Energy storage frequency regulation strategy optimization method and system based on artificial bee colony algorithm

Abstract

This application relates to the field of artificial bee colony algorithm technology, providing a method and system for optimizing energy storage frequency regulation strategies based on the artificial bee colony algorithm, addressing the problems of slow suppression speed and low stability of power grid frequency fluctuations. The method includes: acquiring power grid frequency deviation data under load abrupt changes, extracting frequency change characteristics through sliding time window processing, and performing extreme value normalization to form a standard fluctuation sequence characterizing the dynamic characteristics of the frequency. After converting this sequence into an initial power compensation demand signal, phase compensation preprocessing is performed to obtain an optimized power compensation signal. A parameter optimization model is established with virtual inertial control parameters as optimization variables, and the optimal parameter combination is obtained by iteratively solving using the artificial bee colony algorithm, thereby constructing a frequency-power mapping function. Based on the mapping function, an adaptive virtual inertial control strategy is formed, and after processing the power compensation signal, an energy storage output command is generated. This application improves the suppression speed and stability of power grid frequency fluctuations.

Description

Technical Field

[0001] This application relates to the field of artificial bee colony algorithm technology, and in particular to an optimization method and system for energy storage frequency regulation strategy based on artificial bee colony algorithm. Background Technology

[0002] In power system operation, sudden load changes can cause rapid fluctuations in grid frequency, requiring energy storage systems to provide instantaneous power support to maintain frequency stability. This necessitates that frequency regulation control strategies not only respond quickly to frequency changes but also adaptively adjust output based on the dynamic characteristics of the grid, ensuring the safe operation of energy storage devices while suppressing frequency overshoot.

[0003] Existing solutions employ a virtual inertial control method based on fixed parameters. By measuring the grid frequency deviation, power compensation commands are calculated through preset differential and inertial links, and then the final frequency regulation command is generated by combining the operating state constraints of the energy storage system.

[0004] This approach has limitations in adaptability of control parameters when dealing with frequency fluctuations of varying amplitudes and rates of change, potentially leading to a mismatch between the power compensation response speed and the actual needs of the power grid. Furthermore, fixed control parameters may exhibit fluctuations in their effectiveness in suppressing frequency fluctuations under complex and ever-changing power grid operating conditions, impacting the stable operation of the system. Summary of the Invention

[0005] This application provides a method and system for optimizing energy storage frequency regulation strategies based on artificial bee colony algorithm, in order to solve the problems of slow suppression speed and low stability of power grid frequency fluctuations in the prior art.

[0006] To address the aforementioned technical problems, in a first aspect, this application provides a method for optimizing energy storage frequency regulation strategies based on the artificial bee colony algorithm, comprising:

[0007] Acquire frequency deviation data generated by the energy storage frequency regulation strategy under sudden load changes in the power grid;

[0008] The frequency deviation data is processed by a sliding time window to extract frequency change features, and the frequency change features corresponding to different sliding windows are subjected to extreme value normalization processing to form a standard fluctuation sequence characterizing the dynamic characteristics of the power grid frequency.

[0009] The standard fluctuation sequence is converted into an initial power compensation demand signal, and the initial power compensation demand signal is preprocessed with phase compensation to obtain a preprocessed power compensation demand signal.

[0010] A parameter optimization model is established with the differential gain coefficient and response delay time of virtual inertial control as optimization variables. The artificial bee colony algorithm is used to solve the parameter optimization model in multiple rounds to obtain the optimal parameter combination. Based on the optimal parameter combination, a mapping function between frequency and power is constructed.

[0011] Based on the mapping function, an adaptive virtual inertial control strategy is formed, and the preprocessed power compensation demand signal is processed based on the adaptive virtual inertial control strategy to generate energy storage output commands. The energy storage frequency regulation strategy of the energy storage system is optimized according to the energy storage output commands to achieve optimized suppression of grid frequency fluctuations.

[0012] Optionally, the step of establishing a parameter optimization model with the differential gain coefficient and response delay time of the virtual inertial control as optimization variables, and using the artificial bee colony algorithm to iteratively solve the parameter optimization model in multiple rounds to obtain the optimal parameter combination, and constructing a mapping function between frequency and power based on the optimal parameter combination, includes:

[0013] Establish a parameter optimization model, setting the differential gain coefficient of virtual inertial control as the first optimization variable and the response delay time of virtual inertial control as the second optimization variable;

[0014] Based on the first and second optimization variables, an objective function is defined that aims to minimize the integral of frequency deviation and the rate of power change.

[0015] Based on the characteristics of the objective function, the search parameters of the artificial bee colony algorithm are set, including the number of hired bees, the number of observation bees, and the maximum number of iterations.

[0016] Based on the search parameters, the artificial bee colony algorithm is used to perform multiple rounds of iterative solutions to determine the optimal parameter combination.

[0017] Based on the differential gain coefficient and response delay time in the optimal parameter combination, a mapping function containing a frequency differential term and a delay compensation term is constructed.

[0018] Optionally, the step of using the artificial bee colony algorithm to perform multiple rounds of iterative solving based on the search parameters to determine the optimal parameter combination includes:

[0019] Based on the search parameters, an initial population containing multiple parameter combinations is generated, and each parameter combination includes the values ​​of a first optimization variable and a second optimization variable.

[0020] In the hired bee stage of the artificial bee colony algorithm, a neighborhood search is performed on each parameter combination in the initial population to obtain the updated parameter combination, and the objective function value corresponding to each updated parameter combination is calculated based on the objective function.

[0021] By observing the bee phase of the artificial bee colony algorithm, high-quality parameter combinations with objective function values ​​greater than a preset threshold are selected for deep search to obtain optimized parameter combinations.

[0022] By using the scout bee phase of the artificial bee colony algorithm, the parameters trapped in local optima in the optimized parameter combination are replaced to obtain a new parameter combination, and the new parameter combination is used as the initial population.

[0023] The iterative search process of the hired bee phase, the observation bee phase, and the scout bee phase is repeated until the preset termination condition is met, and the parameter combination with the optimal objective function value is output as the optimal parameter combination.

[0024] Optionally, the step of constructing a mapping function containing a frequency derivative term and a delay compensation term based on the differential gain coefficient and response delay time in the optimal parameter combination includes:

[0025] The frequency deviation data is differentiated to obtain the frequency differential signal;

[0026] The result of multiplying the frequency differential signal by the differential gain coefficient is taken as the frequency differential term;

[0027] The preprocessed power compensation demand signal is subjected to time delay processing to obtain a delayed power signal;

[0028] The difference between the delayed power signal and the preprocessed power compensation demand signal is calculated and used as the delay compensation term;

[0029] The frequency derivative term and the delay compensation term are combined in a weighted manner to construct a mapping function.

[0030] Optionally, the step of converting the standard fluctuation sequence into an initial power compensation demand signal and performing phase compensation preprocessing on the initial power compensation demand signal to obtain a preprocessed power compensation demand signal includes:

[0031] The standard fluctuation sequence is decomposed into multiple frequency sub-band signals, and the power conversion coefficient corresponding to each frequency sub-band signal is determined based on the grid impedance parameters and the energy storage system response characteristics.

[0032] Each frequency sub-band signal is multiplied by its corresponding power conversion coefficient to obtain the power compensation signal for each frequency sub-band signal. The power compensation signals of each frequency sub-band signal are then combined to generate the initial power compensation requirement signal.

[0033] An adaptive filtering algorithm is used to calculate the phase offset of the initial power compensation demand signal relative to the grid voltage signal, and based on the phase offset and a pre-established phase compensation model, the real-time phase compensation value is calculated.

[0034] The real-time phase compensation value is applied to the initial power compensation demand signal to obtain the phase-compensated power compensation demand signal.

[0035] The phase-compensated power compensation demand signal is subjected to amplitude limiting processing to obtain a preprocessed power compensation demand signal.

[0036] Optionally, the step of performing sliding time window processing on the frequency deviation data to extract frequency change features, and performing extreme value normalization processing on the frequency change features corresponding to different sliding windows to form a standard fluctuation sequence characterizing the dynamic characteristics of the power grid frequency, includes:

[0037] Multiple sliding time windows of different durations are set, the sliding time windows including a first window for capturing fluctuation characteristics and a second window for identifying changing trends;

[0038] The first window and the second window are moved sequentially according to a preset time interval, and frequency data segments corresponding to the time period are extracted from the frequency deviation data respectively.

[0039] Differential calculations are performed on the frequency data segments corresponding to each time period of the first window to generate velocity characteristic quantities that reflect instantaneous changes;

[0040] Linear fitting is performed on the frequency data segments corresponding to each second window time period to generate slope feature quantities that reflect the continuous changing trend.

[0041] The velocity feature and the slope feature are combined as corresponding frequency change features to form a frequency change feature set, and the maximum and minimum values ​​of each feature in the frequency change feature set are identified.

[0042] Based on the maximum and minimum values, extreme value normalization is performed on each feature in the frequency change feature set, and the frequency change feature set after extreme value normalization is rearranged in chronological order to form a standard fluctuation sequence.

[0043] Optionally, the step of forming an adaptive virtual inertial control strategy based on the mapping function, and processing the preprocessed power compensation demand signal based on the adaptive virtual inertial control strategy to generate energy storage output commands, includes:

[0044] Based on the frequency derivative term and delay compensation term in the mapping function, a feedforward control channel and a feedback correction channel are constructed respectively.

[0045] The feedforward control channel and the feedback correction channel are connected in parallel to form an adaptive virtual inertial control strategy;

[0046] Based on the adaptive virtual inertial control strategy, the preprocessed power compensation demand signal is subjected to differential gain processing to generate a feedforward power component, and the frequency deviation data is subjected to delay compensation processing to generate a feedback correction component.

