A primary frequency modulation method based on PSO-MPC combined control

Abstract

A primary frequency regulation method based on PSO-MPC combined control belongs to the technical field of power system frequency regulation. The method combines particle swarm optimization (PSO) algorithm and model predictive control (MPC) to realize more efficient frequency regulation control in a wind and storage combined system. The PSO algorithm is used to dynamically optimize key control parameters of the energy storage system, such as the upper and lower limits of SOC and frequency deviation constraints, to ensure that the energy storage system can quickly respond and stably operate when frequency disturbance occurs. The MPC performs rolling prediction and real-time optimization of power distribution for wind power and energy storage systems to ensure that the system can quickly and accurately perform primary frequency regulation under wind speed fluctuations and load changes. Compared with the traditional PI control method, the PSO-MPC combined control strategy of the present application significantly improves the frequency recovery speed of the system, reduces the frequency drop, and maintains excellent steady-state accuracy.

Description

Technical Field

[0001] This invention belongs to the field of power system frequency regulation technology, and relates to a wind-storage joint primary frequency regulation control strategy based on particle swarm optimization (PSO) and model predictive control (MPC) optimization, which aims to improve the frequency stability and response capability of the power system under high wind power penetration conditions. Background Technology

[0002] As global climate change becomes increasingly severe, countries worldwide are accelerating their energy structure transformation and actively promoting the development of renewable energy. Wind power, as one of the most promising renewable energy sources, is widely used globally due to its abundant resources, low pollution, and low operating costs. According to statistics from the International Renewable Energy Agency (IRENA), by 2024, the total installed capacity of wind power globally had exceeded 1.2 terawatts, with wind power penetration rates exceeding 30% in some countries and regions. China's wind power grid connection scale continues to expand, and its penetration rate is rising year by year. However, the large-scale grid connection of wind power has brought severe challenges to the stable operation of the power system. Especially under conditions of high wind power penetration, wind power output exhibits strong volatility and low inertia characteristics, significantly weakening the frequency support capacity of the power system and thus increasing the difficulty of frequency stability control and regulation.

[0003] The volatility and intermittency of wind power are among the main characteristics of wind power systems. Wind power output is affected by wind speed variations, and wind turbines are typically connected to the grid via power electronic devices, resulting in a lack of rotational inertia found in traditional synchronous generators. Traditional thermal and hydroelectric generators, with their large rotational inertia, can respond quickly to system frequency fluctuations and maintain grid frequency stability. However, due to the low inertia of wind turbines, they cannot provide sufficient inertial support during load fluctuations or wind speed changes, leading to severe grid frequency fluctuations and prolonged frequency recovery times. Especially in power systems with high wind power penetration, wind turbines cannot respond to frequency fluctuations as quickly as traditional generators, potentially causing serious problems such as power system frequency collapse and system disconnection. Traditional frequency regulation methods, such as PI control and automatic generation control (AGC), have shown good application results in systems with low wind power penetration, but still face many limitations under conditions of high wind power penetration. PI control, as the most commonly used primary frequency regulation method, is simple in structure and easy to implement. However, in systems with high wind power penetration, due to the drastic fluctuations in wind power, it often exhibits problems such as insufficient frequency regulation accuracy and long response time. Although AGC systems can provide automatic power regulation across the entire network, they are limited by the control cycle and regulation speed when wind power is integrated at high proportions, making it difficult to track rapid wind power fluctuations in a timely manner, and frequency deviations easily exceed the allowable range. The lack of an effective coordination mechanism between wind power and energy storage systems results in low regulation accuracy, failing to meet the needs of dynamic frequency regulation. With the increasing proportion of wind power integration, existing traditional frequency regulation methods are gradually becoming unable to cope with problems such as large frequency fluctuation amplitudes and slow recovery speeds, urgently requiring a new frequency regulation control strategy to improve frequency stability and response capability under conditions of high wind power integration.

[0004] In recent years, wind-storage joint frequency regulation strategies have been widely studied and applied as an effective means to address frequency regulation issues under high wind power penetration. Wind-storage joint frequency regulation, through the coordinated control of wind farms and energy storage systems, leverages the advantages of energy storage devices—fast response speed and high regulation accuracy—to effectively compensate for the insufficient inertia of wind turbines, thereby significantly improving the accuracy and flexibility of system frequency regulation. When facing frequency fluctuations, energy storage systems can provide rapid power response, helping the grid quickly restore frequency and reducing the impact of wind power fluctuations on the grid. The joint dispatch of wind power and energy storage systems makes frequency regulation more precise and efficient, enabling rapid response to wind power output fluctuations and load changes, thus effectively suppressing the safety hazards caused by frequency fluctuations in traditional power systems. However, although wind-storage joint frequency regulation has improved the frequency stability of the power system to some extent, existing control strategies still face technical challenges such as the matching problem between wind power output fluctuations and energy storage system power regulation, insufficient primary frequency regulation response speed, and the challenges of real-time performance and computational complexity in MPC control methods. Among existing control strategies, Model Predictive Control (MPC), as a dynamic optimization method, relies on rolling optimization and feedforward prediction mechanisms to effectively improve frequency regulation accuracy, especially under multiple frequency disturbances and load fluctuation conditions, it exhibits strong robustness and adaptability.

[0005] However, traditional MPC methods are computationally complex and have fixed parameter settings, making them difficult to handle dynamic changes in power systems with high wind power penetration. Furthermore, the computational load of MPC methods increases exponentially with system scale, leading to insufficient real-time performance and efficiency. To further improve MPC control performance, intelligent optimization algorithms such as Particle Swarm Optimization (PSO) are introduced into the MPC framework. By optimizing key control parameters of the energy storage system (such as SOC upper and lower limits and frequency deviation constraints) online, the adaptive capability and real-time adjustment performance of the MPC control strategy are enhanced. While the combination of PSO and MPC can significantly improve frequency regulation accuracy, it still faces many challenges under complex operating conditions, such as how to more rationally coordinate the power distribution between the energy storage system and wind turbines, and how to achieve efficient and stable control strategies in multi-source uncertain environments. In existing research, despite the introduction of intelligent optimization algorithms, the matching problem between wind power and energy storage systems remains a technical bottleneck, and how to achieve precise adjustment in both dynamic and steady-state phases remains an urgent problem to be solved.

[0006] In summary, although the wind-storage joint frequency regulation strategy has a certain research foundation and application exploration, in power systems with a high proportion of wind power integration, the existing frequency regulation methods still have shortcomings in terms of frequency response speed, frequency regulation accuracy, and coordination and matching between energy storage and wind power systems. Summary of the Invention

[0007] To address the problems existing in current technologies, this invention proposes a primary frequency regulation method based on PSO-MPC joint control, aiming to improve the frequency regulation capability of power systems under high wind power penetration conditions through intelligent optimization and precise control. This strategy combines the PSO optimization algorithm and the MPC method to precisely adjust the power distribution between the energy storage system and the wind power system. Simultaneously, it introduces the ACE-ARR time-varying weight mechanism to ensure a balance between rapid response and steady-state accuracy under different disturbance conditions, providing a more efficient and reliable frequency regulation solution for power systems with high wind power penetration.

