mode, comprising: Based on the current operating condition mode, combined with the current head height value and the current water flow rate value, determine the operating condition mode characteristic value under the current operating condition mode; Based on each frequency control model in the preset model library, combined with the current head height value and the current water flow value, the model adaptation characteristic value of each frequency control model under the current operating condition is determined; Based on the characteristic values ​​of the operating condition mode and the model adaptation characteristic values ​​of each frequency modulation control model, the matching degree value of each frequency modulation control model to the current operating condition mode is determined. The frequency modulation control model corresponding to the maximum matching degree value is determined as the target frequency modulation control model.

[0011] According to an embodiment of the present invention, the AGC frequency regulation method for hydropower plants based on a multimodal control strategy, wherein determining the current operating mode of the hydropower plant unit based on the current head height value and the current flow rate value includes: Based on the current head height value and the maximum and minimum head height values ​​of the hydropower plant unit, a first operating condition discrimination value is determined. Based on the current water flow value and the maximum and minimum water flow values ​​of the hydropower plant unit, a second operating condition discrimination value is determined. Based on the first operating condition discrimination value and the second operating condition discrimination value, a comprehensive operating condition discrimination value is determined; Based on the preset fuzzy operating condition discrimination strategy and the comprehensive operating condition discrimination value, the current operating condition mode of the hydropower plant unit is determined.

[0012] Secondly, the present invention also provides a hydropower plant AGC frequency regulation device based on a multimodal control strategy, applied to the hydropower plant AGC frequency regulation method based on a multimodal control strategy as described in the first aspect, wherein the hydropower plant AGC frequency regulation device based on a multimodal control strategy comprises: The data acquisition module is used to collect the unit parameters of the hydropower plant in real time at the first moment; the unit parameters include the current water head, the current water flow, the current unit output, and the historical grid frequency; The determination module is used to determine the current operating condition mode of the hydropower plant unit based on the current head height value and the current water flow value, and to match the target frequency regulation control model based on the current operating condition mode. The first prediction module is used to input the historical power grid frequency into the power grid frequency prediction model to obtain the future power grid frequency of the hydropower plant unit at the second moment, which is output by the power grid frequency prediction model; the power grid frequency prediction model is trained based on the sample power grid frequency and its corresponding labeled data. The second prediction module is used to input the future grid frequency into the target frequency regulation control model to obtain the power adjustment amount output by the target frequency regulation control model; the target frequency regulation control model is trained based on the first power result corresponding to the rated grid frequency under the current operating condition mode, and the sample grid frequency and its corresponding second power result; The adjustment module is used to adjust the current unit output of the hydropower plant unit based on the power adjustment amount to obtain the target unit output of the hydropower plant unit at the second time moment; the first time moment represents the current operating time of the hydropower plant unit, and the second time moment represents the next time moment after the first time moment.

[0013] Thirdly, the present invention also provides an electronic device, comprising: a memory for storing computer software programs; and a processor for reading and executing the computer software programs, thereby implementing the hydropower plant AGC frequency regulation method based on the multimodal control strategy as described above.

[0014] Fourthly, the present invention also provides a non-transitory computer-readable storage medium storing a computer software program, which, when executed by a processor, implements the hydropower plant AGC frequency regulation method based on any of the above-described multimodal control strategies.

[0015] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the hydropower plant AGC frequency regulation method based on a multimodal control strategy as described above.

[0016] The hydropower plant AGC frequency regulation method based on a multimodal control strategy provided in this invention divides different operating conditions according to the water head and flow rate. Then, through the multimodal frequency regulation control model corresponding to each operating condition, it accurately predicts the power adjustment amount at the next moment based on the future grid frequency. Therefore, the control parameters can be flexibly adjusted through the multimodal control strategy, improving the frequency regulation performance of the hydropower plant AGC. At the same time, the grid frequency prediction model can accurately predict the future grid frequency of the hydropower plant units at the next moment. Therefore, the grid frequency of the hydropower plant units at the next moment can be predicted in advance, and the unit output of the hydropower plant units at the next moment can be planned in advance based on the grid frequency. This effectively reduces the lag in frequency regulation response, improves the frequency regulation response of the hydropower plant AGC, and ensures the stable operation of the power system. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the AGC frequency regulation method for hydropower plants based on a multimodal control strategy provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of the hydropower plant AGC frequency regulation device based on a multimodal control strategy provided in an embodiment of the present invention; Figure 3 An embodiment diagram of the electronic device provided in this invention; Figure 4 An embodiment diagram of a computer-readable storage medium provided in accordance with the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0020] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0021] Optional, see below Figure 1 As shown, Figure 1 This is a flowchart illustrating the AGC frequency regulation method for hydropower plants based on a multimodal control strategy provided by the present invention. In this embodiment, the executing entity of the AGC frequency regulation method for hydropower plants based on a multimodal control strategy is the AGC frequency regulation device. Therefore, the AGC frequency regulation method for hydropower plants based on a multimodal control strategy includes: Step 10: Real-time acquisition of the unit parameters of the hydropower plant at the first moment.

[0022] In this embodiment of the invention, a pressure sensor is installed at the water intake pipeline or turbine inlet, an electromagnetic flow meter, an ultrasonic flow meter, or other equipment is installed on the water delivery pipeline, and a power sensor is installed in the hydropower plant unit.

[0023] The first moment represents the current operating moment of the hydropower plant unit, and the unit parameters include the current water head, current water flow, current unit output, and historical grid frequency.

[0024] Therefore, the AGC frequency modulation device collects pressure data by pressure sensor and calculates the current head height of the hydropower plant unit by combining the water level difference. It measures the water flow of the hydropower plant unit by electromagnetic flow meter or ultrasonic flow meter, obtains the current unit output by measuring the generator output power by power sensor, and obtains the historical grid frequency of the hydropower plant unit by retrieving the database of the power system.