[0047] The feedforward power component and the feedback correction component are weighted and fused according to their corresponding dynamic weighting coefficients to generate an untuned control signal.

[0048] The untuned control signal is subjected to power change rate limiting processing to obtain a rate-limited signal, and the rate-limited signal is subjected to power upper and lower limit truncation processing to obtain the energy storage output command.

[0049] Secondly, this application provides an energy storage frequency regulation strategy optimization system based on the artificial bee colony algorithm, comprising:

[0050] The acquisition module is used to acquire frequency deviation data generated by the power grid based on the energy storage frequency regulation strategy under the condition of sudden load change;

[0051] The extraction module is used to perform sliding time window processing on the frequency deviation data to extract frequency change features, and to perform extreme value normalization processing on the frequency change features corresponding to different sliding windows to form a standard fluctuation sequence characterizing the dynamic characteristics of the power grid frequency.

[0052] The conversion module is used to convert the standard fluctuation sequence into an initial power compensation demand signal, and to perform phase compensation preprocessing on the initial power compensation demand signal to obtain a preprocessed power compensation demand signal.

[0053] A module is established to build a parameter optimization model with the differential gain coefficient and response delay time of virtual inertial control as optimization variables. The artificial bee colony algorithm is used to solve the parameter optimization model in multiple rounds to obtain the optimal parameter combination. Based on the optimal parameter combination, a mapping function between frequency and power is constructed.

[0054] The generation module is used to form an adaptive virtual inertial control strategy based on the mapping function, process the preprocessed power compensation demand signal based on the adaptive virtual inertial control strategy, generate energy storage output commands, and optimize the energy storage frequency regulation strategy of the energy storage system according to the energy storage output commands, so as to achieve optimized suppression of grid frequency fluctuations.

[0055] Thirdly, this application provides an electronic device, comprising:

[0056] Memory, used to store computer programs;

[0057] A processor, configured to execute the computer program to implement the steps of the energy storage frequency regulation strategy optimization method based on the artificial bee colony algorithm as described in the first aspect above.

[0058] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the steps of the energy storage frequency regulation strategy optimization method based on the artificial bee colony algorithm described in the first aspect above.

[0059] This application provides a method for optimizing energy storage frequency regulation strategies based on the artificial bee colony algorithm. The method includes: acquiring frequency deviation data generated by the power grid under load abrupt changes based on the energy storage frequency regulation strategy; performing sliding time window processing on the frequency deviation data to extract frequency change features, and performing extreme value normalization processing on the frequency change features corresponding to different sliding windows to form a standard fluctuation sequence characterizing the dynamic frequency characteristics of the power grid; converting the standard fluctuation sequence into an initial power compensation demand signal, and performing phase compensation preprocessing on the initial power compensation demand signal to obtain a preprocessed power compensation demand signal. The system obtains the signal; establishes a parameter optimization model with the differential gain coefficient and response delay time of the virtual inertial control as optimization variables, and uses the artificial bee colony algorithm to iteratively solve the parameter optimization model in multiple rounds to obtain the optimal parameter combination. Based on the optimal parameter combination, a mapping function between frequency and power is constructed. Based on the mapping function, an adaptive virtual inertial control strategy is formed, and the preprocessed power compensation demand signal is processed based on the adaptive virtual inertial control strategy to generate energy storage output commands. The energy storage frequency regulation strategy of the energy storage system is optimized according to the energy storage output commands to achieve optimized suppression of grid frequency fluctuations.

[0060] The technical solution provided in this application has the following beneficial effects:

[0061] This application establishes a real-time sensing capability for the power grid frequency status, providing an accurate data foundation for subsequent frequency regulation control. It accurately captures the dynamic characteristics of frequency changes, unifying fluctuation information at different time scales into a standardized representation. It achieves precise conversion from frequency fluctuations to power demand and improves the synchronization of power signals through phase compensation. Intelligent optimization algorithms obtain control parameters best suited to the current operating conditions, enhancing the adaptability of control. A quantitative correspondence between frequency and power is established, providing a precise mathematical basis for control strategies. A control mechanism capable of autonomously adjusting based on frequency fluctuation characteristics is formed, enhancing the system's responsiveness. Precise control of energy storage power is achieved, improving the stability of the power grid frequency.

[0062] Furthermore, this application establishes a parameter optimization model with virtual inertial control parameters as the optimization object, uses the artificial bee colony algorithm for multiple rounds of iterative solution, determines the optimal combination of differential gain coefficient and response delay time, and constructs a mapping function that includes frequency differentiation and delay compensation based on this, forming a complete parameter optimization and function construction process.

[0063] Furthermore, it achieves automatic matching of control parameters and operating conditions, improves the adaptability and accuracy of virtual inertial control, provides a more accurate control basis for energy storage frequency regulation, and enhances the system's ability to cope with different frequency fluctuations.

[0064] These or other aspects of this application will become more apparent in the following description of the embodiments. Attached Figure Description

[0065] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0066] Figure 1 A flowchart illustrating an energy storage frequency regulation strategy optimization method based on the artificial bee colony algorithm provided in this application embodiment;

[0067] Figure 2 A schematic diagram illustrating a specific implementation of an energy storage frequency regulation strategy optimization method based on the artificial bee colony algorithm provided in this application embodiment;

[0068] Figure 3 This is a schematic diagram of the structure of an energy storage frequency regulation strategy optimization system based on the artificial bee colony algorithm, provided in an embodiment of this application. Detailed Implementation

[0069] Existing virtual inertial control methods based on fixed parameters have significant limitations in dealing with the complex and ever-changing operating conditions of the power grid. These methods use preset control parameters to handle frequency fluctuations of different characteristics, leading to a mismatch between the power compensation response and the actual demands of the power grid. This lack of parameter adaptability causes fluctuations in frequency suppression effectiveness, affecting the stability of system operation, especially when facing frequency disturbances of varying amplitude and rate, where the stability of the control effect needs improvement.

[0070] To address the aforementioned issues, this application proposes an energy storage frequency regulation strategy optimization method based on the artificial bee colony algorithm. By establishing a dynamic parameter optimization mechanism, it achieves adaptive tuning of key parameters for virtual inertial control. This method first performs multi-scale analysis and standardization of frequency fluctuation characteristics, then employs an intelligent optimization algorithm to solve for the optimal parameter combination based on the current operating conditions, and constructs a precise mapping relationship between frequency and power. By forming an adaptive virtual inertial control strategy, this method can dynamically adjust control parameters according to the actual operating state of the power grid, enabling the power compensation response to better match the real-time needs of the power grid. This effectively improves the stability of frequency fluctuation suppression and solves the problem of insufficient adaptability of fixed parameter control methods under complex operating conditions.

[0071] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are merely some embodiments of the present application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0072] The core of this application is to provide an optimization method for energy storage frequency regulation strategy based on the artificial bee colony algorithm. A flowchart of one specific implementation is shown below. Figure 1 As shown, the method includes:

[0073] Step 101: Obtain frequency deviation data generated by the power grid based on the energy storage frequency regulation strategy under the condition of sudden load change.

[0074] In step 101, the load mutation condition refers to the operating state where the electrical load in the power grid changes suddenly and significantly. The energy storage frequency regulation strategy refers to the control method of adjusting the power grid frequency through the charging and discharging of energy storage devices. The frequency deviation data is a sequence of differences between the actual frequency of the power grid and the standard frequency.

[0075] In this embodiment of the application, real-time frequency data of the power grid is continuously collected by a power grid frequency measurement device, and compared and calculated with a standard frequency reference value to obtain a frequency deviation data sequence. This data sequence reflects the fluctuation of the power grid frequency under load change conditions, providing basic data for subsequent analysis.

[0076] For example, in a regional power grid, when large industrial equipment suddenly starts up, the real-time frequency data of the power grid is obtained by a frequency measuring device at a sampling frequency of 100 times per second. The standard frequency reference value is 50 Hz. The frequency deviation data sequence is calculated, in which the maximum frequency deviation reaches 0.15 Hz. This data sequence will be used for subsequent processing and analysis.

[0077] Step 102: Perform sliding time window processing on the frequency deviation data to extract frequency change features, and perform extreme value normalization processing on the frequency change features corresponding to different sliding windows to form a standard fluctuation sequence characterizing the dynamic characteristics of the power grid frequency.

[0078] In step 102, the sliding time window processing is an analytical method that segments continuous data into fixed-length time intervals. Frequency variation characteristics include indicators reflecting frequency fluctuations, such as instantaneous rate of change and trend rate of change. Extreme value normalization is a data processing method that linearly transforms the data to a specific interval based on its maximum and minimum values. The standard fluctuation sequence is a unified data sequence representing the dynamic characteristics of frequency, formed after multi-scale analysis and standardization.

[0079] In this embodiment, firstly, sliding windows of different time lengths are set to segment the frequency deviation data. Instantaneous rate of change characteristics are calculated for the data segments within short time windows, and trend rate of change characteristics are calculated for the data segments within long time windows. Then, extreme value normalization is performed on the feature values ​​obtained from each window. Finally, all normalized feature values ​​are combined in chronological order to form a standard fluctuation sequence.

[0080] For example, frequency deviation data can be processed using two window lengths: 0.2 seconds and 2 seconds. The 0.2-second window is used to calculate the instantaneous rate of change, while the 2-second window is used to calculate the trend rate of change. The slope value is obtained by linear fitting of the trend change rate, where Indicates the change in frequency. This represents the time interval. The calculated eigenvalues ​​are normalized using the following formula: ,in Represents the normalized eigenvalues. These are the original eigenvalues. and These are the minimum and maximum values ​​of the eigenvalues, respectively. Finally, the normalized eigenvalues ​​are arranged in chronological order to form a standard fluctuation sequence.

