Power distribution method for hybrid energy storage, medium and electronic device

Abstract

The application provides a hybrid energy storage power distribution method, medium and electronic equipment, which can collect power grid frequency change data and operation state data of multiple energy storage units, predict power grid frequency modulation demand in a future time period according to the collected data, determine a frequency modulation event type according to a prediction result and an event classification rule, determine a risk index of the energy storage unit according to the operation state data, the prediction result and the frequency modulation event type, determine pre-distribution power of the energy storage unit according to the frequency modulation event type and the risk index, correct the pre-distribution power according to a constraint condition of the energy storage unit and determine a final power instruction, so that the forward-looking of the hybrid energy storage power distribution can be improved, the impact on the energy storage equipment can be reduced, the frequency modulation tracking performance and control stability can be improved, the available adjustment capacity in subsequent frequency modulation events can be improved, and the performance attenuation caused by long-term bias operation of the fast energy storage can be reduced.

Description

Technical Field

[0001] This invention relates to the field of smart energy storage technology, and in particular to a power distribution method, medium, and electronic device for hybrid energy storage. Background Technology

[0002] Currently, with the integration of a high proportion of new energy sources and the advancement of the construction of new power systems, the equivalent inertia of the power grid system has decreased and the randomness of power disturbances has increased. Consequently, the frequency fluctuations of the power grid exhibit characteristics of suddenness, high frequency, and continuous deviation. To better cope with the characteristics of power grid frequency fluctuations, hybrid energy storage technology has emerged.

[0003] Hybrid energy storage technology typically consists of energy-type energy storage and power-type energy storage. Energy-type energy storage, such as battery energy storage, has the advantages of high energy density and strong continuous support capability, while power-type energy storage, such as supercapacitors or flywheel energy storage, has the advantages of high power density and fast response speed. Hybrid energy storage technology combines the advantages of both types of energy storage and has been widely used in primary frequency regulation, secondary frequency regulation, and automatic generation control (AGC) ancillary service scenarios.

[0004] Currently, there are several solutions for achieving grid frequency regulation using hybrid energy storage technology. For example, Chinese patent application (publication number CN117175646A) "A method and system for controlling energy storage participation in primary frequency regulation in a hybrid energy storage system" switches or adjusts the proportion of primary frequency regulation undertaken by the battery and supercapacitor based on real-time frequency difference and energy storage state of charge to improve continuous frequency regulation capability; Chinese patent application (publication number CN117674192A) "A strategy for hybrid energy storage to participate in power distribution and frequency regulation in wind, solar and energy storage stations under multiple scenarios" achieves hybrid energy storage participation in power distribution and frequency regulation under multiple scenarios by adaptively adjusting the low-pass filter time constant and virtual synchronous machine parameters; Chinese patent application (publication number CN111446725A) "A method for hybrid energy storage frequency regulation control in microgrids" is based on a droop control framework, with supercapacitors quickly responding to frequency changes, batteries undertaking power changes and secondary regulation, and superimposed with SOC control, etc.

[0005] Existing hybrid energy storage frequency regulation control schemes can be categorized as follows: The first is a fixed-ratio allocation method, which distributes frequency regulation power commands to battery energy storage and fast energy storage according to a preset ratio. This method has a simple control structure but is difficult to adapt to changes in operating conditions. The second is a frequency division method based on high-pass / low-pass filtering or wavelet decomposition, which allocates low-frequency, slow-changing power components to battery energy storage and high-frequency, fast-changing power components to supercapacitors or flywheel energy storage. This method can solve the problem of division of labor between fast and slow energy storage, but it is essentially still based on static mapping of the frequency band characteristics of current or short-term historical signals. The third is an adaptive allocation method based on SOC, SOH, temperature, and power thresholds, which improves the system's operational capability under state constraints by adjusting the load ratio, frequency division point, droop coefficient, or switching threshold in real time. The fourth is an optimized allocation method based on economic efficiency, life cost, or AGC performance indicators, which achieves power allocation through rolling optimization, two-layer optimization, or model predictive control.

[0006] In summary, the common approach of existing solutions is to allocate power based on the current frequency difference, current power, current SOC / SOH state, or current frequency band decomposition results. The focus is on how to allocate power under the current state, which is essentially a "current state driven" power allocation strategy.

[0007] Therefore, existing hybrid energy storage frequency regulation control schemes have the following drawbacks:

[0008] 1. Lack of foresight: Power allocation is mostly based on the current frequency difference, filter output, or state of charge, lacking a mechanism to judge the future evolution of frequency modulation events, and unable to distinguish events with similar initial amplitudes but significantly different subsequent durations, reversal trends, and recovery processes.

[0009] 2. Crude division of responsibilities: For different types of frequency regulation events, a uniform allocation rule is usually used, which leads to confusion between the responsibilities of rapid energy storage and battery energy storage, resulting in problems such as premature depletion of rapid energy storage, premature high-rate intervention of batteries, or inappropriate timing of replenishment during the recovery phase.

[0010] 3. Insufficient coupling of risk constraints: Although SOC or SOH boundaries are considered, there is a lack of unified characterization for battery life risks, rapid energy storage energy depletion risks, and mode switching jitter risks, making it impossible to form dynamic constraints that are linked to event types.

[0011] 4. Insufficient real-time correction capability: If power boundary, energy boundary, abnormal temperature rise or repeated frequency shift occurs during frequency regulation, the existing static or weak adaptive allocation strategy often cannot quickly trigger redistribution, resulting in a decrease in command tracking performance and even aggravating energy storage degradation.

[0012] Therefore, a new technical solution is needed to address the problems of insufficient foresight, weak event differentiation ability, dispersed risk constraints, and insufficient real-time correction ability of existing hybrid energy storage frequency regulation power allocation schemes. Summary of the Invention

[0013] One objective of this application is to provide a power allocation method, medium, and electronic equipment for hybrid energy storage, in order to solve the problems of insufficient foresight, weak event differentiation capability, dispersed risk constraints, and insufficient real-time correction capability of existing hybrid energy storage frequency regulation power allocation schemes.

[0014] To achieve the above objectives, some embodiments of this application provide a power allocation method for hybrid energy storage, the method comprising:

[0015] Data on grid frequency changes and the operating status of multiple energy storage units are collected according to a preset sampling period. The grid frequency change data includes grid frequency deviation, frequency change rate and automatic generation control command trajectory. The operating status data includes SOC, SOH, terminal voltage, current, temperature, internal resistance, available charging and discharging power and remaining available energy.

[0016] Based on the collected power grid frequency change data and operation status data, the power grid frequency regulation demand is predicted for a future preset time period;

[0017] Based on the prediction results and preset event classification rules, the frequency regulation event type is determined. The prediction results include future total frequency regulation demand power sequence data, predicted maximum amplitude, symbol duration, number of symbol inversions, symbol inversion density, predicted remaining duration, recovery phase slope, and whether the predicted end power enters the dead zone. The frequency regulation event types include impulsive events, persistent deviation events, oscillation events, and recovery events.

