Networking type energy storage capacity optimal matching method based on frequency response model

Abstract

The invention discloses a network construction type energy storage capacity optimal matching method based on a frequency response model, and belongs to the technical field of gas turbine power plant financial supervision and artificial intelligence, and the method comprises the steps: S1, quantifying the power vacancy of a typical station under an extreme short-circuit fault, determining a virtual inertia constant and a virtual droop coefficient of network construction type energy storage, and carrying out the calculation of the virtual inertia constant and the virtual droop coefficient; setting an initial output coefficient and adjusting a step length; s2, constructing and integrating a system frequency response model based on the inertia and frequency modulation characteristics of the conventional unit, the wind turbine generator and the network construction type energy storage; s3, applying power disturbance to the system, gradually increasing an initial output coefficient and performing simulation, and recording the maximum frequency deviation, the maximum frequency change rate and the quasi-steady-state frequency deviation; and S4, when the maximum frequency deviation, the maximum frequency change rate and the quasi-steady-state frequency deviation are respectively smaller than or equal to set threshold values, the energy storage capacity proportion corresponding to the current initial output coefficient is the optimal proportion. According to the method, the problem of accurately measuring the optimal configuration proportion of the network construction type energy storage in the new energy station is solved.

Description

Technical Field

[0001] This invention relates to the fields of financial supervision and artificial intelligence technology for gas turbine power plants, specifically to a method for optimal allocation of grid-type energy storage capacity based on a frequency response model. Background Technology

[0002] With the large-scale integration of new energy sources, power electronic converters such as wind power and photovoltaic power are gradually replacing traditional synchronous machines, which significantly reduces the rotational inertia of the power grid, weakens its self-stabilization capability, and makes its frequency and voltage regulation performance insufficient. At the same time, the intermittency and volatility of new energy output exacerbate power disturbances, further threatening the stability of system operation.

[0003] Currently, extensive research has been conducted both domestically and internationally on capacity configuration and control strategies for energy storage to support grid frequency and voltage. Some scholars combine active power-frequency droop with virtual synchronous machine (VSG) control, simultaneously optimizing energy storage capacity and droop coefficient through multi-objective particle swarm optimization and ideal solution sorting. Others introduce adaptive gain regulation into the VSG inertial link, dynamically tracking frequency fluctuations through an energy manager to meet the three major indicators of transient overshoot, steady-state deviation, and recovery time. Still other studies have proposed virtual inertia and virtual damping limiting strategies based on the system rate of change of frequency (RoCoF) threshold, or prevent transient active power output by limiting the voltage. To prevent inverter overload, other scholars have proposed a rapid reactive power injection model based on the voltage-response curve from the perspective of reactive power support, in order to improve voltage transient drops, and combined it with the energy storage fault ride-through capability index to evaluate the system's anti-disturbance performance under short-circuit faults; other studies have taken the short-circuit ratio of the power station as a key factor to analyze the impact of grid-type energy storage on the short-term and medium-to-long-term stability of the power grid under different short-circuit ratios, providing a multi-dimensional evaluation basis for capacity allocation; in addition, clustering and K-means algorithms and distributed primal dual subgradient methods are used to achieve a decentralized solution for minimizing the cost of capacity allocation on multi-scenario sequences considering spatiotemporal uncertainties.

[0004] In summary, while existing research focuses on improving power supply reliability and reducing operating costs, it has not yet systematically characterized the response characteristics under different capacity ratios based on the dynamic interaction mechanism between energy storage and the power grid. Further research is needed on the proportion of grid-type energy storage in new energy power plants. Summary of the Invention

[0005] In view of the above-mentioned problems, the present invention is proposed.

[0006] Therefore, the technical problem solved by this invention is:

[0007] 1. Quantify the power deficit and corresponding support requirements of typical power plants under extreme short-circuit faults;

[0008] 2. Based on the inertia and frequency regulation characteristics of conventional generating units, wind turbines, and grid-connected energy storage, a system frequency response model is constructed and integrated;

[0009] 3. Record and constrain the maximum frequency deviation Δf through parameter scanning and simulation. max Maximum frequency change rate RoCof max Quasi-steady-state frequency deviation Δf ss ;

[0010] 4. Under the premise of satisfying the above frequency stability indicators, the optimal grid-type energy storage output coefficient and capacity ratio are determined iteratively.

[0011] To address the aforementioned technical problems, this invention provides the following technical solution: a method for optimal allocation of grid-type energy storage capacity based on a frequency response model, comprising the following steps:

[0012] S1. Quantify the power deficit of typical stations under extreme short-circuit faults, determine the virtual inertia constant and virtual droop coefficient of grid-type energy storage, and set the initial output coefficient and adjustment step size.

[0013] S2. Based on the inertia and frequency regulation characteristics of conventional units, wind turbines and grid-type energy storage, construct and integrate the system frequency response model;

[0014] S3. Apply a power disturbance to the system, gradually increase the initial output coefficient and simulate, and record the maximum frequency deviation, the maximum frequency change rate and the quasi-steady-state frequency deviation.

[0015] S4. When the maximum frequency deviation, the maximum frequency change rate, and the quasi-steady-state frequency deviation are all less than or equal to the predetermined thresholds, the energy storage capacity ratio corresponding to the current initial output coefficient is the optimal ratio.

[0016] As a preferred embodiment of the optimal allocation method for grid-type energy storage capacity based on a frequency response model as described in this invention, wherein: S1 specifically includes:

[0017] Fault scenario modeling is performed based on historical fault recording data of the power grid, extracting three-phase short-circuit and single-phase ground fault waveforms of typical power plants; instantaneous power theory is used to decompose the orthogonal components of the fault current and calculate the active power surge value; and the impact of renewable energy disconnection is corrected.