[0025] To calculate the current head height, considering the impact of local and frictional energy losses on measurement accuracy, the collected head height value can be adjusted to obtain the current head height value. The specific adjustment formula is as follows: ; in, This indicates the current water head height value. This represents the collected water head height value. Indicates the water flow velocity. Indicates the length of the water supply pipeline. Indicates the pipe diameter. and It is an empirical coefficient determined based on factors such as pipe material and roughness.

[0026] Step 20: Based on the current head height and current flow rate, determine the current operating mode of the hydropower plant unit, and match the target frequency regulation control model based on the current operating mode.

[0027] Furthermore, the AGC frequency modulation device comprehensively judges the current operating mode of the hydropower plant unit based on the calculated current head height and current water flow value, as described in steps 201 to 204. Different combinations of head height and water flow value will correspond to different operating states. For example, high head and high flow may be a full-load and high-efficiency operating condition, while low head and low flow may be a low-load operating condition.

[0028] Furthermore, for each determined current operating condition mode, the AGC frequency regulation device matches the target frequency regulation control model for the current operating condition mode from the pre-built frequency regulation control model library stored in the AGC frequency regulation device, as described in steps 205 to 208. The target frequency regulation control model is trained based on the first power result corresponding to the rated grid frequency under the current operating condition mode, and the sample grid frequency and its corresponding second power result. In one embodiment, the pre-trained model includes a graph neural network, a front-feedback neural network, and a back-feedback neural network. If the pre-trained model is a front-feedback neural network, it can be understood that the front-feedback neural network is trained based on the first power result corresponding to the rated grid frequency under the current operating condition mode, and the sample grid frequency and its corresponding second power result to obtain the target frequency regulation control model for the current operating condition mode.

[0029] Step 30: Input the historical grid frequency into the grid frequency prediction model to obtain the future grid frequency of the hydropower plant units at the second moment, which is output by the grid frequency prediction model.

[0030] The AGC frequency modulation device in this embodiment of the invention also stores a pre-built power grid frequency prediction model. The power grid frequency prediction model is trained based on the sample power grid frequency and its corresponding labeled data. The specific training process of the power grid frequency prediction model is described in steps 301 to 305.

[0031] Therefore, the AGC frequency regulation device inputs the historical grid frequency into the grid frequency prediction model to obtain the future grid frequency of the hydropower plant units at the second moment, where the second moment represents the next moment after the first moment, i.e., the next moment after the current operating moment. The processing formula of the grid frequency prediction model can be expressed as: ; in, Indicates the future power grid frequency, in units of ,hertz; This represents the average historical power grid frequency. This represents the average frequency difference of the power grid. This represents the optimal model parameters for the power grid frequency prediction model.

[0032] Step 40: Input the future grid frequency into the target frequency regulation control model to obtain the power adjustment amount output by the target frequency regulation control model.

[0033] Furthermore, the AGC frequency modulation device inputs the future grid frequency into the target frequency modulation control model to obtain the power adjustment output of the target frequency modulation control model. The specific training process of the target frequency modulation control model is described in steps 401 to 405. The processing formula of the target frequency modulation control model can be expressed as follows: ; in, Indicates the amount of power adjustment; This represents the optimal model parameters of the target frequency modulation control model, in units of... kilowatts per hertz; This indicates the rated grid frequency under the current operating conditions.

[0034] Step 50: Adjust the current unit output of the hydropower plant based on the power adjustment amount to obtain the target unit output of the hydropower plant at the second moment.

[0035] Furthermore, the AGC frequency regulation device sums the power adjustment amount with the current unit output of the power plant to obtain the target unit output of the hydropower plant at the second moment.

[0036] In one embodiment, the current unit output of the power plant is 50,000 kW, and the power adjustment amount obtained based on the future grid frequency is -2,000. Therefore, the target unit output of the hydropower plant at the second moment is 50,000 + (-2,000) = 48,000 kW.

[0037] This invention divides different operating modes based on water head and flow rate. Then, using a multi-modal frequency control model corresponding to each operating mode, it accurately predicts the power adjustment amount for the next moment based on the future grid frequency. Therefore, control parameters can be flexibly adjusted through a multi-modal control strategy, improving the frequency regulation performance of the hydropower plant's AGC (Automatic Generation Control) system. Simultaneously, the grid frequency prediction model accurately predicts the future grid frequency of the hydropower plant units for the next moment, allowing for advance prediction of the grid frequency and planning of the unit output in advance. This effectively reduces the lag in frequency regulation response, improves the frequency regulation response of the hydropower plant's AGC, and ensures the stable operation of the power system.

[0038] In one embodiment, steps 201 to 204 are described as follows: Step 201: Determine the first operating condition discrimination value based on the current water head height value and the maximum and minimum water head height values ​​of the hydropower plant units.

[0039] Specifically, the AGC frequency modulation device acquires the maximum and minimum head height values ​​of the hydropower plant units. Based on the current head height value and the maximum and minimum head height values ​​of the hydropower plant units, it determines the first operating condition discrimination value. The specific calculation formula is as follows: ; in, This represents the first working condition discrimination value. This indicates the current water head height value. This represents the minimum head height value. This indicates the maximum water head height value.

[0040] Step 202: Determine the second operating condition discrimination value based on the current water flow value and the maximum and minimum water flow values ​​of the hydropower plant units.

[0041] Furthermore, the AGC frequency modulation device acquires the maximum and minimum water flow values ​​of the hydropower plant units. Based on the current water flow value and the maximum and minimum water flow values ​​of the hydropower plant units, it determines the second operating condition discrimination value. The specific calculation formula is as follows: ; in, This indicates the discrimination value for the second working condition. This indicates the current water flow rate. This represents the minimum water flow rate. This indicates the maximum water flow rate.