[0081] Step 103: Convert the standard fluctuation sequence into an initial power compensation demand signal, and perform phase compensation preprocessing on the initial power compensation demand signal to obtain a preprocessed power compensation demand signal.

[0082] In step 103, the initial power compensation demand signal is a preliminary power adjustment command calculated based on frequency fluctuation characteristics. Phase compensation preprocessing is a data processing method that adjusts the phase of the signal to eliminate the effects of time delay. The preprocessed power compensation demand signal is a power command signal that has been phase optimized and can be used for control execution.

[0083] In this embodiment, the standard fluctuation sequence is decomposed into multiple sub-band signals according to different frequency components. The power conversion coefficients corresponding to each sub-band are determined based on the grid impedance characteristics and the energy storage system response capability. The sub-band signals are multiplied by the corresponding coefficients and synthesized into an initial power compensation demand signal. Then, the phase shift of the signal relative to the grid voltage is detected, the compensation value is calculated through the phase compensation model, and the signal is phase-adjusted. Finally, amplitude limiting processing is performed to obtain the pre-processed power compensation demand signal.

[0084] For example, the standard fluctuation sequence is decomposed into three frequency sub-bands: 0.1-0.5Hz, 0.5-2Hz, and 2-5Hz, with power conversion coefficients of 0.8, 1.2, and 1.5 respectively. These sub-bands are then synthesized to obtain the initial power compensation demand signal. A 0.1 radian phase lag is detected relative to the grid voltage. The compensation value is calculated using a phase compensation model, and complex multiplication is employed for phase compensation. Finally, the signal amplitude is limited to ±10 MW to obtain the preprocessed power compensation demand signal.

[0085] Step 104: Establish a parameter optimization model with the differential gain coefficient and response delay time of the virtual inertial control as optimization variables, and use the artificial bee colony algorithm to solve the parameter optimization model in multiple rounds of iteration to obtain the optimal parameter combination. Based on the optimal parameter combination, construct a mapping function between frequency and power.

[0086] In step 104, the parameter optimization model is a deterministic optimization model based on mathematical programming theory. Its model structure includes a decision variable layer, an objective function layer, and a constraint layer. The decision variable layer consists of differential gain coefficients and response delay time. The objective function layer consists of a frequency deviation integral term and a power change rate term. The constraint layer includes parameter range constraints and system stability constraints. The training process is achieved through iterative search using the artificial bee colony algorithm, including five stages: population initialization, fitness calculation, selection operation, crossover and mutation, and population update, ultimately converging to the optimal solution that satisfies the constraints. The artificial bee colony algorithm is a swarm intelligence optimization algorithm that simulates the foraging behavior of bees. The optimal parameter combination is the combination of parameter values ​​that makes the objective function optimal. The mapping function is a functional expression that establishes the mathematical relationship between frequency input and power output.

[0087] In this embodiment, a parameter optimization model is established with the differential gain coefficient and response delay time as optimization variables. An objective function including frequency deviation and power change rate is defined. The search parameters of the artificial bee colony algorithm are set and the population is initialized. Neighborhood search is performed through the hired bee stage. In the observation bee stage, a high-quality solution is selected for deep search. In the scout bee stage, inferior solutions are replaced. The optimal parameter combination is obtained by iterative solution. Based on this combination, a mapping function including frequency differential term and delay compensation term is constructed.

[0088] For example, establish a parameter optimization model with the objective function as follows: ,in Indicates frequency deviation. This represents the rate of change of power. The artificial bee colony parameters were set as follows: 20 hired bees, 20 observation bees, and a maximum of 100 iterations. After 85 iterations, the optimal parameter combination was obtained: a differential gain coefficient of 3.5 and a response delay of 0.15 seconds. Based on this, a mapping function was constructed. Where df / dt represents the rate of change of frequency. This indicates the amount of delay power compensation.

[0089] Step 105: Based on the mapping function, an adaptive virtual inertial control strategy is formed, and the preprocessed power compensation demand signal is processed based on the adaptive virtual inertial control strategy to generate energy storage output command. The energy storage frequency regulation strategy of the energy storage system is optimized according to the energy storage output command to achieve optimized suppression of grid frequency fluctuations.

[0090] In step 105, the adaptive virtual inertial control strategy is a control method that automatically adjusts control parameters based on the system state. The energy storage output command is a control instruction sent to the energy storage system to execute power output. The energy storage system refers to an energy storage device connected to the power grid via a power electronic conversion device. It receives energy storage output commands generated by an artificial bee colony algorithm based on real-time frequency deviation data and frequency change rate data, and dynamically adjusts its charging and discharging power to achieve rapid compensation and stable control of grid frequency fluctuations. The grid provides the energy storage system with frequency deviation signals as the control basis, while the energy storage system provides frequency support to the grid through power output; the two form a closed-loop frequency regulation control circuit.

[0091] In this embodiment, an adaptive virtual inertial control strategy including feedforward control and feedback correction is constructed based on a mapping function. The preprocessed power compensation demand signal is input into the feedforward channel to generate a feedforward power component, and the frequency deviation data is input into the feedback channel to generate a feedback correction component. The two components are fused according to a dynamic weight coefficient to generate a control signal. After being limited by the rate of change and the amplitude, an energy storage output command is formed. The power output of the energy storage system is adjusted according to the command to achieve frequency fluctuation suppression.

[0092] For example, a control strategy is constructed based on a mapping function. The preprocessed power compensation demand signal is differentiated to obtain the feedforward power component, and the frequency deviation data is delayed to obtain the feedback correction component. When the frequency change rate is greater than 0.5 Hz / s, the feedforward weight is 0.7 and the feedback weight is 0.3; otherwise, the feedforward weight is 0.4 and the feedback weight is 0.6. The fused control signal is limited by a change rate of ±20 MW / s and an amplitude limit of ±10 MW to generate an energy storage output command, which adjusts the power output of the energy storage system so that the frequency deviation recovers to within ±0.05 Hz within 2 seconds.

[0093] This method achieves precise suppression of grid frequency fluctuations through multi-scale frequency characteristic analysis, intelligent parameter optimization, and adaptive control strategies. It can automatically adjust control parameters according to the actual operating conditions of the grid, enabling the power output of the energy storage system to better match the grid's frequency regulation needs, thus improving the adaptability and stability of frequency control and effectively enhancing the frequency quality of the grid under sudden load changes.

[0094] To address the insufficient adaptability caused by fixed virtual inertial control parameters, in some embodiments, step 104 involves establishing a parameter optimization model with the differential gain coefficient and response delay time of the virtual inertial control as optimization variables, and using an artificial bee colony algorithm to iteratively solve the parameter optimization model multiple times to obtain the optimal parameter combination. Based on the optimal parameter combination, a mapping function between frequency and power is constructed, such as... Figure 2 As shown, it includes:

[0095] Step 201: Establish a parameter optimization model, set the differential gain coefficient of virtual inertial control as the first optimization variable, and set the response delay time of virtual inertial control as the second optimization variable.

[0096] In step 201, the first optimization variable refers to the differential gain coefficient of the frequency change response intensity in virtual inertial control, and the second optimization variable refers to the response delay time from detection to execution in the control system.

[0097] In this embodiment, the framework of the parameter optimization model is first constructed, and the two key parameters that affect the control performance, namely the differential gain coefficient and the response delay time, are set as variables that need to be optimized. A mathematical model of the optimization problem containing these two variables is established to lay the foundation for subsequent optimization calculations.

[0098] Step 202: Based on the first optimization variable and the second optimization variable, define an objective function that aims to minimize the integral of frequency deviation and the rate of power change.

[0099] In step 202, the objective function is a single-objective function rather than a multi-objective function. It integrates the frequency deviation integral term and the power change rate term into a single comprehensive index through a weighted summation. The weight coefficient is set to 0.05 to balance the contributions of the two optimization objectives, ultimately forming a unified scalar evaluation function to guide the parameter optimization process. The frequency deviation integral is the cumulative amount of frequency deviation from the standard value, and the power change rate is the magnitude of power change per unit time.

[0100] In this embodiment, an objective function is constructed based on two set optimization variables. This function includes a frequency deviation integral term and a power change rate term. By integrating these two indicators reflecting control quality into a unified evaluation standard, a clear direction and objective are provided for parameter optimization.

[0101] Step 203: Based on the characteristics of the objective function, set the search parameters for the artificial bee colony algorithm, including the number of hired bees, the number of observation bees, and the maximum number of iterations.

[0102] In step 203, the characteristics of the objective function refer to the mathematical features exhibited by the function during parameter optimization, including multimodality, nonlinearity, and constraint complexity. These characteristics stem from the dynamic coupling relationship between frequency deviation and power output in virtual inertial control, as well as the safety boundary constraints of power grid operation. Hired bees are responsible for exploring new solutions, while observer bees are responsible for selectively following the optimal solution. The maximum number of iterations is the maximum number of rounds the algorithm can run.

[0103] In this embodiment, the operating parameters of the artificial bee colony algorithm are configured according to the mathematical characteristics of the objective function, including determining the number of hired bees and observer bees, and setting a reasonable maximum number of iterations. The setting of these parameters directly affects the search efficiency and solution quality of the algorithm.

[0104] Step 204: Based on the search parameters, perform multiple rounds of iterative solving using the artificial bee colony algorithm to determine the optimal parameter combination.

[0105] In step 204, the multi-round iterative solution is a process of gradually approaching the optimal solution through repeated improvements.

[0106] In this embodiment, a pre-defined artificial bee colony algorithm is used to perform multiple rounds of optimization calculations. In each iteration, bees are hired to perform neighborhood searches on the current solution. Observer bees select high-quality solutions for in-depth development based on fitness, and scout bees replace stagnant solutions. By continuously updating the population, the optimal parameter combination is finally obtained.