[0018] Considering the low-voltage ride-through failure rate of wind turbines, when the voltage drop exceeds a predetermined proportion, the probability of wind turbines disconnecting from the grid is calculated, and the power deficit is added:

[0019]

[0020] In the formula, ΔP total The total power deficit, i.e., the total system power loss caused by wind turbine disconnection, is usually measured in watts; ΔPfault The basic fault power deficit refers to the power loss caused by fault factors other than wind turbine disconnection, including conventional generator failures and transmission line faults; N WT P represents the total number of wind turbine units in a wind farm or related area; WT,k I represents the rated power or actual output power of the k-th wind turbine, reflecting the power that each wind turbine can provide during normal operation; trip This is the offline indicator function, which is a binary variable.

[0021] As a preferred embodiment of the optimal allocation method for grid-type energy storage capacity based on a frequency response model as described in this invention, wherein: S1 specifically includes:

[0022] The key parameters of the grid-type energy storage are set as virtual inertia constants, and the design is reversed based on the system inertia requirements; a dynamic adaptive mechanism is used to deploy the J function in the dSPACE controller to improve the inertia support strength during high-frequency disturbances;

[0023] The virtual droop coefficient is designed based on the principle of equivalence of adjustment rate; the segmented droop strategy adopts flexible droop when the frequency deviation is less than the predetermined threshold, and switches to rigid droop when the frequency deviation is greater than the predetermined threshold.

[0024] Initial output coefficient, based on power deficit ratio initialization, corrected for renewable energy penetration rate:

[0025] If the proportion of wind power installed capacity is greater than the predetermined proportion, the initial output coefficient increases; if the system includes pumped storage units, the initial output coefficient decreases.

[0026] Adjusting the step size, for high-proportion renewable energy grids, the binary search method and golden ratio are used to accelerate the search; for weak grids, Lyapunov stability verification is added; multi-energy storage collaboration is achieved through pre-configuration of damping controllers.

[0027] As a preferred embodiment of the optimal allocation method for grid-type energy storage capacity based on a frequency response model as described in this invention, wherein: S2 specifically includes:

[0028] In the conventional unit model, the synchronous generator rotor motion equation is established to quantify the suppressive effect of rotational inertia and damping coefficient on the system frequency change rate; in the governor system model, a first-order lag model of valve opening-power response is embedded to simulate the delay characteristics of primary frequency regulation.

[0029] The wind turbine model is reconstructed with low inertia characteristics. For the doubly fed induction generator (DFIG) turbine, only the equivalent inertia of the converter capacitor is retained, and the frequency coupling effect is described by the dynamic equation of the phase-locked loop. For the direct-drive turbine, a virtual inertia control loop is constructed. Frequency regulation capability is limited, considering maximum power point tracking and rotor speed protection, and an upper limit of active power reserve is set.

[0030] The grid-type energy storage model adopts a virtual synchronous machine algorithm and multi-timescale control, which is divided into millisecond-level virtual inertia response suppression RoCoF; second-level droop control to compensate for steady-state frequency deviation; and minute-level state of charge coordination to prevent over-discharge.

[0031] As a preferred embodiment of the optimal allocation method for grid-type energy storage capacity based on a frequency response model as described in this invention, wherein: S2 specifically includes:

[0032] System-level model integration and a unified state-space framework combine the state equations of conventional units, wind power, and energy storage to construct a multi-dimensional state matrix; frequency-power dynamic coupling links the power output of each source with the grid frequency through the node admittance matrix.

[0033] Enhanced high-penetration scenarios and optimized damping ratios are achieved by introducing virtual impedance control in eigenvalue analysis to suppress subsynchronous oscillations caused by wind power grid connection. Communication delay compensation is also implemented by correcting control command transmission errors based on a 5G URLLC time delay prediction model.

[0034] As a preferred embodiment of the optimal allocation method for grid-type energy storage capacity based on a frequency response model as described in this invention, wherein: S2 specifically includes:

[0035] The digital twin verification platform imports the actual power grid topology into the RTDS real-time simulator and compares the simulated frequency curve with historical fault waveform data.

[0036] Data-driven parameter identification employs deep reinforcement learning to dynamically optimize the virtual inertia constant and droop coefficient.

[0037] Robustness testing under extreme conditions involves constructing scenarios with multiple perturbations to verify the model's stability within the confidence interval.

[0038] As a preferred embodiment of the optimal allocation method for grid-type energy storage capacity based on a frequency response model as described in this invention, wherein: S3 specifically includes:

[0039] The types of disturbances selected include step disturbances, which simulate the most severe instantaneous power surges and test the system's shock resistance; ramp disturbances, which simulate the continuous ramping of new energy power and verify the frequency gradual change control performance; and oscillation disturbances, which simulate intermittent power fluctuations and verify the system's damping characteristics.

[0040] The disturbance amplitude is set based on the power deficit quantified in step S1, and a disturbance baseline value is set; the grid operation mode correction is superimposed, with the upper limit taken for high load mode and the lower limit taken for nighttime off-peak mode.

[0041] As a preferred embodiment of the optimal allocation method for grid-type energy storage capacity based on a frequency response model as described in this invention, wherein: S3 specifically includes:

[0042] The incremental step strategy starts from the initial output coefficient set by S1 and gradually increases the proportion of energy storage output according to the preset step size.

[0043] Adjust the rules, and after each adjustment, let it stand still for a predetermined time until the system transitions to a quasi-steady state; if the frequency continues to deteriorate, trigger the step size reduction protection.