[0042] Step 203: Determine the comprehensive working condition discrimination value based on the first working condition discrimination value and the second working condition discrimination value.

[0043] Furthermore, the AGC frequency modulation device calculates and determines the comprehensive operating condition discrimination value based on the first operating condition discrimination value and the second operating condition discrimination value. The specific calculation formula is as follows: ; in, This represents the comprehensive working condition discrimination value. This represents the base of the exponential function, usually e = 2.718.

[0044] Step 204: Based on the preset fuzzy operating condition discrimination strategy and comprehensive operating condition discrimination value, determine the current operating condition mode of the hydropower plant unit.

[0045] Furthermore, the preset fuzzy working condition discrimination strategy in the embodiments of the present invention, such as when the comprehensive working condition discrimination value is in When the value is in the range, it is determined to be a low operating condition mode; when the comprehensive operating condition discrimination value is in the range, it is determined to be a low operating condition mode. When the value is within the specified range, it is determined to be in medium operating condition mode; when the comprehensive operating condition discrimination value is within the specified range... When the value falls within a certain range, it is determined to be in a high-operating-condition mode. Therefore, the AGC frequency regulation device determines the range in which the comprehensive operating condition discrimination value falls, and determines the current operating condition mode of the hydropower plant unit based on the operating condition mode corresponding to that range.

[0046] In one embodiment, the comprehensive operating condition discrimination value is 0.465, which is considered to be in the range of 0.465. Therefore, the current operating mode of the hydropower plant unit is determined to be the medium operating mode.

[0047] This invention divides different operating modes according to the water head and water flow values, enabling the subsequent multi-mode frequency regulation control model corresponding to different operating modes to accurately predict the power adjustment amount at the next moment based on the future grid frequency at the next moment. By flexibly adjusting the control parameters through the multi-mode control strategy, the frequency regulation performance of the hydropower plant's AGC frequency regulation is improved, ensuring the stable operation of the power system.

[0048] In one embodiment, steps 205 to 208 are described as follows: Step 205: Based on the current operating condition mode, combined with the current head height value and the current water flow value, determine the operating condition mode characteristic value under the current operating condition mode.

[0049] In this embodiment of the invention, the preset model library is a matching matrix M. Matching matrix M is a two-dimensional matrix. The rows of the two-dimensional matrix represent different operating conditions, and the columns represent different frequency modulation control models. In one embodiment, there are n operating conditions and m frequency modulation control models, so the matching matrix M is an n*m two-dimensional matrix. The elements in the two-dimensional matrix... Indicates the first Type of operating mode and the first The matching degree value of the frequency modulation control model, the range of the matching degree value is: , 1 indicates a complete mismatch, and 1 indicates a complete match.

[0050] In one embodiment, there are 3 operating condition modes, where 1 represents the first operating condition mode (low operating condition mode), 2 represents the second operating condition mode (medium operating condition mode), and 3 represents the third operating condition mode (high operating condition mode); and 4 frequency modulation control models, where 1 represents the first frequency modulation control model (frequency modulation control model 1), 2 represents the second frequency modulation control model (frequency modulation control model 2), 3 represents the third frequency modulation control model (frequency modulation control model 3), and 4 represents the fourth frequency modulation control model (frequency modulation control model 4). The matching matrix M can then be expressed as: ; in, This represents the matching degree value between the low-operating-condition mode and the frequency modulation control model 1, that is, the matching degree value of the frequency modulation control model 1 in the low-operating-condition mode. The other elements are similar.

[0051] Optionally, the AGC frequency modulation device calculates the current head height and current flow rate based on the current operating condition mode and the corresponding characteristic function to obtain the operating condition mode characteristic value under the current operating condition mode. The specific characteristic function is as follows: ; in, Indicates the first Operating condition characteristic values ​​under various operating conditions Indicates the first The preset mode parameters for each operating condition mode.

[0052] In one embodiment, for the low operating condition mode, , , , , The characteristic values ​​of the operating mode under low operating conditions are: .

[0053] Step 206: Based on each frequency control model in the preset model library, combined with the current head height value and the current water flow value, determine the model adaptation characteristic value of each frequency control model under the current operating condition.

[0054] Furthermore, each frequency control model in the preset model library, combined with its corresponding fitness function, calculates the model fitness characteristic value under the current operating condition by considering the current head height and current flow rate. Therefore, the AGC frequency control device can obtain the model fitness characteristic value of each frequency control model under the current operating condition. The fitness function can be expressed as: ; in, Indicates the first The model adaptation eigenvalues ​​of the frequency modulation control model. Indicates the first The preset model parameters of a frequency modulation control model.

[0055] In one embodiment, for frequency modulation control model 1, , , , , The model adaptation eigenvalues ​​of frequency modulation control model 1 are: .

[0056] Step 207: Based on the characteristic values ​​of the operating condition mode and the model adaptation characteristic values ​​of each frequency regulation control model, determine the matching degree value of each frequency regulation control model to the current operating condition mode.

[0057] Furthermore, the AGC frequency modulation device determines the matching degree value of each frequency modulation control model to the current operating condition mode based on the characteristic values ​​of the operating condition mode and the model adaptation characteristic values ​​of each frequency modulation control model. The specific calculation formula is as follows: ; in, Indicates the first Type of operating mode and the first The matching degree value of the frequency modulation control model, i.e., the first The frequency modulation control model in the first Matching degree value under various operating conditions.

[0058] Continuing with the above embodiments, the matching degree value of frequency modulation control model 1 in low operating condition mode is: .

[0059] Step 208: Determine the frequency modulation control model corresponding to the maximum matching degree value as the target frequency modulation control model.