[0008] The technical solution adopted in this invention is: a primary frequency modulation method based on PSO-MPC joint control, comprising the following steps:

[0009] S1. Establish a wind-storage joint frequency regulation experimental environment and database: Build a primary frequency regulation system including a wind farm, energy storage device, and equivalent power grid in a simulation platform or actual measurement platform, and set up various wind power penetration rate and load disturbance scenarios; continuously collect wind power output power, energy storage power, battery SOC, voltage and current, total system load, frequency and frequency deviation, and organize them according to operating condition labels to form an experimental database for wind-storage joint primary frequency regulation; preprocess the data and divide the data into several typical operating condition sets;

[0010] S2. Extract indicators characterizing the operating status and lifespan of energy storage from the database as health factors on the energy storage side; extract characteristic factors on the wind power side based on the fluctuation characteristics of wind power output; simultaneously construct frequency regulation performance evaluation indicators (maximum frequency difference, recovery time, steady-state error, etc.) reflecting the primary frequency regulation effect; use correlation analysis to calculate the coupling strength between each candidate factor and the frequency regulation performance evaluation indicators, screen key factors, and construct a comprehensive health evaluation function:

[0011] ;

[0012] in: It refers to the state of charge of the energy storage battery. It is the change in SOC. It refers to deep discharge, indicating the range of each charge and discharge cycle. It refers to the number of charge-discharge cycles of the energy storage system. It is the temperature of the energy storage battery. These are weighting coefficients, representing the degree of influence of different indicators or health indicators on the comprehensive health evaluation function;

[0013] S3. Mapping of health status with available capacity and SOC upper and lower limits: Based on the overall health status, the available power capacity of energy storage, the safe upper and lower limits of SOC, and the maximum allowable ramp rate are adaptively set.

[0014] S4. Establish a discrete state-space model for wind and energy storage joint primary frequency regulation: Using system frequency deviation, wind power active power output, energy storage power and SOC as state variables, and wind turbine active power output adjustment and energy storage power command as control inputs, establish a discrete state-space equation describing the dynamics of primary frequency regulation of the power grid.

[0015] S5. Determine the control objectives and constraints, which includes the following sub-steps:

[0016] S5.1 The objective is to minimize the frequency deviation and optimize the power allocation between the energy storage system and the wind power system. The objective function is expressed as:

[0017] ;

[0018] in: It is the first Frequency deviation at time, It is the first Power output of the energy storage system at all times It is the first The state of charge of the energy storage system at all times. It is based on the SOC value. It is a weighting coefficient used to balance the frequency adjustment accuracy and the control of SOC;

[0019] S5.2 MPC Rolling Forecasting and Real-Time Optimization:

[0020] Rolling prediction: at each time step MPC performs future time domain operations Rolling forecasts; predicting the future based on the current system state. Wind power, energy storage power, and load power at each time step;

[0021] Wind power forecast representation:

[0022] ;

[0023] in, It is the first Wind power output at any given time This is the predicted wind speed;

[0024] Real-time optimization: MPC optimizes the allocation of wind power and energy storage power in real time to ensure that system frequency fluctuations are minimized and the SOC value of the energy storage system remains within a reasonable range; the power allocation is as follows:

[0025] ;

[0026] in, It is the output power of wind power. It is the load power. The output power of the energy storage system is positive when discharging to the grid and negative when charging. When wind power cannot meet the load, the energy storage provides the power.

[0027] S5.3 PSO algorithm dynamically adjusts key energy storage parameters:

[0028] Within the framework of MPC optimization, the Particle Swarm Optimization (PSO) algorithm is used to dynamically adjust the key control parameters of the energy storage system. The goal of PSO optimization is to adjust the upper and lower limits of the energy storage's State of Charge (SOC) and frequency deviation constraints. The fitness function of PSO optimization is expressed as:

[0029]

[0030] in: It's a frequency deviation. It is the power of the energy storage system. and These are the minimum and maximum SOC constraints for the energy storage system. These are weighting coefficients;

[0031] The S5.4 system restores frequency by adjusting energy storage and wind power. The restoration process is represented by the following formula:

[0032] Power imbalance:

[0033] Relationship between frequency deviation and power imbalance:

[0034] in, This refers to the instantaneous power imbalance of the system. For frequency deviation, The equivalent coefficient (which can be determined by system parameters or experimentally calibrated);

[0035] S6. Call the typical working condition data in the experimental database, perform simulation calculations on the parameter combination represented by each particle, and obtain a set of frequency response curves; calculate the fitness value according to the fitness function, and update the particle velocity and position according to the PSO mechanism, iterating until convergence; record the optimal parameters obtained by PSO offline optimization as the initial value of the controller, and solidify them in the wind-storage joint frequency regulation control platform as the initial parameters for actual operation.

[0036] S7. When the grid frequency deviates from the rated value by more than the preset threshold, primary frequency regulation is triggered. At the current sampling time, the system frequency, frequency deviation, wind power output, energy storage power and health factor information are collected in real time as the initial state of the MPC prediction model.

[0037] S8. Calculate the overall health of energy storage and wind power based on the latest operating data; adaptively adjust the allowable range of SOC, maximum power of energy storage, and weighting coefficients based on the health; and construct system output predictions for each moment in the prediction time domain by combining the latest system status, load forecast, and wind speed forecast information.

[0038] Within the prediction time domain, using the objective function in step S5 as the optimization objective and combining it with the updated parameters, a constrained linear / quadratic programming optimization problem is constructed.

[0039] S9. An optimization solver is used to solve the problem and obtain the incremental sequence of the active power output of the wind turbine and the sequence of the charging and discharging power of the energy storage in the predicted time domain, so as to minimize the objective function. According to the rolling optimization principle, only the first step control quantity in this sampling period is selected, and the setpoint of the active power output of the wind turbine and the command of the energy storage power are sent to the field control unit. The actual system runs according to this control quantity for one sampling period.

[0040] S10. Collect feedback and update status: When a new sampling time arrives, collect system frequency, wind power output, energy storage power and SOC data again, compare them with the expected values ​​of the previous prediction period, correct model deviation and update health factors; determine whether the system frequency has recovered to within the allowable deviation band and remained there for a certain period of time, and whether the frequency change rate is lower than the set threshold; if the conditions are not met, return to continue to execute rolling optimization and control; if the conditions are met, determine that one frequency regulation process has ended.