[0044] As a preferred embodiment of the optimal allocation method for grid-type energy storage capacity based on a frequency response model as described in this invention, wherein: S4 specifically includes:

[0045] The threshold settings are based on the following: maximum frequency deviation, with a hard limit of ±0.5Hz, which is tightened for scenarios with high penetration of new energy sources; maximum frequency change rate, referencing the tolerance limit of generator bearings; and quasi-steady-state frequency deviation, combined with the dead zone characteristics of primary frequency regulation.

[0046] As a preferred embodiment of the optimal allocation method for grid-type energy storage capacity based on a frequency response model as described in this invention, wherein: S4 specifically includes:

[0047] Capacity ratio conversion and optimization: Convert the output coefficient into a capacity ratio. Basic formula:

[0048]

[0049] In the formula, C ESS α represents the capacity share of energy storage systems, that is, the proportion of capacity that energy storage systems occupy in the overall power system; * ΔP is an adjustment coefficient used to adjust the proportion of energy storage capacity. loss Power loss refers to the reduction in power in a power system caused by various factors, which needs to be compensated by an energy storage system; η is the energy storage cycle efficiency, which refers to the energy utilization efficiency of the energy storage system during charging and discharging; DOD max Maximum depth of discharge refers to the maximum capacity release ratio allowed by the energy storage system during discharge.

[0050] Adjustable capacity compensation: If the system needs to provide inertia support, additional capacity is added.

[0051] The beneficial effects of this invention are as follows: The technical solution of this application first quantifies the power deficit of typical power plants under extreme short-circuit faults and the corresponding grid support requirements, and accurately measures the optimal configuration ratio of grid-type energy storage in new energy power plants under the constraints of three major indicators: frequency deviation, frequency change rate and quasi-steady-state frequency deviation.

[0052] This invention is based on constructing frequency response models for conventional generating units, wind turbine units, and grid-connected energy storage. By gradually increasing the initial output coefficient (KGFM) of the grid-connected energy storage, a series of simulation tests are conducted. This method allows for a relatively convenient determination of the required proportion of grid-connected energy storage configuration.

[0053] This invention can be tested and simulated in various grid-type energy storage experimental scenarios, without being limited by scenario type or operating conditions, and has good versatility and applicability. Attached Figure Description

[0054] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1 SFR model diagram of power system;

[0056] Figure 2 A diagram of the SFR model of a power system that takes into account the penetration rate of new energy sources;

[0057] Figure 3 This is a revised SFR model diagram of a power system that takes into account the penetration rate of new energy sources.

[0058] Figure 4 This is a frequency response model diagram for a conventional generating unit;

[0059] Figure 5 The active power control diagram for grid-type energy storage;

[0060] Figure 6 This is a diagram of the frequency response model for grid-type energy storage.

[0061] Figure 7 The system frequency response model diagram is shown below;

[0062] Figure 8 The frequency response waveform under extreme disturbance power is shown.

[0063] Figure 9 The maximum frequency deviation is plotted under extreme disturbance power.

[0064] Figure 10 The diagram shows the quasi-steady-state frequency deviation under extreme disturbance power.

[0065] Figure 11 Flowchart for solving the optimal allocation ratio of grid-type energy storage capacity;

[0066] Figure 12 This is a waveform diagram of the frequency response under high-frequency events;

[0067] Figure 13 The maximum frequency deviation diagram for high-frequency events;

[0068] Figure 14 A quasi-steady-state frequency deviation diagram under high-frequency events;

[0069] Figure 15 This is a waveform diagram of the frequency response under low-frequency events;

[0070] Figure 16 This is a graph showing the maximum frequency deviation under low-frequency events;

[0071] Figure 17 This is a diagram showing the quasi-steady-state frequency deviation under low-frequency events. Detailed Implementation

[0072] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0073] In this embodiment, firstly, by analyzing the SFR model of the power system, a frequency response model of a conventional generator set is constructed, including the following steps:

[0074] Step 1-1: Establishment of the power system frequency response model.

[0075] The power system frequency model is a closed-loop control system, mainly consisting of synchronous generators and loads, turbines, and their governors. This study aims to investigate the overall frequency characteristics of the power system; therefore, frequency dispersion and power angle stability are not considered, and specific governor control, turbine models, limiting, and other nonlinear elements and secondary frequency regulation are ignored. Any synchronous generator used in thermal or hydroelectric power plants can be simplified into the control closed loop shown in the figure below, and further aggregated into the power system frequency model.

[0076] Figure 1 In this equation, M is the rotor time constant related to the moment of inertia, D is the active frequency response coefficient of the load, and T is the equivalent inertial time constant of the turbine. a is the turbine characteristic coefficient; for a water turbine, a = -2, and for a steam turbine, 0 < a < 1. The governor is simplified as a pure gain of 1 / R.

[0077] For power systems with high penetration of new energy sources, a variable representing the penetration rate needs to be introduced into the above model. To simplify the analysis, it is assumed that the increase in the proportion of new energy power generation is achieved by shutting down conventional synchronous generator units such as hydropower and thermal power plants. The generation coefficient of conventional synchronous generator units is defined as follows:

[0078]

[0079] If the penetration rate of new energy sources continues to increase, the power generation coefficient K of conventional synchronous generator units will continue to decrease, with a value range of 0≤K≤1.

[0080] Figure 1 The model in the text is based on the capacity of the generator (or turbine). If the system load power is used as the benchmark, then the rotor time constant of the synchronous unit and the output power of the turbine should both be multiplied by a coefficient K. This yields a simplified aggregation frequency model for a high-penetration renewable energy power system, such as... Figure 2 As shown.

[0081] Figure 2 In the system diagram: ωref is the reference value of the system's electrical angular frequency; ΔPl is the change in the load's active power; Δω is the change in the system's electrical angular frequency after being disturbed; η is the proportion of new energy units connected to the system; a is the proportion of high-pressure cylinders in the turbine; T is the turbine's equivalent inertia time constant; M is the rotor time constant related to the moment of inertia; and D is the load's active frequency response coefficient. ωref, ΔPl, and Δω are all per-unit values.