[0060] Furthermore, for each operating condition mode The AGC frequency modulation device modulates the matching matrix M for the first... Iterate through the matching values ​​in the rows, and find the first matching value. Maximum matching value in a row and the maximum matching value The corresponding frequency modulation control model is determined to be the same as the operating condition mode. The best-fit target frequency modulation control model. In one embodiment, after calculating the matching degree values ​​between the low-condition mode and all frequency modulation control models, a row of matching degree values ​​is obtained. The maximum matching degree is then Matching score The corresponding frequency modulation control model (frequency modulation control model 3) is the target frequency modulation control model that best matches the low operating condition mode.

[0061] This invention identifies the optimal frequency regulation control model by matching different operating conditions. This allows for accurate prediction of the power adjustment amount at the next moment based on the future grid frequency at the next moment through the multi-mode frequency regulation control model corresponding to different operating conditions. By flexibly adjusting control parameters through multi-mode control strategies, the frequency regulation performance of hydropower plant AGC frequency regulation is improved, ensuring the stable operation of the power system.

[0062] In one embodiment, steps 301 to 305 are described as follows: Step 301: Divide the sample power grid frequency into multiple training sample sequences and multiple test sample sequences according to a preset step size sliding window.

[0063] Specifically, the AGC frequency modulation device divides the sample grid frequency into multiple training sample sequences and multiple test sample sequences according to a preset step size sliding window. The preset step size is set according to the number of sample data, such as 3 or 4, and the ratio between the number of training sample sequences and the number of test sample sequences is set according to the actual situation.

[0064] In one embodiment, a preset step size of 3 is used as a sliding window, and the obtained sample grid frequency is... Therefore, the sample sequence obtained after partitioning is ; ; ; ; ; ; ; ; ; ; ; ; ; .Will ; ; ; ; ; ; ; ; ; Divide into training sample sequences, ; ; ; The sequence was divided into test sample sequences.

[0065] Step 302: Obtain the first target grid frequency for each training sample sequence.

[0066] Furthermore, the AGC frequency modulation device acquires the first target grid frequency for each training sample sequence. The first target grid frequency for each training sample sequence is the next grid frequency for each training sample sequence. The first target grid frequency can be understood as the labeled data of the training sample sequence.

[0067] Continuing with the above embodiments, for the first training sample sequence Its first target grid frequency is For the second training sample sequence Its first target grid frequency is For the third training sample sequence Its first target grid frequency is For the fourth training sample sequence Its first target grid frequency is Similarly, the first target grid frequency of all training sample sequences can be obtained.

[0068] Step 303: Based on the first target grid frequency of each training sample sequence and the mean grid frequency of each training sample sequence, determine the third grid frequency difference of each training sample sequence.

[0069] Furthermore, the AGC frequency modulation device determines the third grid frequency difference for each training sample sequence based on the first target grid frequency and the average grid frequency of each training sample sequence. That is, the third grid frequency difference for each training sample sequence is obtained by subtracting the average grid frequency of each training sample sequence from the first target grid frequency of each training sample sequence.

[0070] Continuing with the above embodiments, for the first training sample sequence Its average grid frequency is Therefore, the third grid frequency difference of the first training sample sequence is... For the second training sample sequence Its average grid frequency is Therefore, the third grid frequency difference of the second training sample sequence is... For the third training sample sequence Its average grid frequency is Therefore, the third grid frequency difference of the second training sample sequence is... Similarly, the third grid frequency difference of all training sample sequences can be obtained.

[0071] Step 304: Obtain the mean value of the grid frequency difference based on the third grid frequency difference of each training sample sequence.

[0072] Furthermore, the AGC frequency modulation device calculates the mean of the third grid frequency difference for each training sample sequence to obtain the mean of the grid frequency difference for all training sample sequences.

[0073] Step 305: Based on the mean of the power grid frequency difference, each test sample sequence and its second target power grid frequency, the model is trained to obtain the power grid frequency prediction model.

[0074] Furthermore, the AGC frequency modulation device trains a model based on the average grid frequency difference, each test sample sequence, and its second target grid frequency to obtain a grid frequency prediction model. The second target grid frequency of each test sample sequence is the next grid frequency of each test sample sequence. Here, the second target grid frequency can be understood as the labeled data of the test sample sequence, as analyzed in detail below: Optionally, the AGC frequency modulation device calculates the average grid frequency of the first test sample sequence, and sums the average grid frequency difference, the second model parameter to be optimized, and the average grid frequency of the first test sample sequence to obtain the final grid frequency of the first test sample sequence. In this embodiment of the invention, the second model parameter to be optimized is a randomly assigned parameter.

[0075] Furthermore, the AGC frequency modulation device inputs the final grid frequency and the second target grid frequency of the first test sample sequence into the second preset loss function to obtain the loss function value of the first test sample sequence, wherein the second preset loss function is expressed as: ; in, This represents the loss function value of the i-th test sample sequence. This represents the final grid frequency of the i-th test sample sequence. This represents the second target grid frequency of the i-th test sample sequence. This represents the mean frequency of the i-th test sample sequence.

[0076] If the loss function value of the first test sample sequence is determined to be less than or equal to the second preset threshold, the AGC frequency modulation device continues to sum the second model parameters to be optimized, the average grid frequency difference, and the average grid frequency of the second test sample sequence to obtain the final grid frequency of the second test sample sequence. The AGC frequency modulation device inputs the final grid frequency of the second test sample sequence and the second target grid frequency into the second preset loss function to obtain the loss function value of the second test sample sequence. This process is repeated until the loss function value of a second consecutive number of test sample sequences is less than or equal to the second preset threshold, thus obtaining the second optimal model parameters and completing the training of the grid frequency prediction model. The second preset threshold and the second consecutive number are set according to actual conditions, such as 50, 60, 100, etc.