[0107] Step 205: Based on the differential gain coefficient and response delay time in the optimal parameter combination, construct a mapping function that includes a frequency differential term and a delay compensation term.

[0108] In step 205, the differential gain coefficient in the optimal parameter combination represents the degree of influence of the frequency change rate on the power compensation amount, and the response delay time represents the time delay from frequency detection to power output. Both of these parameters are numerical solutions that optimize the objective function after evaluating and selecting each parameter combination based on the objective function value during multiple rounds of iterative search using the artificial bee colony algorithm.

[0109] In the embodiments of this application, a complete mapping function is constructed based on the obtained optimal parameter combination. This function includes a frequency differential term weighted by the differential gain coefficient and a delay compensation term determined by the response delay time, forming a complete mathematical relationship from frequency input to power output.

[0110] Here is a specific example:

[0111] In the frequency regulation scenario of the regional power grid responding to sudden load increases from large industrial equipment, based on the preprocessed power compensation demand signal and frequency deviation data sequence, the differential gain coefficient of the virtual inertial control is set as the first optimization variable when establishing the parameter optimization model. The response delay time of virtual inertial control is set as the second optimization variable. ,in The value range is set to 0.1 to 10.0. The value range is set to 0.01 seconds to 1.0 seconds, and these ranges are determined based on the regulation capability of the energy storage system and the grid security requirements; the objective function is defined based on these two optimization variables. ,in The frequency deviation is expressed in Hertz. The power change rate is expressed in megawatts per second. The weighting coefficient is set to 0.05, which is used to balance the two optimization objectives of frequency stability and power fluctuation. The search parameters of the artificial bee colony algorithm are set according to the nonlinear multi-peak characteristics of the objective function: the number of hired bees is set to 20, the number of observer bees is set to 20, and the maximum number of iterations is set to 100. These parameters were determined through preliminary experiments to achieve a balance between computational efficiency and optimization effect. Based on these search parameters, the artificial bee colony algorithm is used for multiple rounds of iterative solution. A population of 40 parameter combinations is initially generated. In each iteration, hired bees perform neighborhood searches on the current parameter combination, observer bees select high-quality parameter combinations for depth searches based on fitness values, and scout bees replace parameter combinations that have not improved for 10 consecutive iterations. After 85 iterations, the iteration stops when the improvement of the optimal objective function value is less than 0.001 for 5 consecutive iterations. The final optimal parameter combination is obtained as follows: With Td = 0.15 seconds, the objective function value decreases from the initial 2.56 to 0.85; a mapping function is constructed based on the differential gain coefficient of 3.5 and the response delay of 0.15 seconds in the optimal parameter combination. ,in The rate of change of frequency is expressed in Hertz per second. This indicates the power compensation amount after a 0.15-second delay, expressed in megawatts. The compensation coefficient is set to 0.8. This coefficient is obtained through regression analysis of historical operating data. The resulting mapping function will be used to generate an adaptive virtual inertial control strategy.

[0112] In this embodiment, the optimal control parameters are automatically obtained through intelligent optimization algorithms, and a precise frequency-power mapping relationship is established, which improves the adaptability and accuracy of virtual inertial control, enabling the energy storage system to provide more matched power support according to the actual state of the power grid, and enhancing the stability of frequency control.

[0113] To further improve the efficiency and accuracy of parameter optimization, in some embodiments, step 204: based on the search parameters, performing multiple rounds of iterative solving using the artificial bee colony algorithm to determine the optimal parameter combination includes:

[0114] Step 301: Based on the search parameters, initialize and generate an initial population containing multiple parameter combinations, where each parameter combination includes the values ​​of a first optimization variable and a second optimization variable.

[0115] In step 301, the initial population is an initial solution set consisting of multiple parameter combinations, each parameter combination containing a specific value of the differential gain coefficient of the first optimization variable and the response delay time of the second optimization variable.

[0116] In this embodiment of the application, an initial population containing multiple parameter combinations is randomly generated according to a preset search parameter scale. The differential gain coefficient and response delay time in each parameter combination are randomly generated within their allowed value range to form an initial search solution set.

[0117] Step 302: Through the hired bee stage of the artificial bee colony algorithm, a neighborhood search is performed on each parameter combination in the initial population to obtain the updated parameter combination, and the objective function value corresponding to each updated parameter combination is calculated based on the objective function.

[0118] In step 302, the hired bee phase is the phase in the artificial bee colony algorithm responsible for exploring new solutions, the neighborhood search is a method of generating new solutions by making small-scale perturbations in the vicinity of the current solution, and the objective function value is a numerical index used to evaluate the quality of parameter combinations.

[0119] In this embodiment, the hired bees perform a neighborhood search for each parameter combination in the initial population. New parameter combinations are generated by adding random perturbations to the current parameter values. Then, the objective function value corresponding to each new parameter combination is calculated based on the objective function, updating the solutions in the population. Specifically, different parameter combinations directly affect the dynamic response characteristics of the virtual inertial control, specifically impacting the integral value of the frequency deviation and the power change rate. The expression used to calculate the objective function value is: This expression is the defined objective function. Based on the sorting rule of the objective function value from smallest to largest, the optimal parameter combination is selected from multiple parameter combinations, that is, the parameter combination that minimizes the objective function value is selected as the optimal solution.

[0120] Step 303: Through the observation bee stage of the artificial bee colony algorithm, select high-quality parameter combinations with objective function values ​​greater than a preset threshold for deep search to obtain optimized parameter combinations.

[0121] In step 303, the observation bee phase is the phase in the artificial bee colony algorithm responsible for selecting the best solution and following up. The optimal parameter combination refers to the parameter combination with a better objective function value. The depth search is a more refined search process in the vicinity of the optimal solution.

[0122] In this embodiment, the observation bee selects high-quality parameter combinations based on the objective function value of each parameter combination. A probability selection mechanism is used to give parameter combinations with smaller objective function values ​​a higher probability of being selected. The selected high-quality parameter combinations are then subjected to a deep search to obtain further optimized parameter combinations.

[0123] Step 304: Through the scout bee stage of the artificial bee colony algorithm, replace the parameters that are trapped in local optima in the optimized parameter combination to obtain a new parameter combination, and use the new parameter combination as the initial population.

[0124] In step 304, the scout bee phase is the phase in the artificial bee colony algorithm responsible for escaping local optima. Local optima refer to the state where the search is stuck in a local region and cannot be improved further. Parameter replacement is the replacement of poorly performing parameter combinations with newly generated parameter combinations.

[0125] In this embodiment, the scout bee detects those parameter combinations that have not been improved for multiple consecutive rounds in the optimized parameter combinations, and replaces these parameter combinations that are trapped in local optima with randomly generated new parameter combinations to maintain the diversity of the population.

[0126] Step 305: Repeat the iterative search process of the hired bee stage, the observation bee stage, and the scout bee stage until the preset termination condition is met, and output the parameter combination with the optimal objective function value as the optimal parameter combination.

[0127] In step 305, the specific termination conditions include reaching the preset maximum number of iterations, the improvement of the objective function value in multiple consecutive iterations being less than a set threshold, or the distribution of the optimal solution in the search population reaching a predetermined convergence criterion.

[0128] In this embodiment, the search process of the hired bee stage, the observation bee stage, and the scout bee stage is repeated. The search stops when the maximum number of iterations is reached or the optimal solution is not improved in multiple consecutive rounds. The parameter combination with the minimum objective function value in the entire search process is output as the optimal parameter combination.

[0129] Here is a specific example:

[0130] In the frequency regulation scenario of the regional power grid responding to sudden load increases from large industrial equipment, based on the pre-set number of hired bees (20), observation bees (20), and a maximum number of iterations (100), an initial population containing 40 parameter combinations is generated. Each parameter combination includes the first optimization variable, the differential gain coefficient. Second optimization variable response delay time The values ​​that can be taken, these values ​​are in The value range is 0.1 to 10.0 and The value is generated uniformly within the range of 0.01 seconds to 1.0 seconds by a random number generator, where the first parameter combination is [ =2.1, =0.08 seconds], the second parameter combination is [ =5.7, =0.25 seconds], and so on, generating a total of 40 different parameter combinations; through the hired bee stage of the artificial bee colony algorithm, a neighborhood search is performed on each parameter combination in the initial population, and the formula for calculating the new parameter combination is: , ,in The search step size for the differential gain coefficient is 0.2. The search step size for the response delay is 0.05. and Given uniformly distributed random numbers between 0 and 1, calculate the objective function value corresponding to each updated parameter combination. The optimal objective function value in the initial population was calculated to be 2.56. Using the artificial bee colony algorithm, high-quality parameter combinations with objective function values ​​greater than a preset threshold were selected for depth-first search during the observation bee phase. The preset threshold was set to 1.50, and the selection probability was calculated using the following formula: ,in Indicates the first The fitness value of a combination of parameters is equal to , Indicates the first The objective function value of each parameter combination is calculated. A deep search is performed on selected high-quality parameter combinations to obtain optimized parameter combinations. During the scout bee phase of the artificial bee colony algorithm, parameters trapped in local optima are replaced. Parameter combinations whose objective function value improvement is less than 0.001 for 10 consecutive iterations are marked as trapped in local optima and replaced with randomly generated new parameter combinations. These new parameter combinations are used as the initial population for the next iteration. This iterative search process of the hired bee phase, observation bee phase, and scout bee phase is repeated 85 times until a preset termination condition is met: the improvement of the optimal objective function value is less than 0.001 for 5 consecutive iterations. At this point, the optimal objective function value decreases from the initial 2.56 to 0.85. The parameter combination with the best objective function value is output as the optimal parameter combination, where the differential gain coefficient is... Equals 3.5, response latency It equals 0.15 seconds.