[0041] Furthermore, in step S1, the raw data in the experimental database is denoised, filtered, outliers removed, and time aligned; according to wind power penetration rate, load disturbance type, and disturbance amplitude, the data is divided into several typical operating condition sets, and further divided into modeling set, optimization set, and verification set.

[0042] Furthermore, in step S1, the wind power output... With wind speed Relatedly, the expression for wind power output is:

[0043]

[0044] in, It is air density. It is the sweeping area of ​​the wind turbine. It is the power factor. It's wind speed;

[0045] Energy storage power Related to the battery's SOC (State of Charge), specifically:

[0046]

[0047] Where η is the efficiency of the energy storage system. It is the charging and discharging power;

[0048] SOC update formula:

[0049]

[0050] in, It is the current moment. , It's the battery capacity. It is the time step.

[0051] Furthermore, in step S2, the energy storage side health factors include at least SOC and its fluctuation amplitude, charge / discharge depth, charge / discharge rate, cumulative cycle count, temperature, and internal resistance change; each health factor is normalized and dimensionally unified to form an energy storage health factor vector.

[0052] Characteristic factors on the wind power side are selected based on the fluctuation characteristics of wind power output, including the active power utilization rate of the wind turbine, the ramp rate, and the number of ramps.

[0053] Furthermore, in step S2, health factors are calculated through correlation analysis. Frequency deviation The degree of influence between them is as follows:

[0054]

[0055] in, It is covariance. That is the standard deviation.

[0056] Furthermore, in step S5, the PSO algorithm dynamically adjusts the energy storage control parameters by updating the position and velocity of the particles. The PSO algorithm updates the velocity and position of each particle using the following formula:

[0057]

[0058] in: It is the first The particle in the first Speed ​​at the next iteration It is the first The current position control parameters of each particle It is a particle The best historical position It is the globally optimal position. It is inertial weight. It is a learning factor. It is a random number.

[0059] Compared with existing wind-storage combined frequency regulation methods, the advantages of this invention are:

[0060] 1. Joint optimization using particle swarm optimization (PSO) and model predictive control (MPC): By optimizing the key parameters of the energy storage system through PSO and combining it with MPC for rolling prediction and optimized allocation of wind power and energy storage power, the response speed and accuracy of the wind-storage system to frequency disturbances can be effectively improved, thereby increasing the efficiency of primary frequency regulation.

[0061] 2. Flexibility and accuracy of control strategy: The constructed PSO-MPC joint control strategy can effectively coordinate the power distribution between the wind power system and the energy storage system, and make dynamic adjustments according to real-time frequency changes. This avoids the fixed setting of system parameters in traditional control methods, enhances the system's adaptability, and improves frequency regulation accuracy.

[0062] 3. High adaptability and ability to handle complex frequency variation scenarios: The PSO-MPC joint control strategy dynamically optimizes the control parameters of energy storage, ensuring that the system can flexibly respond to frequency changes under complex operating conditions with high wind power penetration and large load fluctuations, and has good engineering adaptability and operability.

[0063] Furthermore, this strategy enhances the frequency regulation capability of energy storage systems while reducing reliance on traditional thermal power units, providing a reliable and flexible frequency regulation solution for power systems with a high proportion of wind power integration. Compared to traditional methods, the control strategy of this invention is more efficient and flexible, improving the frequency stability of power systems under high wind power penetration conditions, and providing stronger technical support for the application of wind-storage integrated systems. Attached Figure Description

[0064] To make the objectives, technical solutions, and advantages of the present invention clearer, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, wherein:

[0065] Figure 1 This is the overall flowchart of the present invention;

[0066] Figure 2 This is a schematic diagram of the PSO-MPC joint control strategy of the present invention;

[0067] Figure 3 This is the MPC control prediction frequency diagram of the present invention;

[0068] Figure 4 This is a comparison chart of the effects of the single-frequency modulation PSO-MPC of the present invention;

[0069] Figure 5 This is a frequency deviation diagram after optimization of each algorithm in this invention;

[0070] Figure 6 This is a frequency deviation graph after optimization of each algorithm;

[0071] Figure 7 This is a comparison chart of the effects of frequency modulation PSO-MPC. Detailed Implementation

[0072] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0073] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0074] In view of this, the purpose of this invention is to provide a wind-storage joint primary frequency regulation control strategy based on PSO-MPC optimization. To achieve this objective, this invention provides the following technical solution, specifically including the following steps:

[0075] S1: Establish the experimental environment and database for wind-storage joint frequency regulation.

[0076] Build a primary frequency regulation system containing a wind farm, energy storage device and equivalent power grid in a simulation platform or test platform, and set up various wind power penetration rate and load disturbance scenarios.

[0077] Under various operating conditions, data such as wind power output, energy storage power, battery SOC, voltage and current, total system load, frequency and frequency deviation are continuously collected and organized according to operating condition labels to form an experimental database for wind-storage joint primary frequency regulation, providing basic data support for subsequent modeling and control strategy design.

[0078] S2: Data preprocessing and working condition division.

[0079] The raw data in the experimental database were denoised, filtered, outlier removed, and time aligned. The data were divided into several typical operating condition sets according to wind power penetration rate, load disturbance type (step, random fluctuation, etc.) and disturbance amplitude, and further divided into modeling set, optimization set and validation set.

[0080] S3: Extract health factors from the energy storage side.

[0081] Health factors are extracted from the database to characterize the operating status and lifespan of energy storage, including SOC and its fluctuation amplitude, depth of charge and discharge, charge and discharge rate, cumulative cycle count, temperature, and internal resistance change. Each health factor is normalized and dimensionally unified to form an energy storage health factor vector.

[0082] S4: Extract wind power-side characteristic factors and system performance factors.

[0083] Based on the fluctuation characteristics of wind power output, the active power output utilization rate, ramp rate, and number of ramps of the wind turbine are selected as characteristic factors on the wind power side. At the same time, system performance factors related to frequency regulation quality are extracted, such as maximum frequency deviation, frequency recovery time, overshoot, and steady-state error.

[0084] S5: Correlation analysis of health factors and frequency stability.

[0085] By using correlation coefficient analysis, sensitivity analysis, or grey relational analysis, the correlation between each health factor and the system frequency stability index is calculated, and key health factors that have a significant impact on frequency modulation performance are screened out. The effects of these factors on frequency drop, recovery speed, and steady-state accuracy are analyzed to provide a basis for subsequent weight design and constraint setting.

[0086] S6: Construct a comprehensive health evaluation index.

[0087] Based on the selected key health factors, a comprehensive health evaluation function is constructed, which maps the multidimensional health factors to health values ​​in the range of 0 to 1. When the health value is close to 1, it indicates that the equipment is in good condition and can participate in a larger amplitude frequency modulation. When the health value decreases, the range and intensity of its participation in frequency modulation are automatically tightened.

[0088] S7: Mapping of health status with available capacity and SOC upper and lower limits.