[0018] Based on operational status data, forecast results, and frequency regulation event types, risk indices for multiple energy storage units are determined. These risk indices include battery life risk index, rapid energy storage depletion risk index, and mode switching vibration risk index.

[0019] The pre-allocated power of multiple energy storage units is determined based on the frequency regulation event type and risk index;

[0020] The pre-allocated power is corrected based on the preset constraints of multiple energy storage units, and the final power command is determined.

[0021] Furthermore, based on the collected power grid frequency change data and operational status data, the power grid frequency regulation demand within a preset future time period is predicted, including:

[0022] Historical sliding window data is constructed based on power grid frequency change data and operating status data;

[0023] Calculate the total frequency regulation demand power sequence data based on the power grid frequency change data;

[0024] Based on historical sliding window data and total frequency regulation demand power sequence data, the future total frequency regulation demand sequence data is predicted using a preset time series prediction algorithm.

[0025] Extract event evolution feature data from future total frequency regulation demand sequence data;

[0026] The future total frequency regulation demand sequence data and event evolution characteristic data will be used as the prediction results output.

[0027] Furthermore, based on the prediction results and preset event classification rules, the frequency modulation event types are determined, including:

[0028] If the predicted maximum amplitude is greater than the preset impact threshold, the predicted remaining duration is less than the preset remaining duration threshold, and the sign inversion density is less than the preset sign density threshold, the frequency modulation event type is determined to be an impact event.

[0029] If the symbol duration is greater than the preset symbol duration threshold and the symbol inversion density is less than the preset symbol density threshold in the prediction results, the frequency modulation event type will be determined as a persistent deviation event.

[0030] If the number of symbol inversions in the prediction results is greater than the preset symbol inversion threshold or the symbol inversion density is greater than the preset symbol density threshold, the frequency modulation event type will be determined as an oscillating event.

[0031] If the slope of the recovery phase is less than zero and the predicted end power enters the dead zone in the prediction results, the frequency modulation event type is determined to be a recovery event.

[0032] Furthermore, based on operational status data, forecast results, and frequency regulation event types, risk indices for multiple energy storage units are determined, including:

[0033] The battery life risk index is determined based on the ratio of the available charge / discharge power or current of the battery energy storage unit to the preset rated value, the current change amplitude at adjacent sampling times, the number of power reversals, the temperature rise trend, the internal resistance change trend, and the SOC safety margin. The energy storage unit includes battery energy storage unit and fast energy storage unit.

[0034] The energy depletion risk index of fast energy storage is determined based on the degree of deviation of the SOC from the median, the instantaneous power ratio, the predicted remaining duration, the set maximum duration, and the remaining available energy margin of the fast energy storage unit.

[0035] Based on the sign inversion density, actual power switching frequency, and hold-up margin in the prediction results corresponding to the energy storage unit, the mode switching chatter risk index is determined.

[0036] Furthermore, the pre-allocated power of multiple energy storage units is determined based on the frequency regulation event type and risk index, including:

[0037] The proportion of responsibility for rapid energy storage infrastructure should be determined based on the type of frequency regulation event.

[0038] The target share of rapid energy storage is determined based on the risk index and the basic share of rapid energy storage.

[0039] The pre-allocated power of batteries and the pre-allocated power of fast energy storage are determined based on the target share of fast energy storage and the total future frequency regulation demand.

[0040] Preset constraints are determined and output based on SOC, SOH, temperature, rated power, remaining available energy, and ramp rate. The constraints include SOC boundary constraints, SOH boundary constraints, power boundary constraints, ramp rate boundary constraints, temperature rise boundary constraints, minimum hold time constraints, and compensation window constraints.

[0041] Furthermore, the pre-allocated power is corrected based on the preset constraints of multiple energy storage units to determine the final power command, including:

[0042] The dynamic available power range is determined based on the rated power of the energy storage unit, SOC boundary constraints, SOH boundary constraints, temperature rise boundary constraints, ramp rate boundary constraints, and remaining available energy.

[0043] The pre-allocated power is limited based on the dynamic available power range;

[0044] Determine the power deficit between the future total frequency regulation demand and the pre-allocated power after limiting;

[0045] Apply the ramp rate boundary constraints;

[0046] Implement minimum hold time constraints and hysteresis constraints;

[0047] Determine the final power of the battery and the final power of the fast energy storage that meet the constraints.

[0048] Furthermore, the method also includes:

[0049] When the frequency regulation event ends or enters the recovery phase, the fast energy storage unit is restored to its capacity and SOC return to center control is performed, and the battery energy storage unit undertakes continuous frequency regulation support.

[0050] Furthermore, upon the end of a frequency regulation event or the start of the recovery phase, the fast energy storage unit undergoes capacity restoration and SOC return-to-center control, including:

[0051] Determine whether the FM event has ended or entered the recovery phase;

[0052] When the SOC of a fast energy storage unit deviates from the median range, the SOC deviation is determined.

[0053] Determine the compensation power based on the SOC deviation;

[0054] The target power of the fast energy storage unit is determined based on the power replenishment and the frequency regulation command of the fast energy storage unit.

[0055] When the SOC of the fast energy storage unit returns to the median range, or after a new impulsive or oscillatory frequency regulation event occurs, the return-to-center control is discontinued.

[0056] Some embodiments of this application also provide a computer-readable medium having computer-readable instructions stored thereon, which can be executed by a processor to implement the aforementioned power allocation method for hybrid energy storage.

[0057] Some embodiments of this application also provide an electronic device, which includes a memory for storing computer program instructions and a processor for executing the computer program instructions, wherein when the computer program instructions are executed by the processor, the electronic device performs the aforementioned hybrid energy storage power distribution method.

[0058] Compared with existing technologies, the solution provided in this application can collect grid frequency change data and operating status data of multiple energy storage units, predict grid frequency regulation demand in the future time period based on the collected data, determine the frequency regulation event type based on the prediction results and event classification rules, determine the risk index of the energy storage unit based on the operating status data, prediction results, and frequency regulation event type, determine the pre-allocated power of the energy storage unit based on the frequency regulation event type and risk index, and correct the pre-allocated power based on the constraints of the energy storage unit to determine the final power command. This allows for early identification of the persistence, reversal, and recovery characteristics of frequency regulation events, improving the foresight of hybrid energy storage power allocation; prioritizing the use of fast energy storage under impact conditions, taking over battery energy storage for continuous support as early as possible under persistent deviation conditions, and suppressing frequent battery reversal under oscillation conditions to reduce the impact on energy storage equipment; quickly correcting the allocation results when the energy storage equipment reaches power, energy, or risk boundaries, improving frequency regulation tracking performance and control stability; and performing fast energy storage replenishment and capacity rebalancing control during the recovery phase to improve the available regulation capacity in subsequent frequency regulation events and reduce the performance degradation caused by long-term biased operation of fast energy storage. Attached Figure Description

[0059] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings.

[0060] Figure 1A flowchart illustrating a power allocation method for hybrid energy storage provided for some embodiments of this application.