[0131] In this embodiment, the multi-stage collaborative search mechanism of the artificial bee colony algorithm achieves efficient optimization of parameter combinations, avoids the search process from getting trapped in local optima, ensures that the globally optimal or near-optimal parameter configuration is obtained, and improves the overall performance of the virtual inertial control system.

[0132] To further improve the accuracy and practicality of the mapping function construction, in some embodiments, step 205: constructing a mapping function containing a frequency derivative term and a delay compensation term based on the differential gain coefficient and response delay time in the optimal parameter combination, includes:

[0133] Step 401: Perform differential processing on the frequency deviation data to obtain the frequency differential signal.

[0134] In step 401, the frequency differential signal is a signal that reflects the rate of frequency change. It is obtained by performing differential operations on the frequency deviation data and characterizes the speed of change of the power grid frequency.

[0135] In this embodiment of the application, the frequency deviation data acquired in real time is processed using a differential calculation method to extract the instantaneous change characteristics of the frequency change and obtain a frequency differential signal that reflects the rate of frequency change.

[0136] Step 402: Multiply the frequency differential signal by the differential gain coefficient as the frequency differential term.

[0137] In step 402, the frequency differential term is the power adjustment component formed by multiplying the frequency differential signal by the differential gain coefficient, which reflects the degree of influence of the frequency change rate on power compensation.

[0138] In this embodiment, the frequency differential signal is multiplied by the differential gain coefficient in the optimal parameter combination to amplify the effect of frequency change on power regulation, and the frequency differential term is generated as an important component of the mapping function.

[0139] Step 403: Perform time delay processing on the preprocessed power compensation demand signal to obtain a delayed power signal.

[0140] In step 403, the delayed power signal is the signal obtained after time delay processing of the power compensation demand signal, reflecting the impact of system response delay on power output.

[0141] In this embodiment, the preprocessed power compensation demand signal is processed by a time delay module, and the delay time adopts the response delay time in the optimal parameter combination to obtain a delayed power signal that takes into account system delay.

[0142] Step 404: Calculate the difference between the delayed power signal and the preprocessed power compensation requirement signal, and use it as the delay compensation item.

[0143] In step 404, the delay compensation term is the difference between the delayed power signal and the original power compensation requirement signal, which is used to compensate for the power deviation caused by system delay.

[0144] In this embodiment, the difference between the delayed power signal and the preprocessed power compensation demand signal is calculated. This difference reflects the power compensation error caused by the system delay and is used as a delay compensation term in the construction of the mapping function.

[0145] Step 405: Combine the frequency derivative term and the delay compensation term in a weighted manner to construct a mapping function.

[0146] In step 405, the weighted combination is a calculation method that merges different components according to a specific weight ratio, used to balance the contribution of each component in the final function.

[0147] In this embodiment, the frequency differential term and the delay compensation term are weighted and summed according to preset weighting coefficients to construct a complete mathematical expression for the mapping function, forming a complete conversion relationship from frequency input to power output.

[0148] Here is a specific example:

[0149] In the frequency regulation scenario of the regional power grid responding to a sudden increase in load from large industrial equipment, based on the obtained optimal parameter combination of a differential gain coefficient of 3.5 and a response delay time of 0.15 seconds, the frequency deviation data is first differentiated to obtain the frequency differential signal, which is then calculated using the center difference method. ,in It represents the rate of change of frequency (the frequency differential signal). This indicates the current frequency value, expressed in Hertz (Hz). This indicates the frequency value at the previous moment, expressed in Hertz (Hz). The sampling time interval is 0.01 seconds. When the frequency changes from 49.92 Hz to 49.90 Hz, the calculated frequency differential signal value is -2.0 Hz per second. Multiplying the frequency differential signal by the differential gain coefficient 3.5 yields the frequency differential term. The calculation process is 3.5 multiplied by -2.0 equals -7.0, which is converted to megawatts, resulting in a frequency differential term value of -7.0 megawatts. Simultaneously, the preprocessed power compensation demand signal undergoes a time delay of 0.15 seconds. When the original power compensation demand signal is 8.5 megawatts, after a 0.15-second delay, the delayed power signal value is 8.5 megawatts. The difference between the delayed power signal and the preprocessed power compensation demand signal is calculated as the delay compensation term. Since the signal values ​​before and after the delay are the same, the delay compensation term calculation result is 0 megawatts. Finally, the frequency differential term and the delay compensation term are combined in a weighted manner to construct a mapping function, using weighting coefficients. and The mapping function expression is Substituting the values, we get P equals 0.7 multiplied by -7.0 plus 0.3 multiplied by 0, which equals -4.9 megawatts. The target power value output by the mapping function is expressed in megawatts. The final complete mapping function expression is: .

[0150] In this embodiment, by constructing a mapping function that includes frequency differentiation and delay compensation, an accurate frequency-power conversion relationship is established, which improves the accuracy and response characteristics of control and provides reliable technical support for energy storage systems to participate in grid frequency regulation.

[0151] To further improve the accuracy and practicality of the power compensation signal, in some embodiments, step 103: converting the standard fluctuation sequence into an initial power compensation demand signal, and performing phase compensation preprocessing on the initial power compensation demand signal to obtain a preprocessed power compensation demand signal, includes:

[0152] Step 501: Decompose the standard fluctuation sequence into multiple frequency sub-band signals, and determine the power conversion coefficient corresponding to each frequency sub-band signal based on the grid impedance parameters and the energy storage system response characteristics.

[0153] In step 501, the frequency sub-band signal is the signal component obtained by decomposing the standard fluctuation sequence according to different frequency ranges. The grid impedance parameter is obtained by performing frequency domain impedance scanning tests on grid nodes, characterizing the impedance characteristics of the grid at different frequencies. The energy storage system response characteristics are obtained by performing step response tests on the energy storage converter, characterizing the dynamic response capability of the energy storage system under different operating conditions. The power conversion coefficient is the conversion ratio from each frequency band signal to a power signal determined based on the grid impedance characteristics and the energy storage system response capability. The decomposition of the standard fluctuation sequence is processed using a digital filter bank method, based on the main oscillation mode characteristics of the grid frequency fluctuations. Specifically, the signal is separated by setting three characteristic frequency bands: 0.1-0.5Hz, 0.5-2Hz, and 2-5Hz. These frequency band ranges are determined based on the typical low-frequency oscillation frequency distribution of the power system.

[0154] In this embodiment, the standard fluctuation sequence is decomposed into multiple sub-band signals with different frequency ranges by a filter bank. Based on the frequency characteristics of the grid impedance and the response capability of the energy storage system in different frequency bands, a suitable power conversion coefficient is assigned to each frequency sub-band.

[0155] Step 502: Multiply each frequency sub-band signal by its corresponding power conversion coefficient to obtain the power compensation signal of each frequency sub-band signal, and synthesize the power compensation signals of each frequency sub-band signal to generate the initial power compensation requirement signal.

[0156] In step 502, the power compensation signal is the power adjustment component obtained after power conversion of each frequency sub-band signal, and the initial power compensation demand signal is the total power demand signal after the power compensation signals of each frequency band are synthesized.

[0157] In this embodiment, the power compensation signal of each frequency sub-band is multiplied by the corresponding power conversion coefficient to obtain the power compensation signal of each sub-band. Then, the power compensation signals of all sub-bands are superimposed and synthesized according to time points to generate a complete initial power compensation requirement signal.

[0158] Step 503: Calculate the phase offset of the initial power compensation demand signal relative to the grid voltage signal using an adaptive filtering algorithm, and calculate the real-time phase compensation value based on the phase offset and the pre-established phase compensation model.

[0159] In step 503, the grid voltage signal is the instantaneous three-phase voltage signal directly acquired from the grid connection point. The phase offset is the time lag of the initial power compensation demand signal relative to the grid voltage signal. The construction process of the phase compensation model includes: establishing a switching delay function based on the switching characteristics of the power devices in the energy storage system; constructing a signal transmission delay function in combination with grid transmission line parameters; determining the sampling delay function through the sampling period of the grid frequency measurement unit; and superimposing the three types of delay functions in the time domain to form a complete phase compensation model. This model is a deterministic mathematical model based on grid parameters and device characteristics. The model structure includes a series combination of a switching delay module, a transmission delay module, and a sampling delay module. The specific process of model training includes: collecting actual phase deviation data under different operating conditions as training samples; using grid operating parameters and power commands as model inputs; adjusting the delay parameters in the model through a gradient descent algorithm; using the root mean square error as a loss function to evaluate the degree of agreement between the model output and the actual phase deviation; iteratively optimizing until the model accuracy meets the predetermined requirements; and finally obtaining a trained model that can accurately predict the phase compensation value. The real-time phase compensation value is the phase value that needs to be compensated, calculated based on the current phase offset.

[0160] In this embodiment of the application, an adaptive filtering algorithm is used to compare the waveforms of the initial power compensation demand signal and the grid voltage signal, calculate the phase offset between the two, and then calculate the real-time phase compensation value based on the current phase offset using a pre-established phase compensation model.

[0161] Step 504: Apply the real-time phase compensation value to the initial power compensation demand signal to obtain the phase-compensated power compensation demand signal.

[0162] In step 504, the phase-compensated power compensation demand signal is a power demand signal that has been phase-adjusted and synchronized with the grid voltage.

[0163] In this embodiment, the calculated real-time phase compensation value is applied to the initial power compensation demand signal through a phase rotation operation, synchronizing the power demand signal with the grid voltage signal to obtain a phase-compensated power compensation demand signal. Specifically, applying the real-time phase compensation value to the initial power compensation demand signal is accomplished through complex multiplication. First, the initial power compensation demand signal is converted into a complex number form, and then multiplied by the phase compensation factor. ,in The real-time phase compensation value is used, and the real part of the product is taken as the phase-compensated signal. In the embodiment, when the detected phase offset is 0.1 radians, The initial power compensation demand signal Output in the form of .