[0089] Based on the overall health status, the available power capacity, SOC safety upper and lower limits, and maximum allowable ramp rate of energy storage are adaptively set: the higher the health status, the wider the allowable SOC operating range and the greater the available power; the lower the health status, the narrower the SOC operating range and the more limited the charging and discharging power to protect the lifespan of energy storage.

[0090] S8: Establish a discrete state-space model for wind-storage combined primary frequency regulation.

[0091] Using system frequency deviation, wind power active power output, energy storage power and SOC as state variables, and wind turbine active power output adjustment and energy storage power command as control inputs, a discrete state-space equation describing the dynamics of primary frequency regulation of the power grid is established.

[0092] S9: Determine the control objectives and constraints.

[0093] Minimizing the frequency deviation and its rate of change is taken as the main control objective, while considering the smoothness of wind turbine and energy storage power changes, penalties for SOC deviation from the expected value, etc.; constraints such as wind turbine output and ramp rate, energy storage power and SOC upper and lower limits, and power balance constraints are added to the constraints.

[0094] S10: Construct the MPC objective function.

[0095] The objective function simultaneously considers the following weighted factors: frequency deviation, frequency change rate, energy storage power change, wind power output change, and the deviation of SOC from the reference value. Using the aforementioned health evaluation results, the weights of each factor are dynamically adjusted to reduce the depth of regulation when the energy storage health is low and to fully utilize its frequency regulation capability when the health is high.

[0096] S11: Determine the prediction time domain and the control time domain.

[0097] By combining the time scale and sampling period of the primary frequency regulation of the power system, and setting appropriate prediction time domain length Np and control time domain length Nc, the MPC can cover the main dynamic processes without causing excessive computational burden.

[0098] S12: The set of key parameters to be optimized in the design.

[0099] The upper and lower limits of energy storage SOC, the frequency deviation constraint threshold, the weight coefficients in the objective function, and the wind power / energy storage power allocation coefficient are taken as the key parameter set that needs to be optimized by PSO.

[0100] S13: Construct the PSO fitness function.

[0101] A multi-objective comprehensive fitness function is constructed using comprehensive frequency performance indicators (maximum frequency difference, recovery time, steady-state error), control smoothness (power ramp-up constraint), and energy storage lifetime loss (estimated based on SOC fluctuation and cycle count) as evaluation indicators.

[0102] The smaller the fitness value, the better the frequency modulation effect under this parameter combination.

[0103] S14: Initialize the particle swarm.

[0104] Set PSO parameters such as particle swarm size, maximum number of iterations, learning factor, and inertia weight; randomly generate a number of particles within a preset range, with each particle corresponding to a set of candidate parameters (SOC upper and lower limits, weight coefficients, etc.).

[0105] S15: Train PSO-MPC parameters offline using the experimental database.

[0106] By calling typical working condition data from the experimental database, simulation calculations are performed on the parameter combinations represented by each particle to obtain a set of frequency response curves. The fitness value is calculated according to the fitness function, and the particle velocity and position are updated according to the PSO mechanism. The iteration is continued until convergence, and a set of initial parameters with better performance under most working conditions is obtained.

[0107] S16: Obtain the initial optimal parameters of PSO-MPC and solidify them.

[0108] The optimal parameters obtained from PSO offline optimization are recorded as the initial values ​​of the controller and fixed in the wind-storage joint frequency regulation control platform, serving as the initial parameters and search center during actual operation.

[0109] S17: Frequency modulation event triggering and status sampling.

[0110] When the grid frequency deviates from the rated value by more than the preset threshold, primary frequency regulation is triggered; at the current sampling time, information such as system frequency, frequency deviation, wind power output, energy storage power, SOC, and health factors are collected in real time as the initial state of the MPC prediction model.

[0111] S18: Real-time health calculation and parameter update.

[0112] Using the health assessment methods established in S3 to S7, the comprehensive health of energy storage and wind power is calculated based on the latest operating data. Based on the health, the allowable range of SOC, the maximum power of energy storage, and the weighting coefficient are adaptively adjusted to provide the latest constraints for frequency regulation in this cycle.

[0113] S19: PSO online fine-tuning of key parameters (optional).

[0114] Based on the offline optimization results, and taking into account the current frequency deviation, health status, and recent operating performance, the PSO is used to fine-tune the upper and lower limits of SOC, frequency difference constraints, and some weights online within a small range, so that the control strategy is more in line with the current operating conditions.

[0115] S20: Construct a discrete prediction model for the current moment.

[0116] Based on the state-space model in S8, and combined with the latest system state, load forecast, and wind speed forecast information, prediction expressions for system output (frequency, power, SOC) at each time point in the prediction time domain are constructed.

[0117] S21: Establish the MPC optimization problem.

[0118] Within the prediction time domain, using the objective function in S10 as the optimization objective, and combining all constraints in S9 with the updated parameters in S18 and S19, a constrained linear / quadratic programming optimization problem is constructed.

[0119] S22: Call the optimization solver to solve the control sequence.

[0120] An optimization solver (such as Gurobi) is used to solve the above optimization problem to obtain the wind turbine active power output increment sequence and energy storage charging and discharging power sequence in the predicted time domain, so as to minimize the objective function.

[0121] S23: Issue the control quantity at the current sampling time.

[0122] Based on the rolling optimization principle, only the first step control quantity within the current sampling period is selected, and the active power output setpoint of the wind turbine and the energy storage power command are sent to the field control unit.

[0123] The actual system operates for one sampling cycle according to this control variable.

[0124] S24: Collect feedback and update status.

[0125] When a new sampling time arrives, data such as system frequency, wind power output, energy storage power, and SOC are collected again, compared with the expected values ​​of the previous prediction period, and the model bias is corrected and the health factor is updated.

[0126] S25: Frequency modulation end judgment and loop execution.

[0127] Determine whether the system frequency has recovered to within the allowable deviation band and remained there for a certain period of time, and whether the frequency change rate is lower than the set threshold;

[0128] If the conditions are not met, return to S18~S24 to continue rolling optimization and control; if the conditions are met, determine that one frequency modulation process has ended.

[0129] S26: Frequency modulation process data recording.

[0130] The frequency curve, control commands, SOC trajectory, and health status changes throughout the entire frequency modulation process are fully recorded and stored in the database, providing a data foundation for subsequent performance evaluation and algorithm improvement.

[0131] S27: Set various wind power penetration rates and typical disturbance conditions.

[0132] In simulation or actual systems, wind power penetration rates are set at different levels such as 25%, 40%, and 60%, and various load disturbance forms and amplitudes are superimposed to form a set of representative test conditions.

[0133] S28: Conduct comparative experiments with different control strategies.

