A Flywheel Energy Storage Adaptive Optimization Power Allocation Method and System Based on Mode Decomposition

Abstract

This invention discloses a flywheel energy storage adaptive optimization power allocation method and system based on mode decomposition. The method acquires grid AGC commands in real time through a sliding time window and decomposes the commands into IMF components at different time scales using the CEEMDAN method. Then, it dynamically determines the reconfiguration order based on the maximum ramp rate of the thermal power unit, allocating high-frequency components to the flywheel energy storage and low-frequency components to the thermal power unit. The power allocation commands are periodically updated through rolling optimization. This invention addresses the problem of slow frequency regulation response and insufficient regulation accuracy of traditional thermal power in grids with a high proportion of renewable energy, due to increased uncontrollability on the power source side. By compensating for the inertial delay of thermal power units with the fast response characteristics of flywheel energy storage, it achieves adaptive optimization allocation of frequency regulation power, thereby significantly improving the joint frequency regulation response speed and quality. Simultaneously, it reduces the frequent operation of thermal power units, lowers equipment wear, and extends their service life.

Description

Technical Field

[0001] This invention relates to the field of automatic generation control and energy storage frequency regulation technology in power systems, and particularly to a flywheel energy storage adaptive optimization power allocation method and system based on mode decomposition. Background Technology

[0002] Currently, traditional thermal power units still bear the main responsibility for frequency regulation in my country. With the rapid development of renewable energy generation, significant challenges have been posed to the stable operation of the power system. my country relies heavily on wind and solar power, and because wind and solar energy are affected by geographical environment, climate, and weather changes, their output is highly random and intermittent. This causes the power supply side to shift from being fully controllable to partially uncontrollable, posing an unprecedented challenge to grid frequency stability. Against this backdrop, the demand for frequency regulation technology in the power grid has become even more urgent.

[0003] In recent years, energy storage has received significant attention as a novel frequency regulation technology for power grid auxiliary frequency regulation. Flywheel energy storage, in particular, has been widely applied in power grid frequency regulation projects due to its advantages such as high energy storage density, millisecond-level response speed, fast charge and discharge rates, long cycle life, high conversion efficiency, and the absence of harmful environmental pollutants during operation. Using flywheel energy storage to assist in the frequency regulation of thermal power plants can effectively compensate for the insufficient regulation capacity of thermal power units caused by the large inertia and slow response of their thermal systems. Therefore, research on flywheel energy storage to assist in the frequency regulation of thermal power plants has significant practical and guiding implications.

[0004] Existing research largely focuses on signal decomposition methods for energy storage capacity configuration, with limited research on real-time capture of Automatic Generation Control (AGC) commands and their decomposition into frequency regulation power commands that thermal power units and energy storage need to undertake. Furthermore, current signal decomposition research still employs filtering and traditional empirical mode decomposition methods. Filtering methods suffer from long transition bands and signal distortion, making them unsuitable for decomposing high-frequency signals into high-frequency signals. Traditional empirical mode decomposition methods, on the other hand, suffer from mode aliasing and endpoint effects, resulting in low decomposition accuracy and consequently, insufficient frequency regulation accuracy. Summary of the Invention

[0005] In view of the shortcomings of the prior art, the purpose of this invention is to provide a flywheel energy storage adaptive optimization power allocation method and system based on mode decomposition, which aims to solve the problems of slow response speed and insufficient regulation accuracy of traditional thermal power frequency regulation methods in grids with a high proportion of renewable energy, due to the increased uncontrollability of the power source side.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: Firstly, a flywheel energy storage adaptive optimization power allocation method based on mode decomposition is provided, the adaptive optimization power allocation method comprising: S1. Set a sliding time window and collect the automatic power generation control commands of the power grid in real time within the sliding time window; S2. Perform mode decomposition on the collected automatic power generation control commands to obtain multiple intrinsic mode function components and residual margins that characterize the fluctuation characteristics at different time scales. S3. Set the reconstruction order, and reconstruct the intrinsic mode function components and residual margins with different frequencies based on the reconstruction order to obtain high-frequency components and low-frequency components respectively, and allocate them to the flywheel energy storage and thermal power units. Obtain the frequency regulation power command required by the flywheel energy storage and thermal power units according to the allocation result, and perform frequency regulation power allocation. S4. After a preset control cycle, perform rolling optimization and return to step S2 to re-decompose and reconstruct the automatic power generation control command signal, and update the frequency regulation power command required by the flywheel energy storage and thermal power unit to achieve adaptive optimized power allocation.