[0164] Step 505: Perform amplitude limiting processing on the phase-compensated power compensation demand signal to obtain a preprocessed power compensation demand signal.

[0165] In step 505, amplitude limiting processing is a signal processing method that limits the signal amplitude to an allowable range. The preprocessed power compensation demand signal is the final power command signal obtained after a complete preprocessing process.

[0166] In this embodiment, the amplitude of the phase-compensated power compensation demand signal is checked, and the signal amplitude exceeding the allowable range of the energy storage system is limited to between the maximum and minimum values, so as to obtain a pre-processed power compensation demand signal that meets the system safety requirements.

[0167] Here is a specific example:

[0168] In a frequency regulation scenario for a regional power grid to cope with a sudden increase in load from large industrial equipment, based on an established standard fluctuation sequence, the standard fluctuation sequence is first decomposed into three frequency sub-band signals using a digital filter bank: a low-frequency sub-band signal of 0.1-0.5 Hz, a mid-frequency sub-band signal of 0.5-2 Hz, and a high-frequency sub-band signal of 2-5 Hz. These frequency band divisions are determined based on the characteristics of the main oscillation modes of the power grid. Based on the test results of the power grid impedance parameters and the dynamic response characteristics of the energy storage system, the power conversion coefficients corresponding to each frequency sub-band signal are determined to be 0.8 for low frequency, 1.2 for mid-frequency, and 1.5 for high frequency. These coefficients are obtained by fitting the impedance frequency characteristic curve. The power compensation signal for each frequency sub-band signal is obtained by multiplying each frequency sub-band signal by its corresponding power conversion coefficient. The calculation formula is as follows: ,in Indicates the first The power compensation signal unit for each sub-band is megawatts. Indicates the first The amplitude of each sub-band, Indicates the first The power conversion coefficients corresponding to each sub-band are used. The power compensation signals from the three sub-bands are synthesized to generate the initial power compensation demand signal. The synthesis method uses weighted summation, with weighting coefficients of 0.4, 0.35, and 0.25 based on the importance of each frequency band. The calculated initial power compensation demand signal value is 8.5 MW. An adaptive filtering algorithm is used to calculate the phase shift of the initial power compensation demand signal relative to the grid voltage signal. The phase difference is obtained by comparing the zero-crossing time difference between the two signals. The calculation formula is as follows: ,in The phase offset is expressed in radians. The fundamental frequency of the power grid is 50 Hz. The zero-crossing time difference was measured to be 0.000318 seconds, and the calculated phase offset was 0.1 radians. Based on this phase offset and a pre-established phase compensation model... ,in The real-time phase compensation value is expressed in radians, and the calculated real-time phase compensation value is -0.085 radians. The real-time phase compensation value is applied to the initial power compensation demand signal through complex multiplication, calculated using the following formula: ,in This represents the power compensation demand signal after phase compensation. The initial power compensation demand signal is represented, and the phase-compensated power compensation demand signal is obtained. This signal is then subjected to amplitude limiting processing. Based on the energy storage system capacity, the limiting range is set to -10 MW to +10 MW, limiting the signal amplitude within this range. The final pre-processed power compensation demand signal value is 8.5 MW.

[0169] In this embodiment, a complete preprocessing process, including frequency subband decomposition, power conversion, phase compensation, and amplitude limiting, is used to obtain a power compensation signal that is synchronized with the power grid and meets system safety requirements, providing an accurate and reliable input for subsequent control strategies.

[0170] To further improve the comprehensiveness and accuracy of frequency feature extraction, in some embodiments, step 102: performing sliding time window processing on the frequency deviation data to extract frequency change features, and performing extreme value normalization processing on the frequency change features corresponding to different sliding windows to form a standard fluctuation sequence characterizing the dynamic characteristics of the power grid frequency, including:

[0171] Step 601: Set multiple sliding time windows of different durations, wherein the sliding time windows include a first window for capturing fluctuation characteristics and a second window for identifying changing trends.

[0172] In step 601, the sliding time window is a fixed-length time period that slides along the time axis. The first window is a short time window used to capture rapid fluctuations, and the second window is a long time window used to identify slow trends. The length of the first window is shorter than the length of the second window. The first window uses 0.2 seconds to capture rapid fluctuations, and the second window uses 2 seconds to identify slow trends. This setting ensures that both instantaneous changes and long-term evolutionary features of the frequency can be captured simultaneously.

[0173] In this embodiment of the application, two sliding time windows of different lengths are set according to the characteristics of power grid frequency fluctuations. The shorter first window is used to capture instantaneous fluctuation characteristics, and the longer second window is used to identify long-term change trends.

[0174] Step 602: Move the first window and the second window sequentially according to a preset time interval, and extract the frequency data segments corresponding to the time period from the frequency deviation data respectively.

[0175] In step 602, the preset time interval is the time step of the window movement, and the frequency data segment is a data fragment within a continuous time period extracted from the frequency deviation data.

[0176] In this embodiment, two windows are moved sequentially at fixed time intervals. After each movement, data for the corresponding time period is extracted from the frequency deviation data to form a continuously covered data segment sequence. Specifically, moving the windows sequentially at preset time intervals means sliding the two windows on the time axis with a fixed step size of 0.01 seconds. After each movement, data for the corresponding time period is extracted again to form an analysis sequence that continuously covers all frequency data.

[0177] Step 603: Perform differential calculation on the frequency data segments corresponding to each time period of the first window to generate velocity feature quantities that reflect instantaneous changes.

[0178] In step 603, difference calculation is a mathematical method for calculating the difference between adjacent data, and velocity characteristic quantity is a characteristic index that reflects the instantaneous change rate of frequency.

[0179] In this embodiment of the application, differential calculation is performed on each frequency data segment captured by the first window to obtain a velocity characteristic quantity that reflects the instantaneous rate of frequency change.

[0180] Step 604: Perform linear fitting on the frequency data segments corresponding to each second window time period to generate slope feature quantities that reflect the continuous changing trend.

[0181] In step 604, linear fitting is a method that uses a straight line to approximate the trend of data change, and the slope feature is a trend indicator that reflects the direction of continuous frequency change.

[0182] In this embodiment of the application, each frequency data segment captured by the second window is linearly fitted to obtain a slope feature quantity that reflects the frequency change trend.

[0183] Step 605: Combine the velocity feature and the slope feature as corresponding frequency change features to form a frequency change feature set, and identify the maximum and minimum values ​​of each feature in the frequency change feature set.

[0184] In step 605, the frequency change feature set is a complete set containing all feature quantities, and the maximum and minimum values ​​are the limits of the feature quantities within the statistical range.

[0185] In this embodiment of the application, the calculated velocity and slope features are combined to form a complete feature set, and the maximum and minimum values ​​of each feature are found throughout the entire statistical period.

[0186] Step 606: Based on the maximum value and the minimum value, perform extreme value normalization processing on each feature quantity in the frequency change feature set, and rearrange the frequency change feature set after extreme value normalization processing in chronological order to form a standard fluctuation sequence.

[0187] In step 606, extreme value normalization is a standardization method that linearly transforms the data to a specific interval based on its maximum and minimum values. The standard fluctuation sequence representation is achieved by arranging the normalized values ​​of velocity and slope characteristics in chronological order. The velocity characteristic reflects the rate of frequency change, while the slope characteristic reflects the direction of frequency change. Together, they comprehensively describe the dynamic behavior of the power grid frequency in both time and amplitude dimensions.

[0188] In this embodiment, each feature is normalized based on the found maximum and minimum values, and then the processed feature values ​​are rearranged in chronological order to form a standard fluctuation sequence.

[0189] Here is a specific example:

[0190] In a frequency regulation scenario for a regional power grid to cope with a sudden increase in load from large industrial equipment, based on the obtained frequency deviation data sequence, two sliding time windows of different lengths are first set. The first window has a length of 0.2 seconds to capture fluctuation characteristics, and the second window has a length of 2 seconds to identify changing trends. These window lengths are determined based on the main time constants of power grid frequency fluctuations. The first and second windows are moved sequentially at preset time intervals of 0.01 seconds, respectively extracting frequency data segments corresponding to the time period from the frequency deviation data. The first window extracts 21 sampling points each time, and the second window extracts 201 sampling points each time. Differential calculation is performed on the frequency data segments corresponding to the time period of each first window, using the following formula: ,in Indicates the first The velocity characteristic of each sampling point is measured in Hertz per second. Indicates the first The frequency value of each sampling point is in Hertz. Indicates the first The frequency value of each sampling point is in Hertz. The sampling time interval is 0.01 seconds. When the frequency changes from 49.95 Hz to 49.93 Hz, the calculated velocity characteristic is -2.0 Hz per second. Linear fitting is performed on the frequency data segments corresponding to each second window time period, using the least squares method to fit the straight line. ,in This represents the frequency prediction value after linear fitting. This represents the intercept of the linearly fitted line. The slope characteristic quantity is expressed in Hertz per second. Representing the time variable, a slope of -1.5 Hz per second indicates a decreasing frequency trend. The velocity and slope features are combined as corresponding frequency change features to form a frequency change feature set. Analysis of historical data from the past 24 hours identifies the maximum value of the velocity feature as 15 Hz and the minimum as -15 Hz per second, and the maximum value of the slope feature as 8 Hz and the minimum as -8 Hz per second. Based on these maximum and minimum values, extreme value normalization is performed on each feature in the frequency change feature set, using the following formula: Substituting the velocity characteristic of -2.0 Hz per second into the formula yields a normalized value of 0.433, and substituting the slope characteristic of -1.5 Hz per second into the formula yields a normalized value of 0.406. The frequency variation characteristic set after extreme value normalization is rearranged in chronological order and combined using an alternating arrangement to form a standard fluctuation sequence [0.433, 0.406, 0.450, 0.388, ...]. Odd-numbered positions represent the normalized values ​​of the velocity characteristic, and even-numbered positions represent the normalized values ​​of the slope characteristic. This sequence fully characterizes the dynamic characteristics of the power grid frequency.