[0061] Figure 2 This is a schematic diagram of power allocation at different evolution stages provided for some embodiments of this application. Detailed Implementation

[0062] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0063] Here, the power allocation method of hybrid energy storage in this application embodiment is suitable for scenarios such as primary frequency regulation, secondary frequency regulation and AGC auxiliary services, such as joint frequency regulation of thermal power and energy storage, frequency regulation of energy storage in new energy power plants, independent energy storage auxiliary services and microgrid frequency support.

[0064] In this scenario, the grid frequency will experience sudden, high-frequency, and continuous fluctuations. In order to bring the grid frequency back to normal, it is necessary to adjust the grid frequency by charging and discharging energy storage devices. Currently, hybrid energy storage systems are often used to support primary frequency regulation and secondary frequency regulation. However, how to allocate power between battery energy storage and fast energy storage in hybrid energy storage systems is still a problem worth studying.

[0065] The hybrid energy storage power allocation method provided in this application can collect grid frequency change data and operating status data of multiple energy storage units, predict grid frequency regulation demand in the future time period based on the collected data, determine the frequency regulation event type based on the prediction results and event classification rules, determine the risk index of the energy storage unit based on the operating status data, prediction results, and frequency regulation event type, determine the pre-allocated power of the energy storage unit based on the frequency regulation event type and risk index, and correct the pre-allocated power based on the constraints of the energy storage unit to determine the final power command. This allows for early identification of the persistence, reversal, and recovery characteristics of frequency regulation events, improving the foresight of hybrid energy storage power allocation. Under impact conditions, fast energy storage is prioritized; under continuous deviation conditions, battery energy storage takes over early and provides continuous support; under oscillation conditions, frequent battery reversal is suppressed, reducing the impact on energy storage equipment; when energy storage equipment reaches power, energy, or risk boundaries, the allocation results can be quickly corrected, improving frequency regulation tracking performance and control stability; during the recovery phase, fast energy storage replenishment and capacity rebalancing control are performed, improving the available regulation capacity in subsequent frequency regulation events and reducing the performance degradation caused by long-term biased operation of fast energy storage.

[0066] Figure 1 The following are flowcharts illustrating a power distribution method for hybrid energy storage executed via electronic devices in some embodiments of this application. The electronic devices are the executing entities of this method, such as... Figure 1 As shown, the method may include the following steps:

[0067] Step S101: Collect grid frequency change data and operating status data of multiple energy storage units according to a preset sampling period.

[0068] Here, grid frequency variation data may include, but is not limited to: grid frequency deviation, frequency variation rate, and automatic generation control (AGC) command trajectory.

[0069] The AGC command trajectory is a sequence of frequency regulation power commands generated by the automatic generation control system within a continuous time window. It includes not only the current AGC command value but also the changes in AGC commands at several historical sampling points, such as the rise, fall, maintenance, reversal, and ramp rate. For example, assuming a sampling period of 1 second, the AGC command sequence within a short window might be: 0MW→2MW→5MW→8MW→10MW→10MW→8MW→5MW. This sequence constitutes the AGC command trajectory. From this trajectory, it can be determined that frequency regulation demand initially rises rapidly, then maintains briefly, and then enters a decline and recovery process. In event evolution prediction, this trajectory, along with frequency deviation and frequency change rate, can be combined to determine the persistence, reversal trend, and recovery process of frequency regulation demand within the future short window.

[0070] In some embodiments of this application, the energy storage unit includes a battery energy storage unit and a fast energy storage unit. The fast energy storage unit can be a supercapacitor energy storage unit, a flywheel energy storage unit, a flow battery energy storage unit, or any combination of the above energy storage units.

[0071] The operating status data of the energy storage unit may include, but are not limited to: SOC, SOH, terminal voltage, current, temperature, internal resistance, available charging and discharging power, and remaining available energy.

[0072] Operating status data such as SOC, terminal voltage, current, temperature, available charge / discharge power, and remaining available energy can be used to describe battery energy storage units and fast energy storage units. For flywheel energy storage units, SOC can be understood as the equivalent state of charge calculated based on flywheel speed or kinetic energy; for supercapacitor energy storage units, SOC can be understood as the equivalent state of charge calculated based on terminal voltage or stored energy; for flow battery energy storage units, SOC can be understood as the equivalent state of charge calculated based on terminal voltage or stored energy. Parameters such as SOH and internal resistance are mainly used for battery energy storage units and flow battery energy storage units; for supercapacitor energy storage units or flywheel energy storage units, they can be replaced by equivalent state of health, equivalent internal resistance, speed margin, energy margin, or efficiency degradation indicators, respectively.

[0073] Data acquisition is conducted periodically, with sampling periods that can be fixed or multi-rate. For example, rapidly changing quantities such as frequency deviation, rate of frequency change, terminal voltage, current, and power can use sampling periods of 20ms to 200ms; slowly changing quantities such as SOC, SOH, temperature, internal resistance, and remaining available energy can use sampling periods of 0.5s to 5s. The station-level controller can synchronize data from different sampling periods to a unified control period, such as 100ms, 200ms, or 1s, for subsequent event evolution prediction, risk assessment, and power allocation.

[0074] Step S102: Based on the collected power grid frequency change data and operating status data, predict the power grid frequency regulation demand within a preset time period in the future.

[0075] In some embodiments of this application, predicting the power grid frequency regulation demand within a preset future time period based on collected power grid frequency change data and operating status data may include the following steps:

[0076] 1) Construct historical sliding window data based on power grid frequency change data and operating status data.

[0077] Let the current time be k, the sampling period be Ts, the historical window length be N, the prediction window length be M, and the multivariate time series input data consisting of power grid frequency change data and operating status data be X(k).

[0078] In some embodiments of this application, X(k) = {Δf(kn), df / dt(kn), P_AGC(kn), SOC_b(kn), SOC_f(kn), P_b(kn), P_f(kn), T_b(kn), T_f(kn)}, n = 0, 1, ..., N-1.

[0079] Where Δf is the frequency deviation, df / dt is the frequency change rate, P_AGC is the AGC command, SOC_b is the battery SOC, SOC_f is the fast energy storage SOC, P_b is the available charge / discharge power of the battery energy storage unit, P_f is the available charge / discharge power of the fast energy storage unit, T_b is the battery energy storage unit temperature, and T_f is the fast energy storage unit temperature.

[0080] 2) Calculate the total frequency regulation demand power sequence data based on the power grid frequency change data.

[0081] It can be expressed by the formula: P_req(k) = K_f·Δf(k) + K_r·df / dt(k) + K_a·P_AGC(k).

[0082] Where P_req(k) is the total frequency regulation demand power, K_f is the frequency deviation adjustment coefficient, K_r is the frequency change rate adjustment coefficient, and K_a is the AGC command weighting coefficient. K_f, K_r, and K_a can be preset based on practical experience.

[0083] 3) Based on historical sliding window data and total frequency regulation demand power sequence data, predict future total frequency regulation demand sequence data using a preset time series prediction algorithm.

[0084] Here, the time series forecasting algorithm can be any algorithm that can predict the input sequence data and output the sequence data, such as short-window trend extrapolation, state-space model, recursive least squares model, Kalman filter model, autoregressive exogenous input model or data-driven time series model, etc.