[0134] For each test condition, the following methods were used:

[0135] ① Traditional PI / AGC frequency modulation strategy;

[0136] ② Only the wind-storage joint frequency regulation strategy using MPC is adopted;

[0137] ③ The PSO-MPC joint control strategy proposed in this invention;

[0138] If necessary, two configurations can be set up: "wind power does not participate in frequency regulation" and "wind and energy storage jointly regulate frequency," to compare the effects before and after wind power participates in frequency regulation.

[0139] S29: Conduct a frequency modulation comparison test.

[0140] Under various control strategies and configurations, the same frequency disturbance or load step signal is applied, and the frequency response curve, wind power and energy storage power change curve and SOC trajectory are recorded. Figures 3-7 This is a schematic diagram of the response under some typical operating conditions.

[0141] S30: Statistical evaluation of frequency regulation performance indicators.

[0142] For each experiment, the maximum frequency drop, frequency recovery time, steady-state frequency difference, overshoot, regulation rate, and energy storage charge / discharge depth are calculated to form a unified set of evaluation indicators; and frequency deviation curves under different strategies are plotted for comparison.

[0143] S31: Analyze the performance improvement of PSO-MPC compared to the comparison strategy.

[0144] Based on the statistical results of S30, the differences between traditional PI, simple MPC and PSO-MPC in various indicators are compared, and the improvement ratio of PSO-MPC in frequency drop suppression, recovery speed, steady-state accuracy and energy storage lifetime utilization is quantitatively given, thus verifying the superiority of the strategy of this invention.

[0145] S32: Optimize the control parameters based on the experimental results.

[0146] Based on the deficiencies found in the multi-condition experiments, the weights of the PSO fitness function, particle swarm parameters, and MPC objective function weights were modified; simulations and experiments for some conditions were then repeated under the new parameters until all performance parameters met or exceeded the requirements of power grid dispatching.

[0147] S33: Form the control strategy for engineering applications.

[0148] The optimized PSO-MPC control algorithm is solidified into an engineering application version and deployed to the wind-storage joint frequency regulation control system. Necessary protection logic and fault isolation strategies are set to ensure safe and reliable operation in the actual power grid environment.

[0149] S34: Long-term operation monitoring and data iterative updates.

[0150] During the engineering application phase, the frequency deviation, SOC utilization, and fault records of the system during long-term operation are continuously monitored and statistically analyzed. Newly generated data are continuously added to the experimental database, and the health model and necessary parameters are retrained regularly to maintain the adaptability of the strategy to new operating conditions.

[0151] S35: To form a complete primary frequency modulation method and technology system.

[0152] Through steps S1 to S34, a set of wind and storage joint primary frequency regulation methods is finally formed, which includes "data-driven health assessment + PSO offline / online optimization + MPC rolling control + multi-condition comparison verification".

[0153] This method can achieve coordinated frequency regulation of wind power and energy storage under conditions of high wind power penetration, significantly improve frequency regulation capability, reduce dependence on traditional thermal power units, and has good engineering promotion value.

[0154] A primary frequency modulation method based on PSO-MPC joint control includes the following steps:

[0155] S1. Establish the experimental environment for the wind-storage combined frequency regulation system, collect frequency regulation data under different wind power penetration rates and load fluctuations, and establish the system's experimental database, including wind power output power, energy storage system power, battery SOC, load data, and frequency fluctuation data.

[0156] S1.1 System Experimental Environment Setup:

[0157] The experimental environment setup involves establishing a wind-storage joint frequency regulation system encompassing a wind farm, energy storage system, and power grid on a simulation platform or actual testing platform. The core of the system is the real-time collection and monitoring of various key parameters, including wind power output, energy storage system power, battery SOC (state of charge), load data, and frequency fluctuation data. This process requires establishing an experimental database containing key variables such as wind power output, energy storage power, and battery SOC. To effectively collect and process data, the relationships between relevant parameters can be expressed using the following formula:

[0158]

[0159] Wind power output With wind speed Related. The commonly used expression for wind power is:

[0160]

[0161] in, It is air density. It is the sweeping area of ​​the wind turbine. It is the power factor. It's wind speed.

[0162] Energy storage power Related to the battery's SOC (State of Charge). The power of the energy storage system can be updated using the following relationship:

[0163]

[0164] Where η is the efficiency of the energy storage system. It refers to the charging and discharging power.

[0165] SOC update formula:

[0166]

[0167] in, It is the current moment. , It's the battery capacity. It is the time step.

[0168] S1.2 Establishment of the experimental database:

[0169] To study frequency regulation under different wind power penetration rates and load fluctuations, the experimental database needs to collect the following data: wind power output. Energy storage capacity Load power System frequency deviation .

[0170] Data acquisition methods can be based on the measurement or simulation platform data output of a real-time system. An experimental database is constructed by recording relevant data under different wind power penetration rates and load disturbances (such as step loads, random fluctuations, etc.).

[0171] These data will serve as the basis for subsequent steps, such as optimization and control strategy design.

[0172] S1.3 Experimental Data Collection and Processing:

[0173] At each wind power penetration rate, different load disturbances (such as step loads and random fluctuations) are set, and the corresponding frequency deviation, energy storage power and wind power output power are recorded.

[0174] Data is collected and updated regularly through experimental or simulation systems to form a database that includes time series data.

[0175] Data preprocessing, including noise reduction, filtering, and time alignment, is performed to ensure data accuracy and consistency.

[0176] S1.4 Data Storage:

[0177] A database is used to store the results of each experiment. Each record includes time, wind power, energy storage power, SOC, load power, and frequency deviation. Each set of experimental data is labeled according to different wind power penetration rates and load fluctuation types, and stored in categories for subsequent analysis.

[0178] S2. The relationship between the calculated health factors and the frequency stability of the wind-storage combined system is used to adjust the upper and lower limits of the SOC and frequency deviation constraints of the energy storage system online, so as to ensure that the energy storage system can respond quickly and accurately to frequency changes, avoid overcharging and discharging, and extend the service life of the system.

[0179] S2.1 Definition and Calculation of Health Factors:

[0180] Health factors are used to quantify the state of an energy storage system and assess its suitability for participation in frequency regulation. Health factors are closely related to the performance of the energy storage system (such as State of Charge (SOC), Depth of Charge / Discharge, and Cycle Count). The calculation can be based on the following key indicators:

[0181] ;

[0182] in: It refers to the state of charge of the energy storage battery. It is the change in SOC. It refers to the depth of discharge, indicating the range of each charge and discharge cycle. It refers to the number of charge-discharge cycles of the energy storage system. It is the temperature of the energy storage battery. It is a weighting coefficient, which represents the degree of influence of different indicators on health factors.

[0183] S2.2 Changes in SOC: Changes in SOC can be calculated using the following formula:

[0184]

[0185] in, It is the total capacity of the battery. It is energy storage capacity. It is the time step.