[0191] In this embodiment, through multi-scale window analysis and feature normalization processing, a standard sequence that can comprehensively characterize the dynamic characteristics of the power grid frequency is formed, providing accurate feature input for subsequent power compensation.

[0192] To further improve the adaptability and response performance of the control strategy, in some embodiments, step 105: forming an adaptive virtual inertial control strategy based on the mapping function, and processing the preprocessed power compensation demand signal based on the adaptive virtual inertial control strategy to generate energy storage output commands, includes:

[0193] Step 701: Based on the frequency derivative term and delay compensation term in the mapping function, construct the feedforward control channel and the feedback correction channel respectively.

[0194] In step 701, the feedforward control channel is a fast response channel constructed based on the frequency differential term, and the feedback correction channel is an error correction channel constructed based on the delay compensation term.

[0195] In this embodiment, a feedforward control channel is constructed based on the frequency differential term in the mapping function to achieve fast response, while a feedback correction channel is constructed based on the delay compensation term for system delay compensation. Specifically, the frequency differential term in the mapping function is directly converted into a control channel, i.e., a differential gain module is established to perform differential amplification processing on the input signal; the specific implementation process for constructing the feedback correction channel is to convert the delay compensation term in the mapping function into a feedback loop, i.e., a delay compensation module is established to perform delay processing on the frequency deviation signal before participating in the control.

[0196] Step 702: Connect the feedforward control channel and the feedback correction channel in parallel to form an adaptive virtual inertial control strategy.

[0197] In this embodiment, the feedforward control channel and the feedback correction channel are connected in parallel to form a complete adaptive virtual inertial control strategy framework.

[0198] Step 703: Based on the adaptive virtual inertial control strategy, perform differential gain processing on the preprocessed power compensation demand signal to generate a feedforward power component, and perform delay compensation processing on the frequency deviation data to generate a feedback correction component.

[0199] In step 703, differential gain processing is the process of differentially amplifying the signal to obtain a feedforward component, and delay compensation processing is the process of delay-correcting the signal to obtain a feedback component.

[0200] In this embodiment, the preprocessed power compensation demand signal is input into the feedforward channel for differential gain processing to obtain the feedforward power component, and the frequency deviation data is input into the feedback channel for delay compensation processing to obtain the feedback correction component.

[0201] Step 704: The feedforward power component and the feedback correction component are weighted and fused according to the corresponding dynamic weight coefficients to generate an untuned control signal.

[0202] In step 704, the dynamic weighting coefficient refers to a set of proportional parameters that are automatically adjusted according to the real-time operating status of the power grid. These parameters include the feedforward weighting coefficient and the feedback weighting coefficient, which are used to adjust the relative weights of the feedforward power component and the feedback correction component in the fusion process, respectively. These two coefficients complement each other and their sum remains at 1. When a rapid frequency change is detected, the feedforward weighting coefficient is automatically increased to 0.7 and the feedback weighting coefficient is correspondingly decreased to 0.3. When the frequency change slows down, the feedforward weighting coefficient is decreased to 0.4 and the feedback weighting coefficient is increased to 0.6.

[0203] In this embodiment, a dynamic weighting coefficient is determined based on the real-time operating status of the power grid, and the feedforward power component and the feedback correction component are weighted and fused according to their corresponding weights to generate an untuned control signal.

[0204] Step 705: Perform power change rate limiting processing on the untuned control signal to obtain a rate-limited signal, and perform power upper and lower limit truncation processing on the rate-limited signal to obtain the energy storage output command.

[0205] In step 705, the power change rate limiting process is the process of limiting the speed of signal change, and the power upper and lower limit truncation process is the process of limiting the signal amplitude within the allowable range.

[0206] In this embodiment, the power change rate of the untuned control signal is limited to ensure smooth change, and then the upper and lower power limits are truncated to ensure safe operation, ultimately obtaining the energy storage output command.

[0207] Here is a specific example:

[0208] In a frequency regulation scenario for a regional power grid to cope with a sudden increase in load from large industrial equipment, based on the established mapping function... First, based on the frequency differential term in the mapping function Construct a feedforward control channel based on delay compensation term A feedback correction channel is constructed. The feedforward control channel and the feedback correction channel are connected in parallel to form an adaptive virtual inertial control strategy. Based on this adaptive virtual inertial control strategy, the preprocessed power compensation demand signal of 8.5 MW is subjected to differential gain processing, and the calculation formula is: the feedforward power component equals... ,in This represents the rate of change of the power compensation demand signal after preprocessing. When the detected rate of change of the power signal is -2.0 MW / s, the calculated feedforward power component is -7.0 MW. Simultaneously, delay compensation processing is performed on the frequency deviation data, with a delay time of 0.15 seconds. The calculation formula is: the feedback correction component equals... ,in This represents the frequency deviation after a 0.15-second delay. When the frequency deviation before and after the delay is both -0.1 Hz, the calculated feedback correction component is -0.08 MW. The feedforward power component and the feedback correction component are weighted and fused according to their corresponding dynamic weighting coefficients. When the detected frequency change rate is -2.0 Hz per second, which is greater than the set threshold of 0.5 Hz per second, a feedforward weighting coefficient of 0.7 and a feedback weighting coefficient of 0.3 are used. The calculation formula is: the untuned control signal equals... ,in This indicates an untuned control signal. This indicates that the feedforward weighting coefficient is 0.7. The feedback weighting coefficient is set to 0.3. Substituting the values, the untuned control signal is calculated to be 0.7 multiplied by -7.0 plus 0.3 multiplied by -0.08, which equals -4.924 MW. This untuned control signal is then subjected to a power rate limiting process, with a limit of ±20 MW per second. Since the signal rate of change is within the allowable range, the rate-limited signal value is -4.924 MW. Finally, the rate-limited signal is truncated to upper and lower power limits. Based on the rated capacity of the energy storage system, the upper limit is set to 10 MW and the lower limit to -10 MW. Since -4.924 MW is within the allowable range, the final energy storage output command is -4.9 MW, which will be sent to the energy storage system for power output.

[0209] In the embodiments of this application, a fast-response and stable control strategy is formed through feedforward parallel control and dynamic weight adjustment, which realizes precise control of power compensation and effectively improves the regulation quality of grid frequency.

[0210] Figure 3 This application provides a schematic diagram of the structure of an energy storage frequency regulation strategy optimization system based on the artificial bee colony algorithm. The specific implementation method is described below:

[0211] The acquisition module 31 is used to acquire frequency deviation data generated by the power grid based on the energy storage frequency regulation strategy under the condition of sudden load change.

[0212] The extraction module 32 is used to perform sliding time window processing on the frequency deviation data to extract frequency change features, and to perform extreme value normalization processing on the frequency change features corresponding to different sliding windows to form a standard fluctuation sequence characterizing the dynamic characteristics of the power grid frequency.

[0213] The conversion module 33 is used to convert the standard fluctuation sequence into an initial power compensation demand signal, and to perform phase compensation preprocessing on the initial power compensation demand signal to obtain a preprocessed power compensation demand signal.

[0214] Module 34 is established to build a parameter optimization model with the differential gain coefficient and response delay time of virtual inertial control as optimization variables, and to perform multiple rounds of iterative solution on the parameter optimization model using the artificial bee colony algorithm to obtain the optimal parameter combination. Based on the optimal parameter combination, a mapping function between frequency and power is constructed.

[0215] The generation module 35 is used to form an adaptive virtual inertial control strategy based on the mapping function, process the preprocessed power compensation demand signal based on the adaptive virtual inertial control strategy, generate energy storage output command, and optimize the energy storage frequency regulation strategy of the energy storage system according to the energy storage output command, so as to achieve optimized suppression of grid frequency fluctuations.

[0216] The energy storage frequency regulation strategy optimization system based on the artificial bee colony algorithm in this application embodiment is used to implement the aforementioned energy storage frequency regulation strategy optimization method based on the artificial bee colony algorithm. Therefore, the specific implementation of the energy storage frequency regulation strategy optimization system based on the artificial bee colony algorithm can be seen in the embodiment section of the energy storage frequency regulation strategy optimization method based on the artificial bee colony algorithm above. The specific implementation can be referred to the description of the corresponding embodiment, and will not be repeated here.

[0217] This application also provides an electronic device, including: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of the above-described energy storage frequency regulation strategy optimization method based on the artificial bee colony algorithm.

[0218] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of any of the above-described energy storage frequency regulation strategy optimization methods based on the artificial bee colony algorithm.

[0219] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as USB flash drives, read-only memory, random access memory, portable hard drives, magnetic disks, or optical disks.

[0220] The embodiments of this application also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the embodiments of the energy storage frequency regulation strategy optimization method based on the artificial bee colony algorithm.