[0007] Furthermore, in S1, the sliding time window needs to have a set length, which is set according to the fluctuation characteristics of the power grid AGC signal.

[0008] Furthermore, the length of the sliding time window is determined according to the following formula: ; in, This is the initial setting for the sliding time window; To adjust the intensity coefficient, its value is greater than 0. It is an adjustable gain parameter used to control the sensitivity of the window length to fluctuation changes. This is a volatility indicator for AGC orders over a recent period, calculated using standard deviation. The historical average volatility is typically calculated over the previous complete time window. The average value represents the normal level of recent fluctuations in the power grid system; The value range is from 10 minutes to 60 minutes; Adjusting the strength coefficient Determines when real-time volatility Deviation from historical average volatility At that time, adjust the window length using force. The value is obtained after testing with actual AGC commands.

[0009] Furthermore, in S2, mode decomposition employs a complete set empirical mode decomposition method based on adaptive noise to decompose the acquired automatic generation control command sequence into... One intrinsic mode function component and one margin: ; in, It is a residual sequence. For the m-th intrinsic mode function component, This represents the total number of intrinsic mode function components.

[0010] Furthermore, complete set empirical mode decomposition methods based on adaptive noise include: S201, To the original signal By adding N Gaussian white noise, a total of N preprocessed sequences are obtained. ,in , is represented as: ; In the formula, These are the weighting coefficients of Gaussian white noise; It is Gaussian white noise during the nth processing; S202, For all preprocessed sequences Perform EMD decomposition to obtain the first IMF component of the entire sequence. , The two conditions of the IMF must be met, and their mean is taken as the first IMF component obtained from the Ceemdan decomposition. At the same time, the first residual sequence is obtained. : ; ; S203, similarly, the residual sequence The first IMF component after EMD decomposition with added Gaussian white noise is used to construct N new sequences. After performing EMD decomposition on these N sequences, the mean value is used to obtain the second IMF component. And get the difference And so on, to obtain the m-th IMF component. and the residual of stage m : ; ; in, This represents the noise coefficient that CEEMDAN adds to the input sequence in the (m-1)th stage; S204. Repeat the above steps until the decomposition stops, and finally the residual sequence is obtained. for: ; That is, signal sequence The expression after CEEMDAN decomposition is shown below: .

[0011] Furthermore, in S3, the reconfiguration order is determined based on the maximum ramp rate of the thermal power unit, including: S301. Obtain all possible reconfiguration results, and based on the reconfiguration results, obtain a series of frequency regulation powers required by the thermal power units. ,in ; S302. Obtain the root mean square (RMS) value between the frequency regulation power required for each thermal power plant and the maximum ramp rate, and obtain the index value that minimizes the RMS value to obtain the reconstruction order k: ; in, This represents the maximum ramp rate of the thermal power unit. The time required for the thermal power unit to climb the slope; S303. Based on the reconstruction order k, the eigenmode function components and margins of different frequencies are reconstructed into high-frequency and low-frequency components to obtain the secondary frequency regulation power command required by the thermal power unit and flywheel energy storage at the current moment. Among them, the high-frequency component is allocated to the flywheel energy storage, and the low-frequency component is allocated to the thermal power unit, as expressed as: ; .