[0082] For the SFR model of a power system considering the penetration rate of new energy sources, the system frequency will fluctuate under active power disturbances, and the relevant extreme points of the system frequency curve have been studied in depth. After the system experiences a frequency disturbance of ΔPl, the expression in the complex frequency domain can be written as:

[0083]

[0084] Without utilizing the frequency control capabilities of renewable energy generation, the frequency characteristics of the power system will continuously deteriorate as the generation coefficient of conventional synchronous generators decreases, thus limiting the increase in the penetration rate of renewable energy generation. Model 1 equates renewable energy power electronic converters to power sources without inertia and frequency regulation characteristics. However, in actual operation, whether a generator has inertia and whether it participates in frequency regulation are not necessary and sufficient conditions. Therefore, the coefficients of both the system equivalent inertia 1-ηM and the equivalent droop coefficient 1-ηR are calculated using 1-η, which may not be applicable under certain operating conditions.

[0085] Will Figure 2 In the drooping stage and the unit inertia and load stages, the coefficients η are rewritten as ηR and ηM, respectively, to represent the proportion of units not participating in frequency regulation and the proportion of units without inertia:

[0086]

[0087] In the formula: Represents the total installed capacity of the system; ∑S R=∞ represents the sum of the capacities of units in the system that do not participate in frequency regulation; ∑S M =0 indicates that the system does not have the sum of inertia units or power supply capacity.

[0088] Therefore, 1-ηR can be called the effective droop proportional coefficient of the system, and 1-ηM can be called the effective inertia proportional coefficient of the system. Figure 2 Modified to Figure 3 .

[0089] When the power undergoes a step change, the equation can be rewritten as:

[0090]

[0091] Performing an inverse Laplace transform on the above equation, we obtain:

[0092]

[0093] In the formula, ωn and ξ are rewritten as:

[0094]

[0095] In the formula:

[0096]

[0097] The time when the system frequency reaches its extreme point and the change in maximum frequency are:

[0098]

[0099]

[0100] This model can establish a power system SFR model that is closer to the actual grid operation by distinguishing the proportion of units that actually participate in frequency regulation and those that actually have inertia in the system, thus providing an operating scenario for grid-based energy storage SFR models.

[0101] Step 1-2: Construct a frequency response model for a conventional generator set.

[0102] Conventional generator sets, possessing an actual rotating rotor, can spontaneously and without delay respond to disturbances due to their inertia, converting rotor kinetic energy into mechanical energy to provide inertial power support. For a synchronous generator set, neglecting system damping, its rotor motion equation is:

[0103]

[0104] In the formula, fg is the frequency of the generator set; ΔPm and ΔPe are the increments of the mechanical input power and electromagnetic output power of the generator set, respectively. hg is the inertial constant of the generator set, in seconds. From an energy perspective, the synchronous inertia can also be expressed as:

[0105]

[0106] In the formula, J is the moment of inertia; ω is the mechanical angular velocity; SN is the rated capacity of the generator; and EK is the kinetic energy stored in the rotor, i.e., the inertial quantification value in energy form, with the unit MW·s.

[0107] This paper mainly considers thermal power units (TPU) and gas turbine units (GTU) for conventional power generation, while also including wind turbine generator sets. The overall frequency response model is as follows: Figure 4 As shown in the figure. In the figure, KG and KT are the capacity ratio coefficients of the gas turbine and thermal power unit, and kW is the capacity ratio coefficient of the wind turbine unit; RG and RT are the droop coefficients of the gas turbine and thermal power unit, and RW is the droop coefficient of the wind turbine unit; Dsys is the system damping, and Hsg is the synchronous inertia time constant. The calculation formula is:

[0108] H SG =K G H G +K T H T +K W H W (15)

[0109] To analyze the impact of grid-type energy storage capacity on system transient power support, a system frequency response model including energy storage units is constructed, comprising the following steps:

[0110] Step 2-1: Construct the frequency response model of the energy storage unit

[0111] A grid-based control strategy was selected to build a frequency response model for grid-based energy storage, and the basic control block diagram of grid-based energy storage was established. Figure 5 The SFR model of grid-type energy storage is obtained by disassembling and reconstructing the structure. Figure 6The SFR model adopts a modular approach, independently modeling each control component and dynamic characteristic of grid-connected energy storage, ensuring applicability to different parameter inputs within this framework. Modules such as droop control, SOC control, and power synchronization can be customized and adjusted according to specific needs without affecting the overall system functionality. This flexibility enables this general-purpose model to address frequency regulation requirements under different grid environments. Whether in weak or strong grids, the model effectively captures the frequency impact of the energy storage system. Figure 6 As shown.

[0112] Grid-based technology enables energy storage converters to mimic the inertia and damping effects of synchronous generators, thereby enhancing their grid support capability. Based on this characteristic, the conventional active power control strategy for grid-based systems satisfies:

[0113]

[0114] In the formula: Pm is the mechanical power output of the virtual synchronous machine; ωref is the rated angular frequency of the grid voltage; ωg is the output angular frequency; kd is the equivalent droop coefficient; Pref is the given reference active power; Pe is the active power (electrical power) output of the grid-type converter; PD is the damping power, representing the damping's ability to impede system oscillations, and has the following relationship:

[0115] P D =D·Δω1(17)

[0116] The system rotor angle is obtained from the following relationship:

[0117] δ=∫Δω1(18)

[0118] Based on the principles of grid control and power flow calculation, the calculation method for the active power Pe output by the grid converter is as follows:

[0119]

[0120] In the formula: V is the output voltage amplitude of the grid-connected converter; Eg is the grid voltage amplitude; Xg is the grid line impedance. Eg*V / Xg is defined as the system gain coefficient kp.