[0085] In some embodiments of this application, the future total frequency regulation demand sequence data is predicted by an autoregressive exogenous input model, which can be expressed by the formula: P_hat_req(k+j)=a0+Σa_i·P_req(ki)+Σb_i·Δf(ki)+Σc_i·P_AGC(ki)+Σd_i·r_AGC(ki), j=1,2,...,M.

[0086] Wherein, P_hat_req(k+j) is the predicted total frequency modulation demand power at the j-th sampling time in the future, P_req(ki) is the total frequency modulation demand power at the ith sampling time before the current time, Δf(ki) is the frequency deviation at the ith sampling time before the current time, P_AGC(ki) is the AGC command at the ith sampling time before the current time, r_AGC(ki) is the AGC command ramp rate at the ith sampling time before the current time, and a0, a_i, b_i, c_i, d_i are model parameters, which can be predetermined through historical frequency modulation data or updated in real time.

[0087] 4) Extract event evolution feature data from future total frequency regulation demand sequence data.

[0088] Here, the extracted event evolution feature data may include, but is not limited to: predicted maximum amplitude A_pred, symbol duration T_sign, number of symbol inversions N_rev, symbol inversion density ρ_rev, predicted remaining duration T_rem, and recovery phase slope S_rec, etc.

[0089] The maximum predicted amplitude is expressed by the formula: A_pred = max|P_hat_req(k+j)|. The sign inversion density is expressed by the formula: ρ_rev = N_rev / (M·Ts).

[0090] 5) Output the future total frequency regulation demand sequence data and event evolution characteristic data as the prediction results.

[0091] Here, the future total frequency regulation demand sequence data and event evolution characteristic data, such as the predicted maximum amplitude A_pred, symbol duration T_sign, number of symbol inversions N_rev, symbol inversion density ρ_rev, predicted remaining duration T_rem, and recovery phase slope S_rec, are output as prediction results for use in subsequent event classification and risk assessment processes.

[0092] By using grid frequency change data and operating status data to predict future grid frequency regulation demand, instead of allocating power based on the current frequency difference or current power, the evolution information of future short-term window frequency regulation events is introduced into the control link. That is, the persistence, reversal trend and recovery process of frequency regulation events are used as the pre-input of power allocation, thereby improving the foresight of the division of responsibilities between fast energy storage and slow energy storage such as batteries.

[0093] Step S103: Determine the frequency modulation event type based on the prediction results and preset event classification rules.

[0094] In some embodiments of this application, the prediction results may include, but are not limited to: future total frequency regulation demand power sequence data, predicted maximum amplitude, symbol duration, number of symbol inversions, symbol inversion density, predicted remaining duration, recovery phase slope, and whether the predicted end power has entered the dead zone.

[0095] In some embodiments of this application, the frequency modulation event types include impulsive events, persistent deviation events, oscillating events, and recovery events. Impulsive events correspond to large, short-duration disturbances; persistent deviation events correspond to disturbances with a single symbol lasting for a relatively long time; oscillating events correspond to disturbances with frequent reversals of power or frequency difference signs; recovery events correspond to the stage where the hybrid energy storage system gradually returns to equilibrium and is suitable for performing replenishment and rebalancing control, manifested as a continuous decrease in predicted power amplitude, a recovery segment slope of less than zero, and predicted end power entering the dead zone.

[0096] Here, classifying frequency modulation events into various types provides semantic labels for subsequent power template switching and risk weighting. Specifically, it transforms the predicted evolution characteristics of frequency modulation events into control labels that can be directly used in the subsequent power allocation process. Control labels refer to the correspondence between frequency modulation event types and power allocation strategies, risk correction methods, and constraint enforcement methods.

[0097] For example, in impact events, frequency regulation demand is typically characterized by large amplitude and short duration. Therefore, it is advisable to prioritize increasing the proportion of fast energy storage units to absorb or release instantaneous power quickly. In continuous deviation events, frequency regulation demand lasts for a long time in the same direction. Therefore, battery energy storage units provide the main continuous support to prevent the energy of fast energy storage units from being depleted prematurely. In oscillation events, frequency regulation demand changes frequently in both positive and negative directions. Therefore, fast energy storage units are prioritized to handle the round-trip power, and frequent reversals of battery energy storage units are suppressed through minimum hold time and hysteresis constraints. In recovery events, frequency regulation demand gradually weakens and approaches the dead zone. Therefore, it is suitable to activate the SOC return-to-center control and capacity recovery control of fast energy storage units.

[0098] In some embodiments of this application, the data used to determine the type of frequency modulation event can be calculated directly from the collected data, or it can be generated by further combining the prediction results. For example, it can be one or more of the following data: initial impact amplitude of the event, rate of frequency change, symbol duration, symbol inversion density, power fluctuation intensity, predicted remaining duration, and recovery segment slope, etc.

[0099] In some embodiments of this application, the frequency modulation event type can be determined as an impulse event when the predicted maximum amplitude is greater than a preset impact threshold, the predicted remaining duration is less than a preset remaining duration threshold, and the sign inversion density is less than a preset sign density threshold. For example, when A_pred≥0.25P_N, T_rem≤3s, and ρ_rev≤0.2 times / s, the frequency modulation event type is determined as an impulse event, where P_N is the rated power of the hybrid energy storage unit, T_rem is the predicted remaining duration, and ρ_rev is the sign inversion density.

[0100] In some embodiments of this application, when the symbol duration is greater than a preset symbol duration threshold and the symbol inversion density is less than a preset symbol density threshold in the prediction results, the frequency modulation event type is determined to be a persistent deviation event. For example, when T_sign ≥ 5s and ρ_rev ≤ 0.1 times / s, the frequency modulation event type is determined to be a persistent deviation event, where T_sign is the symbol duration.

[0101] In some embodiments of this application, when the number of symbol inversions in the prediction result is greater than a preset symbol inversion threshold or the symbol inversion density is greater than a preset symbol density threshold, the frequency modulation event type is determined to be an oscillating event. For example, when the number of symbol inversions N_rev ≥ 3 or ρ_rev ≥ 0.3 times / s within the 10s prediction window, the frequency modulation event type is determined to be an oscillating event.

[0102] In some embodiments of this application, when the slope of the recovery phase in the prediction result is less than zero and the predicted end power enters the dead zone, the frequency modulation event type is determined to be a recovery event. For example, when S_rec<0 and |P_hat_req(k+M)|≤0.1P_N, the frequency modulation event type is determined to be a recovery event, where S_rec is the slope of the recovery segment.

[0103] In addition, the relevant thresholds for determining the frequency regulation event types mentioned above can be adjusted according to the power grid frequency regulation assessment requirements, energy storage rated power, and control cycle.

[0104] Step S104: Determine the risk index of multiple energy storage units based on operating status data, prediction results, and frequency regulation event types.