[0012] Furthermore, in S3, the secondary frequency regulation power command that the thermal power unit and flywheel energy storage need to undertake may exceed the limit. The frequency regulation power command needs to be corrected, and the correction method is as follows: When the frequency regulation power command undertaken by the thermal power unit exceeds its maximum adjustable capacity, while the command undertaken by the flywheel energy storage does not exceed its current real-time adjustable capacity, the flywheel energy storage can take on the excess part of the thermal power unit's command within the range of its real-time adjustable capacity to make up for the frequency regulation power gap. When the frequency regulation power command undertaken by the thermal power unit is within the limit, but the command undertaken by the energy storage power station exceeds its current real-time adjustable capacity, the thermal power unit shall undertake the excess part of the command from the energy storage power station within its maximum adjustable capacity. When the frequency regulation power commands undertaken by both thermal power units and energy storage power stations exceed their respective maximum adjustable capabilities, both will output according to their respective upper limits.

[0013] Furthermore, the real-time adjustable capability of flywheel energy storage is the maximum charging and generating power of the energy storage. The frequency regulation power command of flywheel energy storage should be subject to the following constraints: ; ; ; in, and These are the maximum charging and discharging power of the flywheel energy storage, respectively. and These represent the upper and lower limits of the flywheel energy storage SOC, set according to engineering experience. ; Rated capacity for flywheel energy storage; Rated power for flywheel energy storage; The time step is 1 second.

[0014] Furthermore, in S4, the preset control cycle is the length of the time window; the rolling optimization is to slide the sliding time window forward, incorporate the latest automatic power generation control command, and discard the earliest data.

[0015] Secondly, a flywheel energy storage adaptive optimization power allocation system based on mode decomposition includes a data acquisition module, an adaptive window management module, a power command allocation module, and a rolling optimization scheduling module connected in a sequential loop. The data acquisition module acquires the grid AGC command sequence in real time; the adaptive window management module maintains and dynamically adjusts the length of the sliding time window; the signal decomposition and reconstruction module performs online mode decomposition on the AGC command sequence within the window and reconstructs it into high-frequency and low-frequency components according to the reconstruction order k; the power command allocation module allocates the high-frequency components to the flywheel energy storage system and the low-frequency components to the thermal power unit, and corrects the frequency regulation power command; the rolling optimization scheduling module triggers window sliding and starts a new round of optimization at the end of the control cycle.

[0016] The technical solution adopted in this invention has the following beneficial effects: In this invention, a sliding time window is first introduced to collect grid AGC commands in real time. Then, the CEEMDAN method decomposes the commands into intrinsic mode function components at different time scales. Subsequently, the reconfiguration order is determined based on the maximum ramp rate of the thermal power unit. High-frequency components are allocated to flywheel energy storage, and low-frequency components are allocated to the thermal power unit. Then, the power allocation commands are periodically updated through rolling optimization. This solves the problem of slow frequency regulation response and insufficient regulation accuracy of traditional thermal power in grids with a high proportion of renewable energy, due to the increased uncontrollability of the power source side. By compensating for the inertial delay of the thermal power unit through the fast response characteristics of flywheel energy storage, adaptive optimization allocation of frequency regulation power is achieved, thereby significantly improving the joint frequency regulation response speed and frequency regulation quality. At the same time, it reduces the frequent operation of thermal power units, reduces equipment wear, and extends their service life. Attached Figure Description

[0017] Figure 1 This is a flowchart of an adaptive optimization power allocation method for flywheel energy storage based on complete set empirical mode decomposition, as described in this invention. Figure 2 This is a flowchart illustrating the determination of the time window length according to an embodiment of the present invention; Figure 3 This is a graph showing the AGC instruction curves of an embodiment of the present invention; Figure 4 This is a flowchart of the CEEMDAN process according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the IMF after decomposition according to an embodiment of the present invention; Figure 6 This is a flowchart illustrating the determination of the reconstruction coefficient k according to an embodiment of the present invention; Figure 7 The results are MATLAB simulations of the method of this invention. Figure 8 This is a block diagram of a flywheel energy storage adaptive optimization power allocation system based on complete set empirical mode decomposition, as described in this invention. Detailed Implementation

[0018] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] Based on the problems mentioned in the background, this invention proposes a method called Complementary Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN), an advanced signal decomposition technique for processing nonlinear and non-stationary signals. The CEEMDAN algorithm effectively solves the mode mixing and endpoint effects problems in traditional Empirical Mode Decomposition (EMD) by introducing adaptive noise and multiple iterations, thus improving the accuracy and stability of the decomposition.