[0077] If the loss function value of the first test sample sequence is greater than the second preset threshold, the AGC frequency modulation device adjusts the parameters of the second model to be optimized, and sums the adjusted second model parameters, the average grid frequency difference, and the average grid frequency of the second test sample sequence to obtain the final grid frequency of the second test sample sequence. The AGC frequency modulation device inputs the final grid frequency of the second test sample sequence and the second target grid frequency into the second preset loss function to obtain the loss function value of the second test sample sequence. This process is repeated until the loss function value of a second consecutive number of test sample sequences is less than or equal to the second preset threshold, thus obtaining the second optimal model parameters and completing the training of the grid frequency prediction model.

[0078] In one embodiment, the processing formula of the obtained power grid frequency prediction model can be expressed as: ; in, Indicates the future power grid frequency. This represents the average historical power grid frequency to be input. This represents the average frequency difference of the power grid. This represents the second optimal model parameters.

[0079] It should be noted that, in this embodiment of the invention, the second target grid frequency for the last test sample sequence is the last grid frequency in the last test sample sequence.

[0080] In one embodiment, the average value of the grid frequency difference Second optimal model parameters The power grid frequency of the hydropower plant units at the first moment is The power grid frequency of the hydropower plant units two moments before the first moment is: and Therefore, the grid frequency of the hydropower plant units at the second moment is: .

[0081] This invention trains a power grid frequency prediction model, which can then accurately predict the future power grid frequency of hydropower plant units at the next moment. This allows for advance prediction of the power grid frequency of hydropower plant units at the next moment, enabling the planning of the power output of hydropower plant units in advance based on the power grid frequency. This effectively reduces the lag in frequency regulation response, improves the frequency regulation response of hydropower plant AGC, and ensures the stable operation of the power system.

[0082] In one embodiment, steps 401 to 405 are described as follows: Step 401: Divide the sample power grid frequency into training sample data and test sample data.

[0083] Optionally, the AGC frequency modulation device divides the sample grid frequency into training sample data and test sample data. It can be understood that each training sample data and each test sample data is grid frequency data. The ratio of the number of training sample data to the number of test sample data is set according to the actual situation, such as a ratio of 8:2, 7:3, etc.

[0084] Step 402: Calculate the difference between each training sample data and the rated grid frequency to obtain the first grid frequency difference for each training sample data.

[0085] Furthermore, the AGC frequency modulation device calculates the difference between each training sample data and the rated grid frequency to obtain the first grid frequency difference of each training sample data. In this embodiment of the invention, the rated grid frequency is subtracted from each training sample data to obtain the first grid frequency difference of each training sample data, that is, the first grid frequency difference of each training sample data = rated grid frequency - each training sample data.

[0086] Step 403: Calculate the difference between the second power result and the first power result corresponding to each training sample data to obtain the first power difference for each training sample data.

[0087] Furthermore, the AGC frequency modulation device calculates the first power difference for each training sample data based on the second power result and the first power result corresponding to each training sample data. In this embodiment of the invention, the first power result is subtracted from the second power result corresponding to each training sample data to obtain the first power difference for each training sample data, that is, the first power difference = the first power result - the second power result.

[0088] In one embodiment, for each training sample data ,in, Indicates the first Training sample data, training sample data With rated power grid frequency Perform interpolation to obtain the data for each training sample. First grid frequency difference ,therefore, . Training sample data The corresponding second power result Calculate the difference between the first power result and the second power result. To obtain the data for each training sample First power difference .

[0089] Step 404: Determine the first model parameters to be optimized based on the first grid frequency difference and the first power difference of each training sample data.

[0090] Furthermore, the AGC frequency modulation device sums up the first grid frequency difference of each training sample data and takes the average to obtain the grid frequency difference of the training sample data, and sums up the first power difference of each training sample data and takes the average to obtain the power difference average of the training sample data.

[0091] Furthermore, the AGC frequency modulation device calculates the quotient of the average difference between the grid frequency and the power difference in the training sample data to obtain the parameters of the first model to be optimized. Therefore, the first model parameters to be optimized The calculation formula can be expressed as: ; Where n represents the number of training sample data.

[0092] Step 405: Based on the first model parameters to be optimized, the second grid frequency difference between each test sample data and the rated grid frequency, and the second power difference between the second power result and the first power result corresponding to each test sample data, the model is trained to obtain the target frequency regulation control model.

[0093] Furthermore, the AGC frequency modulation device calculates the second power difference between the second power result and the first power result corresponding to each test sample data. The specific process is as described in the above embodiment. The model is trained based on the first model parameters to be optimized, the second grid frequency difference between each test sample data and the rated grid frequency, and the second power difference between the second power result and the first power result corresponding to each test sample data, to obtain the target frequency regulation control model. The specific analysis is as follows: Optionally, the AGC frequency modulation device adjusts the parameters of the first model to be optimized and the first test sample data. Determine the predicted power difference between the first test sample data and the rated grid frequency. Indicates the first Based on the test sample data and the parameters of the first model to be optimized, as well as the second test sample data and the rated grid frequency, the predicted power difference of the second test sample data is determined. Therefore, the formulas for calculating the predicted power difference of the first test sample data and the predicted power difference of the second test sample data are as follows: ; in, This represents the difference in predicted power between the first test sample data. This represents the difference in predicted power between the second and third test sample data. This represents the first test sample data. This represents the second test sample data.

[0094] Furthermore, the AGC frequency modulation device uses the predicted power difference from the first test sample data. The second power difference is input into the first preset loss function to obtain the loss function value of the first test sample data, and the predicted power difference of the second test sample data. The difference between the second and second power values ​​is input into the first preset loss function to obtain the loss function value of the second test sample data. The first preset loss function is expressed as: ; in, This represents the loss function value for the i-th test sample data. This represents the difference in predicted power for the i-th test sample data. This represents the second power difference of the i-th test sample data.