[0186] S2.3 Relationship between health factors and frequency stability: The relationship between health factors and frequency stability can be obtained through correlation analysis. Frequency bias Health status of energy storage systems The relationship between them can be represented as:

[0187]

[0188] in, It's a frequency deviation. It is the health factor of the energy storage system at the current moment. It is the power output of the energy storage system. It is wind power. It is the load power.

[0189] S2.4 Calculate health factors through correlation analysis. Frequency deviation The degree of influence between them can be determined using the correlation coefficient. Or sensitivity analysis method:

[0190]

[0191] in, It is covariance. That is the standard deviation.

[0192] S3. The frequency regulation process is divided into a power allocation phase for the wind power system and the energy storage system, with each system being scheduled and optimized separately. In this phase, Model Predictive Control (MPC) performs rolling predictions and optimized allocation of power between the wind power and energy storage systems, enabling the system to achieve a rapid and coordinated response when frequency disturbances occur. The SOC state and frequency deviation of the energy storage system are used as the main optimization variables, and the PSO algorithm is used to dynamically tune the key operating parameters of the energy storage system, thereby improving the accuracy of frequency regulation and accelerating frequency recovery.

[0193] S3.1 Power Allocation and Optimization Objectives:

[0194] During frequency regulation, the system is divided into power scheduling phases for the wind power system and the energy storage system, which are scheduled and optimized separately. To ensure a fast and coordinated frequency response when disturbances occur, model predictive control (MPC) is used for rolling forecasting and power allocation optimization.

[0195] The objective function is to minimize the frequency deviation and optimize the power allocation between the energy storage system and the wind power system. The objective function of MPC can be expressed as:

[0196]

[0197] in: It is the first Frequency deviation at time, It is the first Power output of the energy storage system at all times It is the first The state of charge of the energy storage system at all times. It is based on the SOC value. It is a weighting coefficient used to balance the frequency adjustment accuracy and the control of the SOC.

[0198] The goal is to optimize the SOC control of the energy storage system and the power output of the wind power system, so as to minimize the system frequency deviation and avoid overcharging and discharging of the energy storage system.

[0199] S3.2 MPC Rolling Forecasting and Optimization:

[0200] Rolling prediction: at each time step MPC performs future time domain operations Rolling forecasts. Based on the current system state, predict the future. The wind power, energy storage power, and load power at each time step.

[0201] Wind power forecasting can be expressed by the following formula:

[0202]

[0203] in, It is the first Wind power output at any given time That is the predicted wind speed.

[0204] Real-time optimization: MPC optimizes the allocation of wind power and energy storage power in real time, ensuring that system frequency fluctuations are minimized and the SOC value of the energy storage system remains within a reasonable range. The power allocation formula can be expressed as:

[0205]

[0206] in, It is backup power, provided by energy storage when wind power cannot meet the load.

[0207] S3.3 PSO algorithm dynamically adjusts key energy storage parameters:

[0208] Within the framework of MPC optimization, the Particle Swarm Optimization (PSO) algorithm is used to dynamically adjust the key control parameters of the energy storage system. The goal of PSO optimization is to enable the system to respond quickly to frequency fluctuations by adjusting the upper and lower limits of the energy storage's State of Charge (SOC) and frequency deviation constraints.

[0209] The fitness function for PSO optimization can be expressed as:

[0210]

[0211] in: It's a frequency deviation. It is the power of the energy storage system. and These are the minimum and maximum SOC constraints for the energy storage system. It is the weighting coefficient.

[0212] The PSO algorithm dynamically adjusts the energy storage control parameters by updating the position and velocity of particles, ensuring the system's rapid recovery and stability under frequency disturbances.

[0213] PSO Update Formula: The PSO algorithm updates the velocity and position of each particle using the following formula:

[0214]

[0215] in: It is the first The particle in the first Speed ​​during the next iteration;

[0216] It is the first The current position (control parameter) of each particle;

[0217] It is a particle The historical best position;

[0218] It is the globally optimal position;

[0219] It is inertial weight. It is a learning factor. It is a random number.

[0220] S3.4 Energy Storage Power Regulation: Based on PSO optimization, MPC adjusts the energy storage power output in real time. This ensures that the energy storage system can respond quickly when frequency disturbances occur.

[0221] S3.5 SOC Update and Optimization: The PSO also dynamically adjusts the upper and lower limits of the energy storage system's SOC. , Avoid overcharging and discharging of the energy storage system, while improving the frequency recovery speed.

[0222] S3.6 Frequency Recovery Process: The system recovers the frequency by adjusting the energy storage and wind power. The recovery process is represented by the following formula:

[0223]

[0224] in, It's a frequency deviation. and These are the power outputs of the wind power and energy storage systems, respectively. It is the load power.

[0225] S4. Through experimental verification and control effects under various wind power penetration rates, the advantages of the PSO-MPC joint control method are established. By comparing with the traditional PI control method, the significant advantages of this method in reducing frequency drop, accelerating frequency recovery speed, and improving steady-state accuracy are verified, ensuring that the system can operate stably under high wind power penetration conditions.

[0226] S4.1 Experimental Design and Setting of Wind Power Penetration Rate:

[0227] In this experiment, we considered the impact of different wind power penetration rates and load fluctuations (step load, random fluctuations, etc.) on the system's frequency regulation performance. These operating conditions were set to test the performance of the control method under different conditions.

[0228] Experimental variables: Wind power penetration rate: set the proportion of wind power in the total system power;

[0229]

[0230] in, It is the first Wind power output at any given time It is the total power of the system. It refers to wind power penetration rate.

[0231] Load fluctuations: Load fluctuations can be simulated using step loads or random load changes, and the load power... It can be represented as:

[0232]

[0233] in, It is the base load. It is a disturbance in the load (step or random fluctuation).

[0234] S4.2 PSO-MPC Joint Control: In the PSO-MPC joint control method, we use the PSO algorithm to dynamically optimize the upper and lower limits of the SOC and frequency deviation constraints of the energy storage system, and use MPC for rolling prediction and power allocation optimization. The goal is to improve the accuracy of frequency regulation by minimizing frequency deviation, frequency recovery time, and SOC fluctuations.

[0235] PSO-MPC objective function:

[0236]

[0237] in: It's a frequency deviation. It is energy storage capacity. It is the SOC of the energy storage system. It is based on the SOC value. It is the weighting coefficient.

[0238] S4.3 PI Control Method: The traditional PI control method involves setting the proportional gain. and integral gain The wind power and energy storage capacity are adjusted based on the frequency deviation. The control formula is:

[0239]

[0240] in, It's a frequency deviation. and These are proportional and integral gains, respectively.

[0241] S4.4 Frequency Decline: The experiment assesses frequency decline by recording the system's frequency deviation under different wind power penetration rates and load fluctuations. It can be represented as:

[0242]

[0243] S4.5 Frequency Recovery Time: The time required for the system to recover to the permissible frequency range. It can be represented as:

[0244]

[0245] in, It is the allowable frequency deviation threshold, which is usually 0.05 Hz.