[0105] Here, the risk indices can include battery life risk index, rapid energy storage energy depletion risk index, and mode switching jitter risk index. The battery life risk index describes the risk of battery energy storage units experiencing lifespan degradation due to high-rate operation, current surges, frequent charge-discharge reversals, temperature rise, and insufficient SOC margin. The rapid energy storage energy depletion risk index describes the risk of reduced rapid regulation capability in supercapacitor energy storage units, flywheel energy storage units, or flow battery energy storage units due to SOC deviating from the median, continuous high power output, or insufficient remaining energy. The mode switching jitter risk index describes the risk of frequent switching of energy storage units, control mode jitter, and command instability caused by frequent reversals in frequency regulation demand.

[0106] The data sources used to determine the risk index can include two parts. One part is the collected operating status data of energy storage units, such as SOC, SOH, temperature, terminal voltage, current, internal resistance, available charge and discharge power, and remaining available energy of battery energy storage units and fast energy storage units. The other part is the event evolution prediction and event classification results, such as the predicted remaining duration, sign inversion density, power fluctuation intensity, event type, and recovery phase slope.

[0107] In some embodiments of this application, the risk index can be normalized to 0 to 1, with a larger value indicating a higher risk.

[0108] In some embodiments of this application, the battery life risk index can be determined based on the ratio of the available charge / discharge power or current of the battery energy storage unit to a preset rated value, the current change amplitude at adjacent sampling times, the number of power reversals, the temperature rise trend, the internal resistance change trend, and the SOC safety margin. This can be expressed by the following formula:

[0109] R_b=w1·C_b+w2·J_I+w3·N_rev,b+w4·G_T+w5·G_R+w6·(1-M_SOC,b)

[0110] Where R_b is the battery life risk index, C_b is the battery rate, J_I is the current surge, N_rev,b is the number of battery power reversals, G_T is the temperature rise trend, G_R is the internal resistance change trend, M_SOC,b is the SOC safety margin of the battery energy storage unit, and w1 to w6 are weighting coefficients.

[0111] Battery capacity C_b represents the current charging / discharging power or current of the battery energy storage unit relative to its rated capacity; current surge J_I represents the magnitude of current change in the battery energy storage unit at adjacent sampling times; battery power reversal number N_rev,b represents the number of times the battery energy storage unit switches from charging to discharging or from discharging to charging; temperature rise trend G_T represents the rate of temperature rise of the battery energy storage unit within a short time window; internal resistance change trend G_R represents the degree of increase or abnormal change in the internal resistance of the battery energy storage unit; and the SOC safety margin M_SOC,b represents the distance between the current SOC and the safety boundary.

[0112] M_SOC,b can be determined as follows: When the battery SOC is between the lower and upper limits, calculate the normalized minimum distance from the upper and lower boundaries. As the battery SOC approaches either boundary, M_SOC,b gradually decreases. When the battery SOC reaches or exceeds the boundary, M_SOC,b can be 0. 1-M_SOC,b in the formula represents the risk term arising from insufficient SOC margin.

[0113] In some embodiments of this application, the energy depletion risk index of fast energy storage can be determined based on the degree of deviation of the SOC from the median of the fast energy storage unit, the instantaneous power ratio, the predicted remaining duration, the set maximum duration, and the remaining available energy margin. This can be expressed by the following formula:

[0114] R_f=v1·D_SOC,f+v2·α_f+v3·T_rem / T_max+v4·(1-M_E,f);

[0115] Where R_f is the energy depletion risk index of fast energy storage, D_SOC,f is the degree of deviation of the SOC of the fast energy storage unit from the median, α_f is the instantaneous power ratio of the fast energy storage unit, T_rem is the predicted remaining duration, T_max is the set maximum duration, M_E,f is the remaining available energy margin of the fast energy storage unit, and v1 to v4 are weighting coefficients.

[0116] D_SOC,f can be expressed by the formula: D_SOC,f=|SOC_f-SOC_mid| / SOC_range, where SOC_f represents the current SOC or equivalent SOC of the fast energy storage unit, SOC_mid represents the target median SOC of the fast energy storage unit, and SOC_range represents the normalized range of deviations from the median allowed for the fast energy storage unit.

[0117] For supercapacitor energy storage units, SOC_f can be calculated based on terminal voltage or stored energy. For flywheel energy storage units, SOC_f can be calculated based on flywheel speed or kinetic energy. For flow battery energy storage units, SOC_f can be calculated based on terminal voltage, electrolyte state, or stored energy.

[0118] SOC_mid can be set to 50%, or it can be a target value within the range of 45% to 55% depending on the actual equipment characteristics. When a fast energy storage unit is maintained near SOC_mid, it typically has both upward and downward rapid adjustment capabilities.

[0119] SOC_range can be determined based on the upper limit of SOC, the lower limit of SOC, and SOC_mid. For example, SOC_range can be the distance between SOC_mid and the lower limit of SOC, the distance between the upper limit of SOC and SOC_mid, or a smaller value of the two. The SOC_range parameter can normalize the SOC deviation into a dimensionless indicator that facilitates risk assessment.

[0120] In some embodiments of this application, the mode switching chattering risk index can be determined based on the sign inversion density, actual power switching frequency, and hold-up margin in the prediction results corresponding to the energy storage unit. This can be expressed by the following formula:

[0121] R_s=q1·ρ_rev+q2·F_sw+q3·(1-M_hold);

[0122] Where R_s is the mode switching jitter risk index, ρ_rev is the sign inversion density, F_sw is the actual power switching frequency, M_hold is the hold time margin, and q1~q3 are weighting coefficients.

[0123] The actual power switching frequency F_sw describes how frequently the energy storage unit switches its power direction or control mode within a preset historical window. This parameter can be calculated based on the actual power command or actual output power of the energy storage unit. For example, within a sliding window of length T_w, if the number of times N_sw occurs when the power sign of the energy storage unit changes from positive to negative, from negative to positive, or from charging mode to discharging mode, or from discharging mode to charging mode, is counted, then the actual power switching frequency can be expressed as F_sw = N_sw / T_w. To facilitate consistency with other risk items, F_sw can also be normalized by dividing it by the preset maximum allowable switching frequency.

[0124] The hold-time margin M_hold describes whether the time the energy storage unit has maintained its current control mode or power direction meets the minimum hold-time requirement. For example, assuming the time since the last power direction switch or control mode switch is T_hold, and the preset minimum hold-time is T_hold,min, then the hold-time margin can be expressed as M_hold = clip(T_hold / T_hold,min,0,1). When T_hold is less than T_hold,min, M_hold is less than 1, indicating that frequent switching is still not advisable; when T_hold reaches or exceeds T_hold,min, M_hold is 1, indicating that the hold-time requirement for switching has been met.

[0125] In the mode switching jitter risk index, the larger F_sw is, the more frequent the actual switching is and the higher the jitter risk; the smaller M_hold is, the less the holding time is and the higher the jitter risk. Therefore, 1-M_hold is used as the risk increase term in the formula.

[0126] By determining a risk index, the risks faced by energy storage units are quantified into intermediate variables that can be used for power allocation correction.

[0127] Step S105: Determine the pre-allocated power of multiple energy storage units based on the frequency regulation event type and risk index.