[0020] Figure 1 This is a flowchart of the adaptive optimization power allocation method for flywheel energy storage based on mode decomposition described in this invention. Figure 1 As shown in the figure, this invention discloses an adaptive optimization power allocation method for flywheel energy storage based on mode decomposition, which includes the following steps: S101: Set a sliding time window and collect AGC commands from the power grid within the time window in real time; Figure 2 This is a flowchart illustrating the determination of the time window length according to an embodiment of the present invention. Figure 3This is a graph illustrating the AGC instruction curves of an embodiment of the present invention. Figure 3 As shown, this retrieves AGC instruction data for a specific time period of 120 minutes. Figure 2 As shown, the length of the sliding time window can be set according to the fluctuation characteristics of the power grid AGC signal.

[0021] A further technical solution involves adjusting the length of the sliding time window. Determined by the following formula: (1) in, The initial setting for the sliding time window can be 20 minutes in a typical implementation. To adjust the intensity coefficient, its value is greater than 0. It is an adjustable gain parameter used to control the sensitivity of the window length to fluctuation changes. This is a volatility indicator for AGC orders over a recent period, calculated using standard deviation. The historical average volatility is typically calculated over the previous complete time window. The average value represents the normal level of recent fluctuations in the system; The value range is from 10 minutes to 60 minutes.

[0022] A further technical solution lies in, Determines when real-time volatility Deviation from historical average volatility At that time, adjust the window length using the following method: Too large: The system will overreact. Window length Drastic changes lead to decreased stability in signal decomposition, and frequent and significant adjustments to the power allocation strategy may trigger system oscillations.

[0023] Too small: The system becomes sluggish. The window length adjusts slowly, and it cannot quickly shorten the window to track high-frequency fluctuations when the power grid conditions change drastically, thus losing its adaptive advantage.

[0024] therefore The value needs to be tested using actual AGC commands: 1. Based on prior knowledge, for Set a reasonable value, let ; 2. Run the algorithm using a representative set of AGC instruction data; 3. Observe the window length during periods of significant fluctuation. Whether the length is shortened, and whether the flywheel effectively handles high-frequency fluctuations. If the system does not respond quickly enough to fluctuations, gradually increase the length. Retest; if the system remains unstable, gradually reduce the size. .

[0025] A further technical solution is that the method for determining the time length indicates that when fluctuations intensify, the window is automatically shortened, which allows the system to forget past stable states more quickly and focus more on recent violent fluctuations; when fluctuations are mild, if the system is in a stable state, the window is automatically extended to provide richer data for modal decomposition.

[0026] S102: Decompose the AGC command to obtain multiple IMF components that characterize fluctuations at different time scales; Figure 4 This is a flowchart of the CEEMDAN process according to an embodiment of the present invention. Figure 4 As shown, the decomposition method employs the Complementary Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) method to decompose the AGC instruction sequence into... One IMF component and one margin: (2) The principle of CEEMDAN decomposition is as follows: set up This represents the j-th IMF component after EMD decomposition of the sequence. This represents the noise coefficient added to the input sequence by CEEMDAN in stage m. The first generation generated for CEEMDAN The IMF component, the intrinsic modal component, has two constraints: Within the entire data segment, the number of extreme points and the number of zero-crossing points must be equal or differ by no more than one.

[0027] At any given time, the average of the upper envelope formed by the local maxima and the lower envelope formed by the local minima is zero, meaning that the upper and lower envelopes are locally symmetrical with respect to the time axis.

[0028] S201: To the original signal By adding N orders of Gaussian white noise, a total of N preprocessed sequences are obtained. ,in : (3) In the formula, These are the weighting coefficients of Gaussian white noise; It is the Gaussian white noise during the nth processing.