[0246] S4.6 Steady-state accuracy: The error after the frequency recovers to steady state, usually expressed as steady-state frequency deviation. To indicate:

[0247]

[0248] The higher the steady-state accuracy, the closer the system frequency is to the target frequency.

[0249] S5. Deploy the PSO-MPC joint control strategy in the actual wind-storage joint frequency regulation system, conduct real-time testing and parameter tuning under various wind speed fluctuations and load changes, verify the frequency regulation performance and applicability of the strategy in the engineering environment, and thus provide stable and reliable frequency regulation technical support for high-proportion wind power access.

[0250] Example 1

[0251] To fully verify the effectiveness and advantages of the wind-storage joint frequency regulation control strategy based on PSO-MPC optimization proposed in this invention, a series of experimental scenarios were designed, covering various operating conditions such as different wind power penetration rates, load fluctuations, and frequency disturbances. A high-fidelity simulation model was used to comprehensively test the wind-storage joint frequency regulation system, aiming to evaluate its dynamic performance and steady-state behavior under various operating environments. The experiments simulated the operation of an actual power system, focusing on analyzing the performance of the wind-storage joint frequency regulation strategy in terms of frequency response capability, system stability, recovery speed, and frequency regulation accuracy, providing technical support and performance verification for practical applications.

[0252] S1. Setting the experimental conditions for the wind-storage integrated frequency regulation system. First, a wind power penetration rate of 25% was selected as the baseline operating condition, and load disturbance experiments were conducted under this condition. The experimental conditions included various frequency disturbances to simulate fluctuations and instability in actual power systems. By setting different load fluctuation and wind speed variation scenarios, the system was ensured to operate stably under various complex conditions. Furthermore, the experimental environment included subsystems such as the wind power system, energy storage system, and power grid to ensure the comprehensiveness and accuracy of the test, and to provide rich data support for the subsequent verification of control strategies.

[0253] S2. During the experiment, Model Predictive Control (MPC) algorithm was used to perform real-time frequency regulation on the wind-storage combined system. Specifically, MPC dynamically adjusts the power output of the wind power and energy storage systems through a real-time rolling predictive control strategy to cope with load surges and frequency fluctuations. Experimental results show that after a load step disturbance, MPC can quickly adjust the power output of the energy storage system, significantly reduce frequency fluctuations, and the recovery speed is significantly faster than traditional control methods. Figure 3 The comparison between the predicted frequency and the actual frequency by MPC control was demonstrated, verifying the accurate prediction and following capabilities of MPC in frequency regulation, especially its advantages in frequency drop and recovery speed.

[0254] S3. Further optimization of key control parameters of the energy storage system, such as the upper and lower limits of SOC and frequency deviation, is achieved through the Particle Swarm Optimization (PSO) algorithm. The PSO algorithm can adjust control parameters online to cope with the challenges brought by frequency changes and wind power fluctuations. Experimental data show that the MPC control strategy optimized by PSO can significantly improve the frequency regulation accuracy of the system, especially under conditions of large frequency fluctuations, it can reduce SOC fluctuations, optimize the charging and discharging strategy of the energy storage system, and improve the overall frequency regulation efficiency. Figure 6 The effects of different optimization algorithms (such as simulated annealing (SA), differential evolution (DE), and PSO) on frequency deviation are demonstrated. PSO optimization is more effective than other methods in reducing frequency fluctuations and improving system stability.

[0255] S4. In the comparative experiment, under the condition of 25% wind power penetration, experiments were conducted using two configurations: "wind power without frequency regulation" and "wind-storage combined frequency regulation". Experimental data show that the wind-storage combined frequency regulation system using the MPC control strategy exhibits significant advantages in frequency recovery speed and steady-state accuracy. Figure 4The diagram illustrates a comparison of frequency response under these two configurations. As shown in the figure, the system employing MPC control can accurately track frequency changes and rapidly adjust the power output of the wind power and energy storage systems, thereby significantly reducing frequency fluctuations and accelerating recovery. Compared to traditional control methods, MPC control exhibits higher accuracy and response speed in frequency drop, recovery slope, and steady-state phases, ensuring the system can quickly recover to the rated frequency after sudden load changes. This fully demonstrates the powerful predictive and precise adjustment capabilities of the MPC control strategy.

[0256] S5. Finally, through comprehensive analysis of the experimental results, the superiority of the PSO-MPC and ACE-ARR integrated control strategy proposed in this invention was verified. Under different disturbance and operating conditions, this strategy can improve the frequency regulation capability of the wind-storage combined system in both dynamic and steady-state phases. Especially under conditions of high wind power penetration, it can effectively reduce dependence on traditional thermal power units and improve the stability and economy of the system. Figure 7 The comparison of frequency stability under different control strategies was demonstrated, showing that PSO-MPC optimized control can significantly improve the frequency recovery speed and steady-state accuracy of the wind-storage integrated system.

[0257] Those skilled in the art will understand that the present invention can be implemented as a method, system, or computer program product. Therefore, the present invention can be implemented in hardware, software, or a combination of both. Furthermore, the present invention can also be implemented as a computer program product by storing computer-usable program code on a computer-usable storage medium (including but not limited to disk storage, optical storage, etc.).

[0258] It should be noted that the above embodiments are merely examples illustrating the technical solutions of the present invention and do not constitute a limitation thereof. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified and equivalently substituted in various ways, provided that such modifications or substitutions do not depart from the technical spirit and scope of the present invention and should be covered within the scope of the claims of the present invention.