[0221] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0222] The foregoing has provided a detailed description of the energy storage frequency regulation strategy optimization method, system, electronic device, and storage medium based on the artificial bee colony algorithm provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and its core ideas. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from its principles, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A method for optimizing energy storage frequency regulation strategies based on artificial bee colony algorithm, characterized in that, include: Acquire frequency deviation data generated by the energy storage frequency regulation strategy under sudden load changes in the power grid; The frequency deviation data is processed by a sliding time window to extract frequency change features, and the frequency change features corresponding to different sliding windows are subjected to extreme value normalization processing to form a standard fluctuation sequence characterizing the dynamic characteristics of the power grid frequency. The standard fluctuation sequence is converted into an initial power compensation demand signal, and the initial power compensation demand signal is preprocessed with phase compensation to obtain a preprocessed power compensation demand signal. A parameter optimization model is established with the differential gain coefficient and response delay time of virtual inertial control as optimization variables. The artificial bee colony algorithm is used to solve the parameter optimization model in multiple rounds to obtain the optimal parameter combination. Based on the optimal parameter combination, a mapping function between frequency and power is constructed. Based on the mapping function, an adaptive virtual inertial control strategy is formed, and the preprocessed power compensation demand signal is processed based on the adaptive virtual inertial control strategy to generate energy storage output command. The energy storage frequency regulation strategy of the energy storage system is optimized according to the energy storage output command to achieve optimized suppression of grid frequency fluctuations. The process involves establishing a parameter optimization model with the differential gain coefficient and response delay time of the virtual inertial control as optimization variables, and using an artificial bee colony algorithm to iteratively solve the parameter optimization model in multiple rounds to obtain the optimal parameter combination. Based on the optimal parameter combination, a mapping function between frequency and power is constructed, including: Establish a parameter optimization model, setting the differential gain coefficient of virtual inertial control as the first optimization variable and the response delay time of virtual inertial control as the second optimization variable; Based on the first and second optimization variables, an objective function is defined that aims to minimize the integral of frequency deviation and the rate of power change. Based on the characteristics of the objective function, the search parameters of the artificial bee colony algorithm are set, including the number of hired bees, the number of observation bees, and the maximum number of iterations. Based on the search parameters, the artificial bee colony algorithm is used to perform multiple rounds of iterative solutions to determine the optimal parameter combination. Based on the differential gain coefficient and response delay time in the optimal parameter combination, a mapping function containing a frequency differential term and a delay compensation term is constructed.

2. The energy storage frequency regulation strategy optimization method based on artificial bee colony algorithm according to claim 1, characterized in that, The process of determining the optimal parameter combination by performing multiple rounds of iterative solving using the artificial bee colony algorithm based on the search parameters includes: Based on the search parameters, an initial population containing multiple parameter combinations is generated, and each parameter combination includes the values ​​of a first optimization variable and a second optimization variable. In the hired bee stage of the artificial bee colony algorithm, a neighborhood search is performed on each parameter combination in the initial population to obtain the updated parameter combination, and the objective function value corresponding to each updated parameter combination is calculated based on the objective function. By observing the bee phase of the artificial bee colony algorithm, high-quality parameter combinations with objective function values ​​greater than a preset threshold are selected for deep search to obtain optimized parameter combinations. By using the scout bee phase of the artificial bee colony algorithm, the parameters trapped in local optima in the optimized parameter combination are replaced to obtain a new parameter combination, and the new parameter combination is used as the initial population. The iterative search process of the hired bee phase, the observation bee phase, and the scout bee phase is repeated until the preset termination condition is met, and the parameter combination with the optimal objective function value is output as the optimal parameter combination.

3. The energy storage frequency regulation strategy optimization method based on artificial bee colony algorithm according to claim 1, characterized in that, The mapping function, which includes a frequency derivative term and a delay compensation term, is constructed based on the differential gain coefficient and response delay time in the optimal parameter combination, including: The frequency deviation data is differentiated to obtain the frequency differential signal; The result of multiplying the frequency differential signal by the differential gain coefficient is taken as the frequency differential term; The preprocessed power compensation demand signal is subjected to time delay processing to obtain a delayed power signal; The difference between the delayed power signal and the preprocessed power compensation demand signal is calculated and used as the delay compensation term; The frequency derivative term and the delay compensation term are combined in a weighted manner to construct a mapping function.

4. The energy storage frequency regulation strategy optimization method based on artificial bee colony algorithm according to claim 1, characterized in that, The step of converting the standard fluctuation sequence into an initial power compensation demand signal and performing phase compensation preprocessing on the initial power compensation demand signal to obtain a preprocessed power compensation demand signal includes: The standard fluctuation sequence is decomposed into multiple frequency sub-band signals, and the power conversion coefficient corresponding to each frequency sub-band signal is determined based on the grid impedance parameters and the energy storage system response characteristics. Each frequency sub-band signal is multiplied by its corresponding power conversion coefficient to obtain the power compensation signal for each frequency sub-band signal. The power compensation signals of each frequency sub-band signal are then combined to generate the initial power compensation requirement signal. An adaptive filtering algorithm is used to calculate the phase offset of the initial power compensation demand signal relative to the grid voltage signal, and based on the phase offset and a pre-established phase compensation model, the real-time phase compensation value is calculated. The real-time phase compensation value is applied to the initial power compensation demand signal to obtain the phase-compensated power compensation demand signal. The phase-compensated power compensation demand signal is subjected to amplitude limiting processing to obtain a preprocessed power compensation demand signal.

5. The energy storage frequency regulation strategy optimization method based on artificial bee colony algorithm according to claim 1, characterized in that, The process of performing sliding time window processing on the frequency deviation data to extract frequency change features, and performing extreme value normalization processing on the frequency change features corresponding to different sliding windows to form a standard fluctuation sequence characterizing the dynamic characteristics of the power grid frequency, includes: Multiple sliding time windows of different durations are set, the sliding time windows including a first window for capturing fluctuation characteristics and a second window for identifying changing trends; The first window and the second window are moved sequentially according to a preset time interval, and frequency data segments corresponding to the time period are extracted from the frequency deviation data respectively. Differential calculations are performed on the frequency data segments corresponding to each time period of the first window to generate velocity characteristic quantities that reflect instantaneous changes; Linear fitting is performed on the frequency data segments corresponding to each second window time period to generate slope feature quantities that reflect the continuous changing trend. The velocity feature and the slope feature are combined as corresponding frequency change features to form a frequency change feature set, and the maximum and minimum values ​​of each feature in the frequency change feature set are identified. Based on the maximum and minimum values, extreme value normalization is performed on each feature in the frequency change feature set, and the frequency change feature set after extreme value normalization is rearranged in chronological order to form a standard fluctuation sequence.

6. The energy storage frequency regulation strategy optimization method based on artificial bee colony algorithm according to claim 1, characterized in that, The process of forming an adaptive virtual inertial control strategy based on the mapping function, and processing the preprocessed power compensation demand signal based on the adaptive virtual inertial control strategy to generate energy storage output commands, includes: Based on the frequency derivative term and delay compensation term in the mapping function, a feedforward control channel and a feedback correction channel are constructed respectively. The feedforward control channel and the feedback correction channel are connected in parallel to form an adaptive virtual inertial control strategy; Based on the adaptive virtual inertial control strategy, the preprocessed power compensation demand signal is subjected to differential gain processing to generate a feedforward power component, and the frequency deviation data is subjected to delay compensation processing to generate a feedback correction component. The feedforward power component and the feedback correction component are weighted and fused according to their corresponding dynamic weighting coefficients to generate an untuned control signal. The untuned control signal is subjected to power change rate limiting processing to obtain a rate-limited signal, and the rate-limited signal is subjected to power upper and lower limit truncation processing to obtain the energy storage output command.

7. A system for optimizing energy storage frequency regulation strategies based on artificial bee colony algorithm, characterized in that, include: The acquisition module is used to acquire frequency deviation data generated by the power grid based on the energy storage frequency regulation strategy under the condition of sudden load change; The extraction module is used to perform sliding time window processing on the frequency deviation data to extract frequency change features, and to perform extreme value normalization processing on the frequency change features corresponding to different sliding windows to form a standard fluctuation sequence characterizing the dynamic characteristics of the power grid frequency. The conversion module is used to convert the standard fluctuation sequence into an initial power compensation demand signal, and to perform phase compensation preprocessing on the initial power compensation demand signal to obtain a preprocessed power compensation demand signal. A module is established to build a parameter optimization model with the differential gain coefficient and response delay time of virtual inertial control as optimization variables. The artificial bee colony algorithm is used to solve the parameter optimization model in multiple rounds to obtain the optimal parameter combination. Based on the optimal parameter combination, a mapping function between frequency and power is constructed. The generation module is used to form an adaptive virtual inertial control strategy based on the mapping function, process the preprocessed power compensation demand signal based on the adaptive virtual inertial control strategy, generate energy storage output command, and optimize the energy storage frequency regulation strategy of the energy storage system according to the energy storage output command, so as to achieve the optimization and suppression of grid frequency fluctuations. The process involves establishing a parameter optimization model with the differential gain coefficient and response delay time of the virtual inertial control as optimization variables, and using an artificial bee colony algorithm to iteratively solve the parameter optimization model in multiple rounds to obtain the optimal parameter combination. Based on the optimal parameter combination, a mapping function between frequency and power is constructed, including: Establish a parameter optimization model, setting the differential gain coefficient of virtual inertial control as the first optimization variable and the response delay time of virtual inertial control as the second optimization variable; Based on the first and second optimization variables, an objective function is defined that aims to minimize the integral of frequency deviation and the rate of power change. Based on the characteristics of the objective function, the search parameters of the artificial bee colony algorithm are set, including the number of hired bees, the number of observation bees, and the maximum number of iterations. Based on the search parameters, the artificial bee colony algorithm is used to perform multiple rounds of iterative solutions to determine the optimal parameter combination. Based on the differential gain coefficient and response delay time in the optimal parameter combination, a mapping function containing a frequency differential term and a delay compensation term is constructed.

8. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor, configured to execute the computer program to implement the steps of the energy storage frequency regulation strategy optimization method based on the artificial bee colony algorithm as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, enables the optimization method for energy storage frequency regulation strategy based on the artificial bee colony algorithm as described in any one of claims 1 to 6.

Images