[0029] S202: For all preprocessed sequences Perform EMD decomposition to obtain the first IMF component of the entire sequence. , The two conditions of the IMF must be met, and their mean is taken as the first IMF component obtained from the Ceemdan decomposition. At the same time, the first residual sequence is obtained. : (4) (5) S203: Similarly, the residual sequence The first IMF component after EMD decomposition with added Gaussian white noise is used to construct N new sequences. After performing EMD decomposition on these N sequences, the mean value is used to obtain the second IMF component. And get the difference And so on, to obtain the m-th IMF component. and the residual of stage m : (6) (7) S204: Repeat the above steps until the decomposition stops, and finally the residual sequence is obtained. for: (8) That is, signal sequence The expression after CEEMDAN decomposition is shown below: (9) Figure 5 This is a schematic diagram of the disassembled IMF according to an embodiment of the present invention. Figure 5 As shown, the AGC signal was decomposed into 10 IMF components and one residual signal using the CEEMDAN decomposition method. The decomposition results progressed from high frequency to low frequency. The lower-ranked mode components had high frequency and high instantaneous power, while the higher-ranked mode components had low frequency and long power period.

[0030] S103: Set the reconstruction order The IMF components with different frequencies and the residual margin are reconstructed to obtain high-frequency components and low-frequency components respectively. The obtained high-frequency components and low-frequency components are allocated to the flywheel energy storage and the thermal power unit respectively. At this time, the frequency regulation power command to be undertaken by the flywheel energy storage and the thermal power unit is obtained. Figure 6This is a flowchart illustrating the determination of the reconstruction coefficient k according to an embodiment of the present invention. Figure 6 As shown, the reconstruction order k is determined based on the maximum ramp rate of the thermal power unit, and the specific steps are as follows: S301: Calculate all possible reconfiguration results to obtain a series of frequency regulation power requirements for thermal power plants. ,in ; S302: Calculate the root mean square (RMS) value between the frequency regulation power required for each thermal power plant and the maximum ramp rate, and take the index value that minimizes the RMS value to obtain the reconstruction order k. (10) in This represents the maximum ramp rate of the thermal power unit. This refers to the time required for the thermal power unit to climb the slope.

[0031] S303: Based on the reconstruction order k, the IMF components of different frequencies and the margin are reconstructed into high-frequency components and low-frequency components to obtain the secondary frequency regulation power command that the thermal power unit and flywheel energy storage need to undertake at the current moment. Among them, the high-frequency component is allocated to the flywheel energy storage, and the low-frequency component is allocated to the thermal power unit. (11) (12) A further technical solution is that the secondary frequency regulation power command that the thermal power unit and flywheel energy storage need to bear may exceed the limit in practice. The frequency regulation power command is corrected according to the following basic rules: When the frequency regulation power command undertaken by the thermal power unit exceeds its maximum adjustable capacity, while the command undertaken by the flywheel energy storage does not exceed its current real-time adjustable capacity, the flywheel energy storage can take on the excess part of the thermal power unit's command within the range of its real-time adjustable capacity to make up for the frequency regulation power gap. When the frequency regulation power command undertaken by the thermal power unit is within the limit, but the command undertaken by the energy storage power station exceeds its current real-time adjustable capacity, the thermal power unit shall undertake the excess part of the command from the energy storage power station within its maximum adjustable capacity. When the frequency regulation power commands undertaken by both thermal power units and energy storage power stations exceed their respective maximum adjustable capabilities, both will output according to their respective upper limits.

[0032] A further technical solution is that the real-time adjustable capability of the flywheel energy storage is the maximum charging and generating power of the energy storage, and the frequency regulation power command of the flywheel energy storage should be within the following constraints: (13) (14) (15) in and These are the maximum charging and discharging power of the flywheel energy storage, respectively. and These represent the upper and lower limits of the flywheel energy storage SOC, set according to engineering experience. ; Rated capacity for flywheel energy storage; Rated power for flywheel energy storage.