[0128] Here, when an event is identified as an impact-type or oscillation-type event, the proportion of the fast energy storage unit can be increased to reduce the short-term high-rate following of the battery energy storage unit; when an event is identified as a continuous deviation-type event, the proportion of the battery energy storage unit can be increased to enable the fast energy storage unit to provide edge support and rapid compensation; when an event is identified as a recovery-type event, the recovery start point and recovery intensity can be determined based on the state of charge offset of the fast energy storage unit and the predicted remaining duration.

[0129] Specifically, pre-allocated power and dynamic control boundaries can be generated through a combination of "basic allocation ratio for event types + risk index correction + dynamic boundary constraints".

[0130] In some embodiments of this application, determining the pre-allocated power of multiple energy storage units based on the frequency regulation event type and risk index may include the following steps:

[0131] 1) Determine the proportion α0 of rapid energy storage infrastructure to be undertaken based on the type of frequency regulation event.

[0132] For example, α0 can be taken as 0.7 to 0.9 in impact events, α0 can be taken as 0.2 to 0.4 in continuous deviation events, α0 can be taken as 0.6 to 0.8 in oscillation events, and α0 is determined according to the SOC return requirement of the fast energy storage unit in recovery events.

[0133] 2) Determine the target share of rapid energy storage based on the risk index and the basic share of rapid energy storage.

[0134] It can be expressed by the formula: α=clip(α0+λ_b·R_b-λ_f·R_f+λ_s·R_s, α_min, α_max).

[0135] Where α is the proportion of the target to be undertaken by fast energy storage, R_b is the battery life risk index, R_f is the energy depletion risk index of fast energy storage, R_s is the mode switching jitter risk index, α_min is the lower limit of the proportion of the target to be undertaken by fast energy storage, α_max is the upper limit of the proportion of the target to be undertaken by fast energy storage, clip() is the clipping function, and λ_b, λ_f, and λ_s are weighting coefficients.

[0136] When the battery life risk index is high, the proportion of power borne by the fast energy storage unit should be appropriately increased; when the energy depletion risk index of the fast energy storage unit is high, the proportion of power borne by the fast energy storage unit should be reduced; when the mode switching jitter risk index is high, the fast energy storage unit should be given priority to bear the round-trip power in oscillating events to reduce the frequent reversal of the battery energy storage unit.

[0137] 3) Determine the pre-allocated power of batteries and the pre-allocated power of fast energy storage based on the target proportion of fast energy storage and the total future frequency regulation demand.

[0138] It can be expressed by the formula: P_f_pre=α·P_req, P_b_pre=(1-α)·P_req.

[0139] Where P_f_pre is the pre-allocated power for fast energy storage, P_b_pre is the pre-allocated power for the battery, and P_req is the total frequency regulation demand power.

[0140] 4) Determine and output the preset constraints based on SOC, SOH, temperature, rated power, remaining available energy and ramp rate.

[0141] Here, the dynamic control boundary is updated in real time based on SOC, SOH, temperature, rated power, remaining available energy, and ramp rate, forming the dynamic power upper limit, dynamic power lower limit, dynamic dead zone, dynamic hold time, and dynamic recovery start time. The obtained dynamic control boundary is output as a preset constraint condition.

[0142] Here, the dynamic control boundary is a control constraint that is updated in real time based on the current SOC, SOH, temperature, rated power, remaining available energy and ramp rate of the energy storage unit. It is used to correct the pre-allocated power to the final power command that can be executed in engineering.

[0143] The dynamic power limit represents the maximum discharge power that the energy storage unit is allowed to output at the current moment. This limit is restricted by factors such as rated power, lower SOC limit, remaining available energy, SOH, temperature, and ramp rate. When the energy storage unit's SOC is low, temperature is high, or SOH is low, the dynamic power limit will decrease accordingly.

[0144] The dynamic power lower limit represents the maximum charging power that an energy storage unit is allowed to absorb at any given moment, and is usually expressed as a negative value. This lower limit is limited by factors such as rated charging power, state of charge (SOC) upper limit, rechargeable energy capacity, temperature, and ramp rate. When the SOC or temperature of the energy storage unit is too high, the absolute value of the dynamic power lower limit will decrease accordingly.

[0145] Dynamic dead zone is used to represent the range within which small power fluctuations or minute frequency differences do not trigger frequent operations of the energy storage unit. By setting a dynamic dead zone, the energy storage unit can be prevented from responding frequently to noise, minor disturbances, or invalid frequency modulation demands, thereby reducing power jitter and equipment losses.

[0146] Dynamic hold-up time represents the minimum time an energy storage unit needs to remain in a specific power direction or control mode. This parameter can be adjusted in real time based on the type of frequency regulation event and the risk index. For example, in oscillating events, the dynamic hold-up time of the battery energy storage unit can be appropriately increased to reduce frequent charge-discharge reversals; in impact events, the hold-up time of the fast energy storage unit can be appropriately shortened to improve the fast energy storage's response to transient power.

[0147] The dynamic compensation start time indicates the moment when the fast energy storage unit begins SOC return-to-center control after the frequency regulation event ends or enters the recovery phase. This time is determined based on the predicted remaining duration, the recovery phase slope, whether the predicted end power enters the dead zone, and the degree of SOC deviation of the fast energy storage. By setting a dynamic compensation start time, premature compensation by the fast energy storage can be avoided when the frequency regulation demand is still strong, thereby reducing the conflict between compensation power and frequency regulation power.

[0148] In some embodiments of this application, the constraints may include, but are not limited to: SOC boundary constraints, SOH boundary constraints, power boundary constraints, ramp rate boundary constraints, temperature rise boundary constraints, minimum hold time constraints, and compensation window constraints.

[0149] For example, if the total frequency regulation demand power P_req=10MW, the frequency regulation event is classified as an impact event, the basic share of fast energy storage α0=0.8, and after risk index correction, the target share of fast energy storage α=0.75, then the pre-allocated power of fast energy storage P_f_pre=7.5MW, the pre-allocated power of battery P_b_pre=2.5MW, and if the current maximum available discharge power of fast energy storage calculated by dynamic control boundary is 7MW, then the fast energy storage power is limited to 7MW, and the remaining 0.5MW is transferred to batteries or other regulation resources.

[0150] By coupling frequency regulation event classification with risk index, fast energy storage units can be prioritized under impact conditions, battery energy storage units can take over and provide continuous support as early as possible under continuous deviation conditions, and frequent reversals of battery energy storage units can be suppressed under oscillation conditions, thereby reducing the impact on energy storage equipment.

[0151] Step S106: Correct the pre-allocated power according to the preset constraints of multiple energy storage units and determine the final power command.

[0152] Here, the final power command is a control command that can be executed in real time in engineering.

[0153] In some embodiments of this application, modifying the pre-allocated power and determining the final power command based on preset constraints of multiple energy storage units may include the following steps:

[0154] 1) Determine the dynamic available power range based on the rated power of the energy storage unit, SOC boundary constraints, SOH boundary constraints, temperature rise boundary constraints, ramp rate boundary constraints, and remaining available energy.