[0121] The transfer function is derived as follows:

[0122]

[0123] Step 2-2: Construct the system frequency response model.

[0124] Based on the analysis and modeling of the inertial response and primary frequency regulation characteristics of traditional generating units and energy storage units, a system is established as follows: Figure 7The system frequency response model is shown. When a disturbance occurs in the system, the conventional unit provides an instantaneous inertial response through rotor inertia, while the grid-type energy storage simulates inertial behavior through virtual synchronous generator (VSG) control to compensate for the inertia deficiency and enhance the initial frequency support. At the same time, the droop control of the energy storage and the governor of the conventional unit work together to regulate frequency fluctuations: the energy storage system suppresses oscillations through the equivalent damping coefficient DGFM, while the conventional unit uses the regulation coefficients RG and RT to recover the frequency. Together, they constitute a coordinated frequency support mechanism to ensure the stable operation of the system.

[0125] Figure 7 In this context, the capacity ratio of KGFM energy storage and the Hsys time constant are defined as follows:

[0126] H sys =K G H G +K H H H +K T H T +K GFM H GFM (twenty one)

[0127] Frequency is a key indicator of power grid stability. Load power fluctuations can cause system frequency shifts, requiring energy storage devices to provide transient power support and accelerate frequency recovery. Therefore, the rational allocation of energy storage capacity is crucial. Based on the established frequency response model, the optimal allocation of energy storage capacity in grid-type power stations is further calculated and verified, including the following steps:

[0128] Step 3-1: Determine the three-dimensional evaluation index of frequency response.

[0129] To quantify the support effect of grid-based energy storage on grid frequency, this patent proposes a three-dimensional frequency evaluation index system that includes transient and steady-state parameters, based on GB / T40595-2021 "Power System Power Balance Guidelines" and IEEE Std1547-2018.

[0130] (1) Maximum frequency deviation (Δfmax): The maximum absolute value of the instantaneous frequency deviation (50Hz) of the system after being disturbed. It reflects the instantaneous support capability of energy storage for power deficit in the early stage of a fault, and is positively correlated with the power response speed of energy storage (such as droop control gain) and reserve capacity. Its constraint condition is Δfmax≤0.5Hz.

[0131] (2) Maximum Rate of Frequency Change (RoCofmax): The maximum rate of frequency change per unit time, characterizing the system's inertia level. Grid-type energy storage simulates the inertial response of a synchronous machine through virtual inertia control, suppressing frequency abrupt changes. Its value depends on the ratio of the virtual inertia constant (H) to the disturbance power amplitude. The constraint is RoCofmax ≤ 0.5 Hz / s, and the calculation formula is:

[0132]

[0133] (3) Quasi-steady-state frequency deviation (Δfss): The frequency offset of the system when it enters the quasi-steady state after a disturbance. It reflects the continuous power regulation capability of the energy storage and needs to be achieved through droop control or secondary frequency modulation, and is directly related to the energy storage capacity configuration. In this paper, the frequency value 10 seconds after the system starts operating is taken as the quasi-steady-state frequency, and its constraint is Δfss≤0.2Hz. The calculation formula is:

[0134]

[0135] (4) Disturbance power: By setting up a 750KV substation open circuit test in a real example system of a power grid in western China, the extreme power disturbance that does not meet the three major frequency indicators under load shedding event is obtained and standardized to obtain the disturbance power ΔPl under extreme event.

[0136] During the 2s simulation, the power grid simulator was set with input disturbance power, and the corresponding frequency response model was obtained as follows: Figure 8 As shown.

[0137] from Figure 9 It can be seen from this that the maximum frequency deviation (Δfmax) at this time is 0.50009Hz; from Figure 10 It can be seen that the quasi-steady-state frequency deviation (Δfss) is 0.38617Hz at this time; the maximum frequency change rate (RoCofmax) can be calculated to be 0.95699Hz / s.

[0138] Step 3-2: Determine the optimal capacity allocation that meets the frequency assessment criteria.

[0139] In step 3-1, the power disturbance value ΔPl under extreme events has been obtained. At this point, the frequency index clearly does not meet the standards for stable grid operation, requiring grid-based energy storage to provide power support to maintain grid stability. First, the output coefficients KG and KT of conventional generating units and wind turbines are fixed. The output coefficient KGFM of the grid-based energy storage is changed to alter the magnitude of the supporting power. The experiment is set with an initial energy storage output of 0 and a power disturbance of ΔPl. Then, the maximum frequency deviation, maximum frequency change rate, and quasi-steady-state frequency deviation are recorded for each experiment and compared with the indicators. Finally, the relevant indicator data are evaluated. If the requirements are not met, the coefficients are iteratively processed, and the experiment is repeated until the provided power support can maintain all three major grid frequency indicators within the constraints. The grid-based energy storage output coefficient at this point is the optimal energy storage output coefficient that satisfies the frequency requirements.

[0140] In summary, this invention proposes quantitative evaluation indicators to characterize the transient stability of GFM from three aspects: maximum frequency deviation, maximum frequency change rate, and quasi-steady-state frequency change rate, which facilitates the analysis in the following text.

[0141] Based on the above analysis, this embodiment proposes an optimal capacity allocation algorithm based on the frequency response (SFR) model, such as... Figure 11 As shown, it includes the following steps:

[0142] Step 1: First, quantify the power deficit ΔPl of a typical power station under extreme short-circuit faults, and determine the values ​​of the virtual inertia constant and virtual droop coefficient of the grid-type energy storage system. Then, set an initial output coefficient KGFM and an adjustment step size ΔKGFM for the grid-type energy storage to prepare for subsequent parameter scanning and optimization.