[0095] Furthermore, the AGC frequency modulation device calculates the difference between the loss function value of the first test sample data and the loss function value of the second test sample data. If the difference is less than or equal to a first preset difference, the AGC frequency modulation device continues to calculate the predicted power difference of the third test sample data using the first model parameters to be optimized. Based on the predicted power difference of the third test sample data, it calculates the loss function value of the third test sample data and calculates the difference between the loss function values ​​of the second and third test sample data. This process is repeated until a first consecutive number of differences are less than or equal to a first preset threshold, thus obtaining the first optimal model parameters and completing the training of the target frequency modulation control model. The first preset difference and the first consecutive number are set according to actual conditions, such as the first consecutive number being 50.

[0096] If the difference is greater than the first preset difference, the AGC frequency modulation device adjusts the parameters of the first model to be optimized, calculates the predicted power difference of the third test sample data based on the adjusted parameters, calculates the loss function value of the third test sample data based on the predicted power difference and the predicted power difference, and calculates the difference between the loss function value of the second test sample data and the loss function value of the third test sample data. This process is repeated to obtain the difference between the loss function values ​​of two adjacent test sample data until a first consecutive number of differences are less than or equal to the first preset threshold, thus obtaining the first optimal model parameters and completing the training of the target frequency modulation control model.

[0097] In one embodiment, the processing formula of the obtained target frequency modulation control model can be expressed as: ; in, Indicates the amount of power adjustment. Represents the first optimal model parameters. This indicates the rated grid frequency under the current operating conditions.

[0098] In one embodiment, the future power grid frequency First optimal model parameters The rated grid frequency of the hydropower plant units under the current operating conditions is: Therefore, the power adjustment of the hydropower plant units at the second moment is: .

[0099] This invention trains a target frequency regulation control model, which can then accurately predict the power adjustment of hydropower plant units at the next moment based on the future grid frequency. This allows for advance prediction of the power adjustment of hydropower plant units at the next moment, enabling advance planning of unit output based on the power adjustment, effectively reducing the lag in frequency regulation response, improving the frequency regulation response of hydropower plant AGC, and ensuring the stable operation of the power system.

[0100] Furthermore, the hydropower plant AGC frequency regulation device based on multimodal control strategy provided by the present invention will be described below. The hydropower plant AGC frequency regulation device based on multimodal control strategy described below can be referred to in correspondence with the hydropower plant AGC frequency regulation method based on multimodal control strategy described above.

[0101] Optional, refer to Figure 2 , Figure 2 This is a schematic diagram of the structure of the hydropower plant AGC frequency regulation device based on a multimodal control strategy provided by the present invention. The hydropower plant AGC frequency regulation device based on a multimodal control strategy includes: The data acquisition module 210 is used to acquire the unit parameters of the hydropower plant unit in real time at the first moment; the unit parameters include the current water head value, the current water flow value, the current unit output and the historical grid frequency; The determination module 220 is used to determine the current operating mode of the hydropower plant unit based on the current head height value and the current water flow value, and to match the target frequency regulation control model based on the current operating mode. The first prediction module 230 is used to input historical power grid frequencies into the power grid frequency prediction model to obtain the future power grid frequency of the hydropower plant units at the second moment, which is output by the power grid frequency prediction model; the power grid frequency prediction model is trained based on the sample power grid frequencies and their corresponding labeled data; The second prediction module 240 is used to input the future grid frequency into the target frequency regulation control model to obtain the power adjustment amount output by the target frequency regulation control model. The target frequency regulation control model is trained based on the first power result corresponding to the rated grid frequency under the current operating condition, as well as the sample grid frequency and its corresponding second power result.

[0102] The adjustment module 250 is used to adjust the current unit output of the hydropower plant based on the power adjustment amount, so as to obtain the target unit output of the hydropower plant at the second moment; the first moment represents the current operating moment of the hydropower plant unit, and the second moment represents the moment after the first moment.

[0103] This invention divides different operating modes based on water head and flow rate. Then, using a multi-modal frequency control model corresponding to each operating mode, it accurately predicts the power adjustment amount for the next moment based on the future grid frequency. Therefore, control parameters can be flexibly adjusted through a multi-modal control strategy, improving the frequency regulation performance of the hydropower plant's AGC (Automatic Generation Control) system. Simultaneously, the grid frequency prediction model accurately predicts the future grid frequency of the hydropower plant units for the next moment, allowing for advance prediction of the grid frequency and planning of the unit output in advance. This effectively reduces the lag in frequency regulation response, improves the frequency regulation response of the hydropower plant's AGC, and ensures the stable operation of the power system.

[0104] Please see Figure 3 , Figure 3 An embodiment diagram of an electronic device provided in accordance with the present invention. For example... Figure 3 As shown, this embodiment of the invention provides an electronic device 300, including a memory 310, a processor 320, and a computer program 311 stored in the memory 310 and executable on the processor 320. When the processor 320 executes the computer program 311, it performs the following steps: Real-time acquisition of hydropower plant unit parameters at the first moment; unit parameters include current water head, current water flow, current unit output, and historical grid frequency; Based on the current head height and current flow rate, determine the current operating mode of the hydropower plant unit, and match the target frequency regulation control model based on the current operating mode. Historical grid frequencies are input into the grid frequency prediction model to obtain the future grid frequency of the hydropower plant units at the second moment, which is output by the grid frequency prediction model. The grid frequency prediction model is trained based on sample grid frequencies and their corresponding labeled data. The future grid frequency is input into the target frequency regulation control model to obtain the power adjustment output of the target frequency regulation control model. The target frequency regulation control model is trained based on the first power result corresponding to the rated grid frequency under the current operating condition, as well as the sample grid frequency and its corresponding second power result. The current unit output of the hydropower plant is adjusted based on the power adjustment amount to obtain the target unit output of the hydropower plant at the second moment; the first moment represents the current operating moment of the hydropower plant unit, and the second moment represents the moment after the first moment.