[0155] Here, the dynamic available power range can be calculated separately for each energy storage unit i. The dynamic available power range can include the upper limit of dynamic discharge power P_i_max,dyn(k) and the lower limit of dynamic charging power P_i_min,dyn(k). The following explanation uses the example of positive discharge power and negative charging power. The calculation process can include the following steps:

[0156] a) First, determine the base available power based on the rated power, SOH and temperature of the energy storage unit.

[0157] Let the rated discharge power of energy storage unit i be P_i_dis,N, the rated charging power be P_i_ch,N, the SOH derating factor be μ_SOH,i(k), and the temperature derating factor be μ_T,i(k). Then the basic available discharge power P_i_dis,0(k) and the basic available charging power P_i_ch,0(k) can be expressed by the following formulas:

[0158] P_i_dis,0(k)=P_i_dis,N·μ_SOH,i(k)·μ_T,i(k),

[0159] P_i_ch,0(k)=P_i_ch,N·μ_SOH,i(k)·μ_T,i(k),

[0160] Among them, μ_SOH,i(k) and μ_T,i(k) can be set according to the lower limit of SOH, the upper limit of temperature and the equipment operation requirements. When SOH is low or temperature is high, the corresponding derating factor decreases, thereby reducing the current allowable power.

[0161] b) Next, determine the energy-constrained power based on the SOC boundary and the remaining available energy.

[0162] Let the rated energy of energy storage unit i be E_i,N, the current SOC be SOC_i(k), the lower limit of SOC be SOC_i,min, the upper limit of SOC be SOC_i,max, and the time scale for energy constraint calculation be T_c. Then the energy constraint power in the discharge direction and the charging direction can be expressed as follows:

[0163] P_i_dis,E(k)=max{0,(SOC_i(k)-SOC_i,min)·E_i,N / T_c},

[0164] P_i_ch,E(k)=max{0,(SOC_i,max-SOC_i(k))·E_i,N / T_c},

[0165] Where P_i_dis,E(k) represents the power that can be released at present under the condition that it is not lower than the lower limit of SOC, and P_i_ch,E(k) represents the power that can be absorbed at present under the condition that it does not exceed the upper limit of SOC.

[0166] c) Next, determine the allowable power range within this control cycle based on the ramp rate constraint.

[0167] Let the power of energy storage unit i in the previous control cycle be P_i(k-1), the maximum ramp rate be R_i, and the sampling period or control period be Ts. Then the power change in this cycle should satisfy:

[0168] P_i(k)≤P_i(k-1)+R_i·Ts,

[0169] P_i(k)≥P_i(k-1)-R_i·Ts.

[0170] d) Finally, the rated power constraint, SOH constraint, temperature constraint, SOC energy constraint, and ramp rate constraint are combined to obtain the dynamic available power range:

[0171] P_i_max,dyn(k)=min{P_i_dis,0(k),P_i_dis,E(k),P_i(k-1)+R_i·Ts},

[0172] P_i_min,dyn(k)=max{-P_i_ch,0(k),-P_i_ch,E(k),P_i(k-1)-R_i·Ts}.

[0173] The resulting P_i_min,dyn(k) and P_i_max,dyn(k) are the dynamic power boundaries that energy storage unit i can execute within the current control cycle. Subsequent pre-allocated power needs to be limited and corrected within this range.

[0174] In some embodiments of this application, the power command of energy storage unit i in the kth control cycle must be within the dynamic available power range, which can be expressed by the formula: P_i_min,dyn(k)≤P_i(k)≤P_i_max,dyn(k).

[0175] Where i is the energy storage unit number, which can be a battery energy storage unit, a fast energy storage unit, or other energy storage units participating in frequency regulation; k is the current sampling time or the current control cycle; P_i(k) is the power command or the corrected execution power issued to energy storage unit i in the k-th control cycle; P_i_max,dyn(k) is the maximum discharge power allowed to be output by energy storage unit i at the current time, i.e., the dynamic power upper limit; P_i_min,dyn(k) is the lower limit value corresponding to the maximum charging power allowed to be absorbed by energy storage unit i at the current time, which is usually a negative value.

[0176] This constraint ensures that the final power command issued to each energy storage unit will not exceed the power, energy, temperature, SOH, and ramp rate boundaries allowed by the current equipment state.

[0177] 2) Limit the pre-allocated power based on the dynamic available power range.

[0178] This can be expressed by the formula: P_i_lim(k)=sat(P_i_pre(k), P_i_min,dyn(k), P_i_max,dyn(k)). Where P_i_pre(k) is the pre-allocated power of energy storage unit i calculated based on the frequency regulation event type and risk index, P_i_min,dyn(k) is the current allowed lower limit of dynamic power for energy storage unit i, P_i_max,dyn(k) is the current allowed upper limit of dynamic power for energy storage unit i, P_i_lim(k) is the power result after limiting, and sat() is the limiting function.

[0179] The function of the limiting function is as follows: when P_i_pre(k) is greater than P_i_max,dyn(k), P_i_lim(k) takes the value of P_i_max,dyn(k); when P_i_pre(k) is less than P_i_min,dyn(k), P_i_lim(k) takes the value of P_i_min,dyn(k); when P_i_pre(k) is between P_i_min,dyn(k) and P_i_max,dyn(k), P_i_lim(k) takes the value of P_i_pre(k).

[0180] This limiting process can convert the theoretical pre-allocation results into executable power commands that meet the current operating boundaries of the energy storage device.

[0181] 3) Determine the power deficit between the future total frequency regulation demand and the pre-allocated power after limiting.

[0182] It can be expressed by the formula: ΔP(k)=P_req(k)-ΣP_i_lim(k).

[0183] If ΔP(k) is not zero, then redistribution is carried out based on the remaining power margin and the risk index.

[0184] 4) Implement ramp rate boundary constraints.

[0185] It can be expressed by the formula: |P_i(k)-P_i(k-1)|≤R_i·Ts.

[0186] Where i is the energy storage unit number, k is the current control cycle, P_i(k) is the power command of energy storage unit i in the current control cycle, P_i(k-1) is the power command of energy storage unit i in the previous control cycle, R_i is the maximum ramp rate of energy storage unit i, that is, the maximum power allowed to change per unit time, in MW / s or kW / s, Ts is the sampling period or control period, and R_i·Ts is the maximum power change allowed to increase or decrease within a control cycle.

[0187] For example, when the maximum ramp rate of energy storage unit i is 5 MW / s and the control period Ts is 0.2 s, the maximum allowable power change of this energy storage unit within one control period is 1 MW. If the pre-allocated power requires a power change of 2 MW within one period, it needs to be corrected to a change of no more than 1 MW based on the ramp rate boundary constraint.

[0188] Therefore, the ramp rate boundary constraint can make the power command change smoothly, reduce current surges and control oscillations, and improve the stability of frequency regulation control and equipment safety of energy storage systems.

[0189] 5) Implement minimum hold time constraints and hysteresis constraints.