[0143] Step 2: Based on the inertia and frequency regulation characteristics of conventional units, wind turbines and grid-type energy storage, construct frequency response models for each unit and integrate them into the system frequency response model;

[0144] Step 3: In the system frequency response model, initialize parameters KG, KT, and KGFM, and apply the disturbance ΔPl obtained in Step 1. Gradually increase the initial output coefficient KGFM of the grid-type energy storage, perform simulations, and record three key frequency stability indicators: maximum frequency deviation Δf. max Maximum rate of change of frequency RoCof max Quasi-steady-state frequency deviation Δf ss The above indicators are then compared one by one with the pre-set constraint limits to determine whether the candidate KGFM meets the frequency stability requirements.

[0145] Step 4: Determine the optimal grid-type energy storage capacity ratio: If Δf max RoCof max and Δfss If all frequency stability constraints are met, then the current KGFM is the optimal output coefficient, and the algorithm ends. Otherwise, increase KGFM (KGFM←KGFM+ΔKGFM), return to step three, and continue iterating until all constraints are met or the design upper limit is reached.

[0146] Further preferably, the frequency stability index includes:

[0147] Frequency stability indicators: maximum frequency deviation, maximum frequency change rate, and quasi-steady-state frequency deviation are further optimized. Step one includes the following sub-steps:

[0148] Step S101: Quantify the power deficit ΔPl of a typical power station under extreme short-circuit faults.

[0149] Step S102: Determine the values ​​of the virtual inertia constant H and the virtual droop coefficient Kd of the grid-type energy storage in the system.

[0150] Assuming the grid-connected energy storage VSG operates at rated power and is in a discharging state, when the grid frequency deviates, the larger the virtual droop coefficient, the greater the output power of the grid-connected energy storage VSG, and the more significant its effect on improving system frequency stability. However, the virtual droop coefficient cannot be set too large, lest the output power of the grid-connected energy storage VSG exceed the maximum value of the energy storage converter due to grid frequency disturbances. The frequency deviation range of the power system during quasi-stable operation is ±0.2Hz; therefore, the virtual droop coefficient should satisfy the following condition:

[0151]

[0152] In the formula, Pmax is the maximum discharge power of the VSG, and ηch and ηdisch are the charging efficiency and discharging efficiency of the energy storage, respectively.

[0153] Simultaneously, when designing the virtual droop factor, the SOC value of the grid-connected energy storage system must also be considered. When the SOC value of the grid-connected energy storage system is low, it may not have sufficient remaining capacity to respond to the additional output required for high-frequency support. To ensure that the grid-connected energy storage VSG can operate continuously as much as possible under low power conditions and avoid downtime, the droop factor can be reduced to decrease the additional active power output required to respond to grid frequency deviations. After the grid frequency recovers, the grid-connected energy storage VSG can continue to operate according to the dispatch instructions from the higher-level system. Ignoring the energy consumed by the resistance of the lithium battery pack, the virtual droop factor should satisfy the following:

[0154]

[0155] In the formula, SOCt0 is the SOC value at the initial moment of the grid frequency disturbance, SOCmin is the minimum SOC value that the lithium battery can operate at, and SOCmax is the maximum SOC value that the lithium battery can operate at.

[0156] The virtual droop factor needs to satisfy both of the above conditions to maintain the operation of the grid-type energy storage converter. Therefore, the virtual droop factor selected in this paper is as follows:

[0157]

[0158] According to the national standard "General Technical Principles for Virtual Synchronizers", the active power frequency regulation response time should not exceed 500ms, the active power frequency regulation adjustment time should not exceed 1s, and the power error should be controlled within 2%. Therefore, the active power closed-loop design is an underdamped system. The typical time of an underdamped system meets the national standard as follows:

[0159]

[0160] If D is set to a value of 5000, then J∈(0, 34.09).

[0161] Furthermore, in power systems, power imbalances caused by large-scale disturbances are one of the main factors leading to abnormal frequency drops. This was clearly demonstrated in the major blackout in the UK on August 9, 2019. In that event, the consecutive tripping of two power sources created a significant power shortage, which in turn caused a decrease in system frequency. As the system frequency dropped to a certain threshold, it triggered the first-level low-frequency load shedding measures, leading to the major blackout. Maintaining sufficient system inertia is a crucial measure to prevent the frequency from falling below the safety threshold, such as the minimum frequency stipulated by 49Hz, when facing large disturbances.

[0162] Adequate system inertia is crucial for ensuring power systems maintain frequency stability after large-scale disturbances, especially preventing the frequency from exceeding the minimum frequency threshold. When the frequency exceeds the primary frequency regulation threshold after a disturbance, the speed governor activates and increases active power. As the inertia and primary frequency regulation work together to continuously suppress frequency changes, the frequency eventually reaches the minimum frequency point fnadir.

[0163] After the disturbance occurs, the lowest point of the system frequency can be represented as:

[0164]

[0165] As can be seen from the above formula, when the minimum frequency value and the disturbance power are determined, the inertia requirement of the current system can be calculated. The minimum inertia requirement considering the minimum frequency constraint can be represented by Hnadir. Based on the set minimum frequency constraint, the minimum inertia required by the system under this disturbance can be calculated, as shown in the following formula:

[0166]

[0167] For grid-type energy storage, the equivalent synchronous machine inertia can be calculated as follows:

[0168]

[0169] In the formula, H and SNB are the virtual inertia, kinetic energy change, and rated capacity of the grid-type energy storage device, respectively, and J = 2H.

[0170] Therefore, virtual inertia should meet the following requirements:

[0171]

[0172] Step S103: Set the initial output coefficient KGFM for grid-type energy storage and adjust the step size ΔKGFM.