[0105] Please see Figure 4 , Figure 4 An embodiment diagram of a computer-readable storage medium provided in accordance with an embodiment of the present invention is shown. Figure 4 As shown, this embodiment provides a computer-readable storage medium 400 on which a computer program 311 is stored. When the computer program 311 is executed by a processor, it performs the following steps: Real-time acquisition of hydropower plant unit parameters at the first moment; unit parameters include current water head, current water flow, current unit output, and historical grid frequency; Based on the current water head height and current water flow, determine the current operating mode of the hydropower plant unit, and match the target frequency regulation control model based on the current operating mode; Historical grid frequencies are input into the grid frequency prediction model to obtain the future grid frequency of the hydropower plant units at the second moment, which is output by the grid frequency prediction model. The grid frequency prediction model is trained based on sample grid frequencies and their corresponding labeled data. The future grid frequency is input into the target frequency regulation control model to obtain the power adjustment output of the target frequency regulation control model. The target frequency regulation control model is trained based on the first power result corresponding to the rated grid frequency under the current operating condition, as well as the sample grid frequency and its corresponding second power result. The current unit output of the hydropower plant is adjusted based on the power adjustment amount to obtain the target unit output of the hydropower plant at the second moment; the first moment represents the current operating moment of the hydropower plant unit, and the second moment represents the moment after the first moment.

[0106] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the hydropower plant AGC frequency regulation method based on the multimodal control strategy provided by the above methods. The method includes: Real-time acquisition of hydropower plant unit parameters at the first moment; unit parameters include current water head, current water flow, current unit output, and historical grid frequency; Based on the current water head height and current water flow, determine the current operating mode of the hydropower plant unit, and match the target frequency regulation control model based on the current operating mode; Historical grid frequencies are input into the grid frequency prediction model to obtain the future grid frequency of the hydropower plant units at the second moment, which is output by the grid frequency prediction model. The grid frequency prediction model is trained based on sample grid frequencies and their corresponding labeled data. The future grid frequency is input into the target frequency regulation control model to obtain the power adjustment output of the target frequency regulation control model. The target frequency regulation control model is trained based on the first power result corresponding to the rated grid frequency under the current operating condition, as well as the sample grid frequency and its corresponding second power result. The current unit output of the hydropower plant is adjusted based on the power adjustment amount to obtain the target unit output of the hydropower plant at the second moment; the first moment represents the current operating moment of the hydropower plant unit, and the second moment represents the moment after the first moment.

[0107] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0108] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0109] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A frequency regulation method for AGC in hydropower plants based on a multimodal control strategy, characterized in that, include: Real-time acquisition of hydropower plant unit parameters at the first moment; the unit parameters include the current water head, current water flow, current unit output, and historical grid frequency; Based on the current head height and the current flow rate, the current operating mode of the hydropower plant unit is determined, and a target frequency regulation control model is matched based on the current operating mode. The historical power grid frequency is input into the power grid frequency prediction model to obtain the future power grid frequency of the hydropower plant unit at the second moment, which is output by the power grid frequency prediction model; the power grid frequency prediction model is trained based on the sample power grid frequency and its corresponding labeled data; The future grid frequency is input into the target frequency regulation control model to obtain the power adjustment amount output by the target frequency regulation control model; the target frequency regulation control model is trained based on the first power result corresponding to the rated grid frequency under the current operating condition mode, and the sample grid frequency and its corresponding second power result; The current unit output of the hydropower plant is adjusted based on the power adjustment amount to obtain the target unit output of the hydropower plant at the second time moment; the first time moment represents the current operating time of the hydropower plant unit, and the second time moment represents the next time moment after the first time moment.

2. The hydropower plant AGC frequency regulation method based on multimodal control strategy according to claim 1, characterized in that, The training process of the target frequency modulation control model is as follows: The sample power grid frequency is divided into training sample data and test sample data. The first grid frequency difference for each training sample data is obtained by calculating the difference between each training sample data and the rated grid frequency. The difference between the second power result and the first power result corresponding to each training sample data is calculated to obtain the first power difference of each training sample data. Based on the first grid frequency difference and the first power difference of each training sample data, the first model parameters to be optimized are determined; The target frequency regulation control model is obtained by training the model based on the first model parameters to be optimized, the second grid frequency difference between each test sample data and the rated grid frequency, and the second power difference between the second power result and the first power result corresponding to each test sample data.

3. The hydropower plant AGC frequency regulation method based on multimodal control strategy according to claim 2, characterized in that, The target frequency regulation control model is obtained by training the model based on the first model parameters to be optimized, the second grid frequency difference between each test sample data and the rated grid frequency, and the second power difference between the second power result and the first power result corresponding to each test sample data, including: Based on the parameters of the first model to be optimized and the first test sample data and the rated grid frequency, the predicted power difference of the first test sample data is determined, and based on the parameters of the first model to be optimized and the second test sample data and the rated grid frequency, the predicted power difference of the second test sample data is determined. Based on the first preset loss function and the predicted power difference and the second power difference of the first test sample data, the loss function value of the first test sample data is determined, and based on the second preset loss function and the predicted power difference and the second power difference of the second test sample data, the loss function value of the second test sample data is determined. The parameters of the first model to be optimized are adjusted based on the difference between the loss function values ​​of the first test sample data and the loss function values ​​of the second test sample data. The difference between the loss function values ​​of two adjacent test sample data is obtained based on the adjusted parameters of the first model to be optimized until the first consecutive number of differences are less than or equal to the first preset threshold. The first optimal model parameters are obtained, and the training of the target frequency modulation control model is completed. The first preset loss function is expressed as: ； in, This represents the loss function value for the i-th test sample data. This represents the difference in predicted power for the i-th test sample data. This represents the second power difference of the i-th test sample data.