Claims

1. A method for primary frequency modulation based on PSO-MPC joint control, characterized in that, Comprising the following steps: S1, Establishing the experimental environment and database of wind storage combined primary frequency modulation: build a primary frequency modulation system containing wind farm, energy storage device and equivalent power grid in the simulation platform or actual measurement platform, set multiple wind power penetration and load disturbance scenarios; continuously collect wind power output, energy storage power, battery voltage and current, system total load, frequency and frequency deviation, and organize them according to the working condition label to form the experimental database of wind storage combined primary frequency modulation; preprocess the data and divide them into several typical working condition sets; S2, Establishing the wind storage combined primary frequency modulation model: build the wind turbine model, energy storage device model and equivalent power grid model, and establish the wind storage combined primary frequency modulation model; S2, extracting the indicators representing the state of energy storage operation and life from the database as the health factors of the energy storage side; based on the fluctuation characteristics of wind power output, extracting the characteristic factors of the wind power side; At the same time, the frequency modulation performance evaluation index reflecting the primary frequency modulation effect is constructed: maximum frequency difference, recovery time, steady-state error; the coupling strength between each candidate factor and the frequency modulation performance evaluation index is calculated by correlation analysis, the key factors are screened, and the comprehensive health degree evaluation function is constructed: ； Wherein: is the state of charge of the energy storage battery, is is the change amount of is the deep discharge, indicating the range of each charge and discharge; is the number of charge and discharge cycles of the energy storage system, is the temperature of the energy storage battery; is the weight coefficient, the weight coefficient is the influence degree of different indexes on the comprehensive health degree evaluation function, and each index has been homogenized and normalized. S3, the comprehensive health degree and the available capacity, mapping of upper and lower limits: according to the comprehensive health degree, the available power capacity of the energy storage, adaptive setting of the safety upper and lower limits and the maximum allowed ramp rate; S4, establish discrete state space model of wind storage combined primary frequency modulation: take system frequency deviation, wind power active output, energy storage power and As state variables, the wind turbine active power adjustment amount and the energy storage power instruction are taken as control inputs to establish the discrete state space equation describing the dynamic of the grid primary frequency modulation. S5, determine the control target and constraint condition, including the following sub-steps: S5.1 The target is to minimize the frequency deviation and optimize the power distribution of the energy storage system and the wind power system, and the objective function is represented as: ； wherein: is the frequency deviation at the is the frequency deviation at the is the frequency deviation at the is the energy storage output power at the is the energy storage output power at the is the state of charge of the energy storage system at the is the reference SOC value, is the weight coefficient used to balance the frequency adjustment accuracy and the control of S5.2 MPC rolling prediction and real-time optimization: Rolling prediction: at each time step , the MPC performs a rolling prediction of the future time horizon ; based on the current system state, the wind power, the storage power and the load power are predicted for the next time steps; The wind power prediction is represented as: ； wherein, is the first wind power output at the time instant t, is the predicted wind speed; Real-time optimization: MPC optimizes the allocation of wind power and energy storage power in real time, ensuring that system frequency fluctuations are minimized and the energy storage system... The value remains within a reasonable range; power allocation is as follows: ； wherein, is the wind power output, is the load power, is the energy storage system output power, positive for discharging to the grid and negative for charging; is the reserved standby power for meeting the primary frequency regulation margin requirement; S5.3 PSO algorithm dynamically adjusts key parameters of energy storage: within the framework of MPC optimization, the particle swarm optimization (PSO) algorithm is used to dynamically adjust the key control parameters of the energy storage system; the goal of PSO optimization is to minimize the deviation of the frequency of the power grid from the target value by adjusting the upper and lower limits of the energy storage and the frequency deviation constraints, and the fitness function of the PSO optimization is expressed as: ； wherein: is the frequency deviation, is the energy storage output power, positive for discharging, negative for charging; and are the minimum and maximum constraints of the energy storage system, respectively, is the weight coefficient; S5.4 The system restores the frequency by adjusting the energy storage and wind power, and the recovery process is represented by the following formula: Power imbalance: ; Frequency deviation and power imbalance relationship: ; wherein is the system instantaneous power imbalance, is the frequency deviation, is the equivalent coefficient; S6, call the typical working condition data in the experimental database, simulate and calculate each particle representing the parameter combination to obtain a group of frequency response curves; calculate the fitness value according to the fitness function, and update the particle speed and position according to the PSO mechanism, and iterate to convergence; record the optimal parameters obtained by PSO offline optimization as the initial value of the controller, and solidify them in the wind storage combined frequency modulation control platform as the initial parameters in actual operation; S7, when the grid frequency deviates from the rated value by more than the preset threshold, the primary frequency modulation is triggered; at the current sampling time, the system frequency, frequency deviation, wind power output, energy storage power and health factor information are collected in real time as the initial state of the MPC prediction model; S8, calculate the comprehensive health degree of energy storage and wind power according to the latest operation data; according to the health degree, adjust the weight coefficient of the energy storage and the wind power The allowed range, the maximum power of the energy storage, and the weight coefficient are adaptively adjusted; combined with the latest system state, load prediction, and wind speed prediction information, the system output prediction at each time in the prediction time domain is constructed. In the prediction time domain, taking the objective function in step S5 as the optimization objective, combining the updated parameters, a linear / quadratic programming optimization problem with constraints is constructed; S9, use the optimization solver to solve and obtain the wind turbine active power output increment sequence and energy storage charging and discharging power sequence in the prediction time domain, so that the objective function is minimized; according to the rolling optimization principle, only the first step control quantity in the current sampling period is selected, and the wind turbine active power output given value and energy storage power instruction are issued to the field control unit; the actual system runs for one sampling period according to the first step control quantity. S10, collecting feedback and updating state: when the new sampling time comes, the system frequency, wind power output, energy storage power and data are collected again, the model deviation is corrected and the health factor is updated by comparing with the expected value of the previous prediction period; it is judged whether the system frequency is restored to the allowable deviation band and maintained for a certain time, and the frequency change rate is lower than the set threshold; if the conditions are not met, return to continue rolling optimization and control; if the conditions are met, it is determined that the primary frequency modulation process is over. 2.The primary frequency modulation method based on PSO-MPC joint control according to claim 1, wherein: In step S1, the original data in the experimental database is denoised, filtered, and abnormal points and time alignment are processed; according to the wind power penetration, load disturbance type and disturbance amplitude, the data is divided into several typical working condition sets, and further divided into modeling set, optimization set and verification set. 3.The primary frequency modulation method based on PSO-MPC joint control according to claim 2, characterized in that: In step S1, the wind power output is related to the wind speed and the wind power output expression is: ； wherein, is the air density, is the swept area of the wind turbine, is the power coefficient, is the wind speed; Energy storage power With the state of charge of the battery Specifically, the state of charge of the battery ； wherein η is the efficiency of the energy storage system, is the charge and discharge power; Update Formula: ； wherein, is the current time step, , is the battery capacity, is the time step.

4. The primary frequency modulation method based on PSO-MPC combined control according to claim 3, characterized in that: In step S2, the energy storage side health factors at least include and the fluctuation amplitude, the charge and discharge depth, the charge and discharge rate, the cumulative cycle number, the temperature, the internal resistance change; each health factor is normalized and dimensionally unified to form an energy storage health factor vector; the wind power side characteristic factors at least include the wind turbine active power output utilization rate, the climbing rate, and the climbing number.

5. The method of claim 4, wherein the method is based on a PSO-MPC combined control method. In step S2, the comprehensive health degree evaluation function is calculated through correlation analysis The influence degree between the frequency deviation and the frequency deviation, specifically: ； wherein, is the covariance, is the standard deviation. 6.The primary frequency modulation method based on PSO-MPC joint control according to claim 5, characterized in that: In step S5, the PSO algorithm dynamically adjusts the energy storage control parameters by updating the position and speed of the particles, and the PSO algorithm updates the speed and position of each particle by the following formula: ； in: It is the first The particle in the first Speed ​​at the next iteration It is the first The current position control parameters of each particle It is a particle The best historical position It is the globally optimal position. It is inertial weight. It is a learning factor. It is a random number.

Images