[0173] Further preferably, step two includes the following sub-steps:

[0174] Step S201: Based on the inertia and primary frequency regulation characteristics of conventional units, wind turbine units, and grid-connected energy storage, construct the corresponding frequency response models:

[0175] Step S202: Integrate the frequency response models of the three types of units to obtain the system frequency response model.

[0176] Further preferably, step three includes the following sub-steps:

[0177] Step S301: Apply the disturbance ΔPl obtained in Step 1. Gradually increase the initial output coefficient KGFM of the grid-type energy storage and perform simulation;

[0178] Step S302: Record the maximum frequency deviation Δf max Maximum rate of change of frequency RoCof max Quasi-steady-state frequency deviation Δf ss Frequency stability index value;

[0179] Step S303: Compare the candidate KGFM with the set constraint limits one by one to determine whether the candidate KGFM meets the frequency stability requirements.

[0180] Further preferably, step four includes the following sub-steps:

[0181] Step S401: If Δf max RoCof max and Δf ss If all of them satisfy the frequency stability constraint, then the current KGFM is the optimal output coefficient;

[0182] Step S402: If not satisfied, return to step three and continue iterating until all constraints are satisfied or the design limit is reached;

[0183] A detailed SFR model of a wind-solar-storage coupled power station in a power grid example system in western China was established. To verify the proposed optimal capacity ratio evaluation method for grid-connected energy storage power stations, two sets of typical operating condition experiments were designed: under different disturbance conditions, the output coefficient of the energy storage system was gradually increased, and the system frequency response and key operating indicators were recorded until the frequency stability requirements were met. The simulation results of Experiment 1 are as follows: Figure 12-14 As shown, the simulation results of Experiment 2 are as follows: Figure 15-17 As shown.

[0184] Scenario 1: At a certain moment, the shedding of a large load causes the system to experience an increased frequency of events.

[0185] (1) When the energy storage output coefficient is 0.00807, the grid-type energy storage capacity accounts for 1.558%. When extreme power disturbances occur, the grid-type energy storage cannot provide sufficient power support and cannot maintain the stable operation of the power grid.

[0186] (2) When the energy storage output coefficient is 0.01614, the grid-type energy storage capacity accounts for 3.068%. When extreme power disturbances occur, the grid-type energy storage cannot provide sufficient power support and cannot maintain the stable operation of the power grid.

[0187] (3) When the energy storage output coefficient is 0.0269, the grid-type energy storage capacity accounts for 5.010%. When extreme power disturbances occur, the grid-type energy storage can provide sufficient power support to maintain the stable operation of the power grid and meet the power index constraints. The specific proportion is shown in Table 1.

[0188] Scenario 2: A large load is received at a certain moment, causing the system to experience a frequency reduction event.

[0189] (1) When the energy storage output coefficient is 0.10088, the grid-type energy storage capacity accounts for 1.9396%. When extreme power disturbances occur, the grid-type energy storage cannot provide sufficient power support and cannot maintain the stable operation of the power grid.

[0190] (2) When the energy storage output coefficient is 0.02018, the grid-type energy storage capacity accounts for 3.8053%. When extreme power disturbances occur, the grid-type energy storage cannot provide sufficient power support and cannot maintain the stable operation of the power grid.

[0191] (3) When the energy storage output coefficient is 0.0269, the grid-type energy storage capacity accounts for 5.0102%. When extreme power disturbances occur, the grid-type energy storage can provide sufficient power support to maintain the stable operation of the power grid and meet the power index constraints.

[0192] Based on the above experimental results Δfmax RoCof max and Δf ss When the frequency stability constraint is met, the grid-type energy storage capacity ratio at this point is the optimal capacity ratio.

[0193] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for optimal allocation of grid-type energy storage capacity based on a frequency response model, characterized in that, include: S1. Quantify the power deficit of typical stations under extreme short-circuit faults, determine the virtual inertia constant and virtual droop coefficient of grid-type energy storage, and set the initial output coefficient and adjustment step size. S2. Based on the inertia and frequency regulation characteristics of conventional units, wind turbines and grid-type energy storage, construct and integrate the system frequency response model; S3. Apply a power disturbance to the system, gradually increase the initial output coefficient and simulate, and record the maximum frequency deviation, the maximum frequency change rate and the quasi-steady-state frequency deviation. S4. When the maximum frequency deviation, the maximum frequency change rate, and the quasi-steady-state frequency deviation are all less than or equal to the predetermined thresholds, the energy storage capacity ratio corresponding to the current initial output coefficient is the optimal ratio.

2. The optimal allocation method for grid-type energy storage capacity based on a frequency response model as described in claim 1, characterized in that, S2 specifically includes: In the conventional unit model, the synchronous generator rotor motion equation is established to quantify the suppressive effect of rotational inertia and damping coefficient on the system frequency change rate; in the governor system model, a first-order lag model of valve opening-power response is embedded to simulate the delay characteristics of primary frequency regulation. The wind turbine model is reconstructed with low inertia characteristics. For the doubly fed induction generator (DFIG) turbine, only the equivalent inertia of the converter capacitor is retained, and the frequency coupling effect is described by the dynamic equation of the phase-locked loop. For the direct-drive turbine, a virtual inertia control loop is constructed. Frequency regulation capability is limited, considering maximum power point tracking and rotor speed protection, and an upper limit of active power reserve is set. The grid-type energy storage model adopts a virtual synchronous machine algorithm and multi-timescale control, which is divided into millisecond-level virtual inertia response suppression RoCoF; second-level droop control to compensate for steady-state frequency deviation; and minute-level state of charge coordination to prevent over-discharge.