[0033] S104: After a preset control cycle, perform rolling optimization and return to step S102 to re-decompose and reconstruct the AGC signal and update the frequency regulation power commands that the flywheel energy storage and thermal power units need to undertake.

[0034] The preset control period is the length of the time window; the scrolling optimization is to slide the sliding time window forward, incorporate the latest AGC instructions, and remove the earliest data.

[0035] Finally, the proposed method was simulated and tested in the MATLAB / SIMULINK simulation environment.

[0036] Figure 7 The MATLAB simulation results of the algorithm of this invention show that the strategy proposed in this invention performs better in following AGC commands. Its dynamic response speed is significantly faster than the traditional strategy, with smaller overshoot and higher steady-state accuracy. This proves that through CEEMDAN decomposition, the system can accurately extract the rapidly changing high-frequency components in the AGC commands, which are then handled by the flywheel energy storage in milliseconds, while the slowly changing low-frequency and residual components are smoothly tracked by the thermal power unit, thus achieving complementary advantages overall. This result also indirectly reflects the effectiveness of the proposed adaptive sliding time window mechanism and the model based on the maximum ramp rate of the thermal power unit to dynamically determine the reconstruction order k. They can optimize the decomposition and allocation strategy in real time according to the actual fluctuation characteristics of the power grid, rather than using fixed parameters.

[0037] In summary, the flywheel energy storage adaptive optimization power allocation method based on complete set empirical mode decomposition according to the embodiments of the present invention has the following beneficial effects: By introducing a sliding time window to collect grid AGC commands in real time, and then using the CEEMDAN method to decompose the commands into intrinsic mode function components at different time scales, the reconfiguration order is determined based on the maximum ramp rate of the thermal power unit. High-frequency components are allocated to flywheel energy storage, and low-frequency components are allocated to the thermal power unit. Power allocation commands are then periodically updated through rolling optimization. This solves the problem of slow frequency regulation response and insufficient regulation accuracy of traditional thermal power in grids with a high proportion of renewable energy, due to the increased uncontrollability of the power source side. By using the fast response characteristics of flywheel energy storage to compensate for the inertial delay of thermal power units, adaptive optimization allocation of frequency regulation power is achieved, which significantly improves the joint frequency regulation response speed and frequency regulation quality. At the same time, it reduces the frequent operation of thermal power units, reduces equipment wear, and extends their service life.

[0038] Based on the same inventive concept, this invention also provides a flywheel energy storage adaptive optimization power allocation system based on complete set empirical mode decomposition. Since the principle of this system in solving the problem is similar to that of the flywheel energy storage adaptive optimization power allocation method based on complete set empirical mode decomposition, the implementation of this system can refer to the implementation of the method, and the repeated parts will not be described again.

[0039] Figure 8 The block diagram of a flywheel energy storage adaptive optimization power allocation system based on complete set empirical mode decomposition described in this invention includes: The data acquisition module is used to acquire the power grid AGC command sequence in real time; An adaptive window management module is used to maintain and dynamically adjust the length of the sliding time window; The signal decomposition and reconstruction module is used to perform online mode decomposition on the AGC instruction sequence within the window, and reconstruct it into high-frequency and low-frequency components according to the reconstruction order k. The power command allocation module is used to allocate high-frequency components to the flywheel energy storage system, low-frequency components to the thermal power unit, and to correct the frequency regulation power command. The rolling optimization scheduling module is used to trigger window sliding and start a new round of optimization process at the end of the control cycle.