4. The hydropower plant AGC frequency regulation method based on multimodal control strategy according to claim 1, characterized in that, The training process of the power grid frequency prediction model is as follows: The sample power grid frequency is divided according to a sliding window with a preset step size to obtain multiple training sample sequences and multiple test sample sequences; Obtain the first target grid frequency for each training sample sequence; the first target grid frequency for each training sample sequence is the next grid frequency for each training sample sequence. Based on the first target grid frequency of each training sample sequence and the mean grid frequency of each training sample sequence, the third grid frequency difference of each training sample sequence is determined. The mean value of the grid frequency difference is obtained based on the third grid frequency difference of each training sample sequence; The power grid frequency prediction model is obtained by training the model based on the mean difference of the power grid frequency, each test sample sequence and its second target power grid frequency; the second target power grid frequency of each test sample sequence is the next power grid frequency of each test sample sequence.

5. The hydropower plant AGC frequency regulation method based on multimodal control strategy according to claim 4, characterized in that, The process of training the model based on the mean of the power grid frequency difference, each test sample sequence, and its second target power grid frequency to obtain the power grid frequency prediction model includes: Based on the average grid frequency difference, the parameters of the second model to be optimized, and the average grid frequency of the first test sample sequence, the final grid frequency of the first test sample sequence is determined. Based on the second preset loss function and the final grid frequency and the second target grid frequency of the first test sample sequence, the loss function value of the first test sample sequence is determined; The second model parameters to be optimized are adjusted based on the loss function value of the first test sample sequence. The loss function value of the second test sample sequence is obtained based on the adjusted second model parameters to be optimized. This process continues until the loss function value of the second consecutive number of test sample sequences is less than or equal to the second preset threshold. The second optimal model parameters are then obtained, and the training of the power grid frequency prediction model is completed. The second preset loss function is expressed as: ； in, This represents the loss function value of the i-th test sample sequence. This represents the final grid frequency of the i-th test sample sequence. This represents the second target grid frequency of the i-th test sample sequence. This represents the mean frequency of the i-th test sample sequence.

6. The hydropower plant AGC frequency regulation method based on multimodal control strategy according to claim 1, characterized in that, The process of matching the target frequency modulation control model based on the current operating condition mode includes: Based on the current operating condition mode, combined with the current head height value and the current water flow rate value, determine the operating condition mode characteristic value under the current operating condition mode; Based on each frequency control model in the preset model library, combined with the current head height value and the current water flow value, the model adaptation characteristic value of each frequency control model under the current operating condition is determined; Based on the characteristic values ​​of the operating condition mode and the model adaptation characteristic values ​​of each frequency modulation control model, the matching degree value of each frequency modulation control model to the current operating condition mode is determined. The frequency modulation control model corresponding to the maximum matching degree value is determined as the target frequency modulation control model.

7. The hydropower plant AGC frequency regulation method based on a multimodal control strategy according to any one of claims 1 to 6, characterized in that, The process of determining the current operating mode of the hydropower plant unit based on the current head height and the current flow rate includes: Based on the current head height value and the maximum and minimum head height values ​​of the hydropower plant unit, a first operating condition discrimination value is determined. Based on the current water flow value and the maximum and minimum water flow values ​​of the hydropower plant unit, a second operating condition discrimination value is determined. Based on the first operating condition discrimination value and the second operating condition discrimination value, a comprehensive operating condition discrimination value is determined; Based on the preset fuzzy operating condition discrimination strategy and the comprehensive operating condition discrimination value, the current operating condition mode of the hydropower plant unit is determined.

8. A frequency regulation device for AGC in a hydropower plant based on a multimodal control strategy, characterized in that, The hydropower plant AGC frequency regulation method based on a multimodal control strategy, as described in any one of claims 1 to 7, comprises: The data acquisition module is used to collect the unit parameters of the hydropower plant in real time at the first moment; the unit parameters include the current water head, the current water flow, the current unit output, and the historical grid frequency; The determination module is used to determine the current operating condition mode of the hydropower plant unit based on the current head height value and the current water flow value, and to match the target frequency regulation control model based on the current operating condition mode. The first prediction module is used to input the historical power grid frequency into the power grid frequency prediction model to obtain the future power grid frequency of the hydropower plant unit at the second moment, which is output by the power grid frequency prediction model; the power grid frequency prediction model is trained based on the sample power grid frequency and its corresponding labeled data. The second prediction module is used to input the future grid frequency into the target frequency regulation control model to obtain the power adjustment amount output by the target frequency regulation control model; the target frequency regulation control model is trained based on the first power result corresponding to the rated grid frequency under the current operating condition mode, and the sample grid frequency and its corresponding second power result; The adjustment module is used to adjust the current unit output of the hydropower plant unit based on the power adjustment amount to obtain the target unit output of the hydropower plant unit at the second time moment; the first time moment represents the current operating time of the hydropower plant unit, and the second time moment represents the next time moment after the first time moment.

9. An electronic device, comprising: Memory, used to store computer software programs; A processor for reading and executing the computer software program, characterized in that, when the processor executes the computer software program, it implements the hydropower plant AGC frequency regulation method based on a multimodal control strategy as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium, wherein a computer software program is stored therein, characterized in that, When the computer software program is executed by the processor, it implements the hydropower plant AGC frequency regulation method based on a multimodal control strategy as described in any one of claims 1 to 7.

Images