3. The optimal allocation method for grid-type energy storage capacity based on a frequency response model as described in claim 2, characterized in that, S2 specifically includes: System-level model integration and a unified state-space framework combine the state equations of conventional units, wind power, and energy storage to construct a multi-dimensional state matrix; frequency-power dynamic coupling links the power output of each source with the grid frequency through the node admittance matrix. Enhanced high-penetration scenarios and optimized damping ratios are achieved by introducing virtual impedance control in eigenvalue analysis to suppress subsynchronous oscillations caused by wind power grid connection. Communication delay compensation is also implemented by correcting control command transmission errors based on a 5G URLLC time delay prediction model.

4. The optimal allocation method for grid-type energy storage capacity based on a frequency response model as described in claim 3, characterized in that, S2 specifically includes: The digital twin verification platform imports the actual power grid topology into the RTDS real-time simulator and compares the simulated frequency curve with historical fault waveform data. Data-driven parameter identification employs deep reinforcement learning to dynamically optimize the virtual inertia constant and droop coefficient. Robustness testing under extreme conditions involves constructing scenarios with multiple perturbations to verify the model's stability within the confidence interval.

5. The optimal allocation method for grid-type energy storage capacity based on a frequency response model as described in claim 4, characterized in that, S1 specifically includes: Fault scenario modeling is performed based on historical fault recording data of the power grid, extracting three-phase short-circuit and single-phase ground fault waveforms of typical power plants; instantaneous power theory is used to decompose the orthogonal components of the fault current and calculate the active power surge value; and the impact of renewable energy disconnection is corrected. Considering the low-voltage ride-through failure rate of wind turbines, when the voltage drop exceeds a predetermined proportion, the probability of wind turbines disconnecting from the grid is calculated, and the power deficit is added: In the formula, ΔP total The total power deficit, i.e., the total system power loss caused by wind turbine disconnection, is usually measured in watts; ΔP fault The basic fault power deficit refers to the power loss caused by fault factors other than wind turbine disconnection, including conventional generator failures and transmission line faults; N WT P represents the total number of wind turbine units in a wind farm or related area; WT,k I represents the rated power or actual output power of the k-th wind turbine, reflecting the power that each wind turbine can provide during normal operation; trip This is the offline indicator function, which is a binary variable.

6. The optimal allocation method for grid-type energy storage capacity based on a frequency response model as described in claim 5, characterized in that, S1 specifically includes: The key parameters of the grid-type energy storage are set as virtual inertia constants, and the design is reversed based on the system inertia requirements; a dynamic adaptive mechanism is used to deploy the J function in the dSPACE controller to improve the inertia support strength during high-frequency disturbances; The virtual droop coefficient is designed based on the principle of equivalence of adjustment rate; the segmented droop strategy adopts flexible droop when the frequency deviation is less than the predetermined threshold, and switches to rigid droop when the frequency deviation is greater than the predetermined threshold. Initial output coefficient, based on power deficit ratio initialization, corrected for renewable energy penetration rate: If the proportion of wind power installed capacity is greater than the predetermined proportion, the initial output coefficient increases; if the system includes pumped storage units, the initial output coefficient decreases. Adjusting the step size, for high-proportion renewable energy grids, the binary search method and golden ratio are used to accelerate the search; for weak grids, Lyapunov stability verification is added; multi-energy storage collaboration is achieved through pre-configuration of damping controllers.

7. The optimal allocation method for grid-type energy storage capacity based on a frequency response model as described in claim 6, characterized in that, S3 specifically includes: The types of disturbances selected include step disturbances, which simulate the most severe instantaneous power surges and test the system's shock resistance; ramp disturbances, which simulate the continuous ramping of new energy power and verify the frequency gradual change control performance; and oscillation disturbances, which simulate intermittent power fluctuations and verify the system's damping characteristics. The disturbance amplitude is set based on the power deficit quantified in step S1, and a disturbance baseline value is set; the grid operation mode correction is superimposed, with the upper limit taken for high load mode and the lower limit taken for nighttime off-peak mode.

8. The method for optimal allocation of grid-type energy storage capacity based on a frequency response model as described in claim 7, characterized in that, S3 specifically includes: The incremental step strategy starts from the initial output coefficient set by S1 and gradually increases the proportion of energy storage output according to the preset step size. Adjust the rules, and after each adjustment, let it stand still for a predetermined time until the system transitions to a quasi-steady state; if the frequency continues to deteriorate, trigger the step size reduction protection.

9. The method for optimal allocation of grid-type energy storage capacity based on a frequency response model as described in claim 8, characterized in that, S4 specifically includes: The threshold settings are based on the following: maximum frequency deviation, with a hard limit of ±0.5Hz, which is tightened for scenarios with high penetration of new energy sources; maximum frequency change rate, referencing the tolerance limit of generator bearings; and quasi-steady-state frequency deviation, combined with the dead zone characteristics of primary frequency regulation.

10. The method for optimal allocation of grid-type energy storage capacity based on a frequency response model as described in claim 9, characterized in that, S4 specifically includes: Capacity ratio conversion and optimization: Convert the output coefficient into a capacity ratio. Basic formula: In the formula, C ESS α represents the capacity share of energy storage systems, that is, the proportion of capacity that energy storage systems occupy in the overall power system; * ΔP is an adjustment coefficient used to adjust the proportion of energy storage capacity. loss Power loss refers to the reduction in power in a power system caused by various factors, which needs to be compensated by an energy storage system; η is the energy storage cycle efficiency, which refers to the energy utilization efficiency of the energy storage system during charging and discharging; DOD max Maximum depth of discharge refers to the maximum capacity release ratio allowed by the energy storage system during discharge. Adjustable capacity compensation: If the system needs to provide inertia support, additional capacity is added.

Images