[0040] In summary, the flywheel energy storage adaptive optimization power allocation system based on complete ensemble empirical mode decomposition of this invention introduces a sliding time window to collect grid AGC commands in real time. Then, the CEEMDAN method decomposes the commands into intrinsic mode function components at different time scales. Subsequently, the reconfiguration order is determined according to the maximum ramp rate of the thermal power unit. High-frequency components are allocated to flywheel energy storage, and low-frequency components are allocated to the thermal power unit. The power allocation commands are then periodically updated through rolling optimization. This solves the problem of slow frequency regulation response and insufficient regulation accuracy of traditional thermal power in grids with a high proportion of renewable energy, due to the increased uncontrollability of the power source side. By compensating for the inertial delay of the thermal power unit through the fast response characteristics of flywheel energy storage, adaptive optimization allocation of frequency regulation power is achieved, thereby significantly improving the joint frequency regulation response speed and frequency regulation quality. At the same time, it reduces the frequent operation of thermal power units, reduces equipment wear, and extends their service life.

Claims

1. A flywheel energy storage adaptive optimization power allocation method based on mode decomposition, characterized in that, Adaptive optimization power allocation methods include: S1. Set a sliding time window and collect the automatic power generation control commands of the power grid in real time within the sliding time window; S2. Perform mode decomposition on the collected automatic power generation control commands to obtain multiple intrinsic mode function components and residual margins that characterize the fluctuation characteristics at different time scales. S3. Set the reconstruction order, and reconstruct the intrinsic mode function components and residual margins with different frequencies based on the reconstruction order to obtain high-frequency components and low-frequency components respectively, and allocate them to the flywheel energy storage and thermal power units. Obtain the frequency regulation power command required by the flywheel energy storage and thermal power units according to the allocation result, and perform frequency regulation power allocation. S4. After a preset control cycle, perform rolling optimization and return to step S2 to re-decompose and reconstruct the automatic power generation control command signal, and update the frequency regulation power command required by the flywheel energy storage and thermal power unit to achieve adaptive optimized power allocation.

2. The adaptive optimization power allocation method for flywheel energy storage based on mode decomposition according to claim 1, characterized in that, In S1, the sliding time window needs to be set with a length that is determined based on the fluctuation characteristics of the power grid AGC signal.

3. The adaptive optimization power allocation method for flywheel energy storage based on mode decomposition according to claim 2, characterized in that, The length of the sliding time window is determined by the following formula: ； in, This is the initial setting for the sliding time window; To adjust the intensity coefficient, its value is greater than 0. It is an adjustable gain parameter used to control the sensitivity of the window length to fluctuation changes. This is a volatility indicator for AGC orders over a recent period, calculated using standard deviation. The historical average volatility is typically calculated over the previous complete time window. The average value represents the normal level of recent fluctuations in the power grid system; The value range is from 10 minutes to 60 minutes; Adjusting the strength coefficient Determines when real-time volatility Deviation from historical average volatility At that time, adjust the window length using force. The value is obtained after testing with actual AGC commands.

4. The adaptive optimization power allocation method for flywheel energy storage based on mode decomposition according to claim 1, characterized in that, In S2, mode decomposition employs a complete set empirical mode decomposition method based on adaptive noise to decompose the acquired automatic generation control command sequence into... One intrinsic mode function component and one margin: ； in, It is a residual sequence. For the m-th intrinsic mode function component, This represents the total number of intrinsic mode function components.

5. The adaptive optimization power allocation method for flywheel energy storage based on mode decomposition according to claim 4, characterized in that, Complete set empirical mode decomposition methods based on adaptive noise include: S201, To the original signal By adding N Gaussian white noise, a total of N preprocessed sequences are obtained. ,in , is represented as: ； In the formula, These are the weighting coefficients of Gaussian white noise; It is Gaussian white noise during the nth processing step; S202, For all preprocessed sequences Perform EMD decomposition to obtain the first IMF component of the entire sequence. , The two conditions of the IMF must be met, and their mean is taken as the first IMF component obtained from the Ceemdan decomposition. At the same time, the first residual sequence is obtained. : ； ； S203, similarly, the residual sequence The first IMF component after EMD decomposition with added Gaussian white noise is used to construct N new sequences. After performing EMD decomposition on these N sequences, the mean value is used to obtain the second IMF component. And get the difference And so on, to obtain the m-th IMF component. and the residual of stage m : ； ； in, This represents the noise coefficient that CEEMDAN adds to the input sequence in the (m-1)th stage; S204. Repeat the above steps until the decomposition stops, and finally the residual sequence is obtained. for: ； That is, signal sequence The expression after CEEMDAN decomposition is shown below: 。 6. The adaptive optimization power allocation method for flywheel energy storage based on mode decomposition according to claim 1, characterized in that, In S3, the reconfiguration order is determined based on the maximum ramp rate of the thermal power unit, including: S301. Obtain all possible reconfiguration results, and based on the reconfiguration results, obtain a series of frequency regulation powers required by the thermal power units. ,in ; S302. Obtain the root mean square (RMS) value between the frequency regulation power required for each thermal power plant and the maximum ramp rate, and obtain the index value that minimizes the RMS value to obtain the reconstruction order k: ； in, This represents the maximum ramp rate of the thermal power unit. The time required for the thermal power unit to climb the slope; S303. Based on the reconstruction order k, the eigenmode function components and margins of different frequencies are reconstructed into high-frequency and low-frequency components to obtain the secondary frequency regulation power command required by the thermal power unit and flywheel energy storage at the current moment. Among them, the high-frequency component is allocated to the flywheel energy storage, and the low-frequency component is allocated to the thermal power unit, as expressed as: ； 。 7. The adaptive optimization power allocation method for flywheel energy storage based on mode decomposition according to claim 6, characterized in that, In S3, the secondary frequency regulation power command that the thermal power unit and flywheel energy storage need to undertake may exceed the limit. The frequency regulation power command needs to be corrected, and the correction method is as follows: When the frequency regulation power command undertaken by the thermal power unit exceeds its maximum adjustable capacity, while the command undertaken by the flywheel energy storage does not exceed its current real-time adjustable capacity, the flywheel energy storage can take on the excess part of the thermal power unit's command within the range of its real-time adjustable capacity to make up for the frequency regulation power gap. When the frequency regulation power command undertaken by the thermal power unit is within the limit, but the command undertaken by the energy storage power station exceeds its current real-time adjustable capacity, the thermal power unit shall undertake the excess part of the command from the energy storage power station within its maximum adjustable capacity. When the frequency regulation power commands undertaken by the thermal power unit and the energy storage power station both exceed their respective maximum adjustable capabilities, both will output according to their respective upper limits.

8. The adaptive optimization power allocation method for flywheel energy storage based on mode decomposition according to claim 7, characterized in that, The real-time adjustable capability of flywheel energy storage is the maximum charging and generating power of the energy storage. The frequency regulation power command of flywheel energy storage should be subject to the following constraints: ； ； ； in, and These are the maximum charging and discharging power of the flywheel energy storage, respectively. and These represent the upper and lower limits of the flywheel energy storage SOC, set according to engineering experience. ; Rated capacity for flywheel energy storage; Rated power for flywheel energy storage; The time step is 1 second.

9. The adaptive optimization power allocation method for flywheel energy storage based on mode decomposition according to claim 1, characterized in that, In S4, the preset control cycle is the length of the time window; the rolling optimization is to slide the sliding time window forward, incorporate the latest automatic power generation control command, and discard the earliest data.

10. A flywheel energy storage adaptive optimization power allocation system employing the mode decomposition-based flywheel energy storage adaptive optimization power allocation method as described in claim 1, characterized in that, The system comprises a data acquisition module, an adaptive window management module, a power command allocation module, and a rolling optimization scheduling module, which are connected in a sequential loop. The data acquisition module acquires the grid AGC command sequence in real time. The adaptive window management module maintains and dynamically adjusts the length of the sliding time window. The signal decomposition and reconstruction module performs online mode decomposition on the AGC command sequence within the window and reconstructs it into high-frequency and low-frequency components based on the reconstruction order k. The power command allocation module allocates the high-frequency components to the flywheel energy storage system and the low-frequency components to the thermal power unit, and corrects the frequency regulation power command. The rolling optimization scheduling module triggers window sliding and starts a new round of optimization at the end of the control cycle.

Images