[0190] Here, the minimum hold-up time constraint is used to limit the frequent switching of energy storage units between control modes such as charging, discharging, or standby. Specifically, when an energy storage unit switches from charging to discharging, or from discharging to charging, it needs to maintain the new power direction or control mode for at least a preset time T_hold,min. If a reverse switching requirement occurs again before T_hold,min is reached, the original power direction can be maintained, the standby state can be maintained, or a smooth transition can be performed until the minimum hold-up time requirement is met.

[0191] In addition, when the frequency modulation event is an oscillating event, the minimum hold time of the battery energy storage unit can be increased to extend its power switching hysteresis; when the frequency modulation event is an impulsive event, the action hold time of the fast energy storage unit can be shortened to improve its ability to withstand transient impacts and pulse power.

[0192] Hysteresis constraints are used to set different entry and exit thresholds during power direction switching, mode switching, or SOC recovery start-up and shutdown to prevent the system from repeatedly operating near the thresholds. For example, for power response, an entry threshold P_on and an exit threshold P_off can be set, where P_on > P_off. When the frequency regulation demand exceeds P_on, the energy storage unit enters the response state, and exits the response state only when the frequency regulation demand falls below P_off. For SOC recovery control, a recovery start-up interval and a recovery stop interval can be set. For example, recovery can be started when the fast energy storage SOC is below 45% or above 55%, and recovery can be stopped when the SOC returns to around 50% or within a preset median range.

[0193] 6) Determine the final power of the battery and the final power of the fast energy storage that meet the constraints.

[0194] Determine the final battery power P_b_final and the final fast energy storage power P_f_final that satisfy the above constraints, and send P_b_final and P_f_final to the energy storage converter (PCS) of the battery energy storage unit and the PCS of the fast energy storage unit, respectively.

[0195] For example, assuming the total frequency regulation demand is 10MW, and the power pre-allocation result is 7.5MW for fast energy storage units and 2.5MW for battery energy storage units, real-time boundary constraint calculation results show that the maximum discharge power of the fast energy storage units is only 6.5MW due to low SOC, while the maximum increase in discharge power of the battery energy storage units is 3MW. The ramp-up rate constraint of the battery energy storage units allows an increase from 2.5MW to 3.5MW in this cycle. Therefore, the fast energy storage units need to be limited to 6.5MW first, resulting in a 1MW shortfall. This 1MW shortfall is made up by the battery energy storage units, adjusting their power from 2.5MW to 3.5MW. The final output power command for both the fast energy storage units and the battery energy storage units is 6.5MW, still meeting the total frequency regulation demand of 10MW. Furthermore, if the temperature, SOC, or SOH boundaries of the battery energy storage units do not allow for further power increases, the remaining shortfall is treated as a tracking residual or reallocated by the higher-level frequency regulation resources.

[0196] By introducing a real-time correction and redistribution mechanism, the allocation results can be quickly corrected when the energy storage device reaches the power, energy or risk boundary, thereby improving frequency regulation tracking performance and control stability.

[0197] Figure 2 This application illustrates power allocation at different evolutionary stages according to some embodiments, such as... Figure 2 As shown, different evolutionary stages correspond to different frequency regulation event types. In different frequency regulation event types, the total frequency regulation power demand (i.e., total frequency regulation power demand) is different, and the power undertaken by battery energy storage units and fast energy storage units are also different.

[0198] In some embodiments of this application, when the frequency modulation event ends or enters the recovery phase, the fast energy storage unit can also be restored in capacity and controlled to return to SOC. The battery energy storage unit can then provide continuous frequency modulation support, thereby improving the adjustability to subsequent events and reducing the performance degradation caused by long-term biased operation of the fast energy storage unit.

[0199] Here, the end of a frequency modulation event refers to the fact that the frequency deviation, frequency change rate, AGC command change, and total frequency modulation power demand all enter the set dead zone and remain there for a certain period of time. For example, if |Δf|≤Δf_dead, |P_req|≤P_dead, and |dP_AGC / dt|≤r_dead are satisfied, and this condition is maintained for 2s to 5s, the end of the frequency modulation event can be determined.

[0200] The frequency regulation event has entered the recovery phase. This means that the event evolution prediction results show that the future frequency regulation demand amplitude continues to decline, the prediction end enters the dead zone or approaches zero, and the slope of the recovery segment indicates that the hybrid energy storage system is returning to a balanced state. For example, if S_rec<0 and |P_hat_req(k+M)|≤P_dead, it can be determined that the frequency regulation event has entered the recovery phase.

[0201] Capacity recovery and SOC return-to-center control for fast energy storage units refers to restoring the SOC, equivalent energy state, or flywheel speed of the fast energy storage unit to near the median value after the frequency regulation demand weakens or the event ends, so that it regains its bidirectional fast regulation capability. For supercapacitor energy storage units and flow battery energy storage units, SOC return-to-center refers to restoring its voltage or equivalent SOC to 45% to 55%. For flywheel energy storage units, SOC return-to-center refers to restoring its speed or kinetic energy to the rated median range.

[0202] In some embodiments of this application, when the frequency modulation event ends or enters the recovery phase, the process of restoring the capacity of the fast energy storage unit and controlling the SOC return to neutron may include the following steps:

[0203] 1) Determine whether the FM event has ended or entered the recovery phase.

[0204] 2) Determine the SOC deviation when the SOC of the fast energy storage unit deviates from the median range.

[0205] The formula is: e_f = SOC_mid - SOC_f. Where e_f is the SOC deviation of the fast energy storage unit, SOC_mid is the median SOC of the fast energy storage unit, and SOC_f is the current SOC or equivalent SOC of the fast energy storage unit.

[0206] 3) Determine the compensation power based on the SOC deviation.

[0207] The formula is: P_rep = clip(K_soc·e_f, -P_rep_max, P_rep_max). Here, P_rep is the recovery power of the fast energy storage unit, used to gradually return the SOC or equivalent energy state of the fast energy storage unit to near the median value when the frequency regulation event ends or enters the recovery phase; K_soc is the SOC recovery coefficient, used to convert the SOC deviation into recovery power, and its magnitude can be set according to the rated power of the fast energy storage unit, the allowable recovery time, and the system frequency regulation tracking requirements; P_rep_max is the maximum allowable recovery power, used to limit the excessive impact of the recovery process on external frequency regulation power tracking; clip() is a limiting function used to limit K_soc·e_f between -P_rep_max and P_rep_max.

[0208] Here, P_rep_max can be determined based on the current chargeable / dischargeable power of the fast energy storage unit, the compensable power of the battery energy storage unit, the allowable tracking error of the hybrid energy storage system, and the current frequency regulation event type. When the SOC of the fast energy storage unit deviates significantly from the median, the compensation power increases accordingly; when the SOC approaches the median, the compensation power gradually decreases; when the compensation power exceeds the maximum allowable value, it is limited by the clip() function.

[0209] By using the above-mentioned power recovery calculation method, the fast energy storage unit can be gradually restored to a state suitable for bidirectional fast adjustment without significantly affecting the external frequency regulation power tracking of the hybrid energy storage system, thereby improving its ability to cope with subsequent impact-type or oscillating frequency regulation events.

[0210] 4) Determine the target power of the fast energy storage unit based on the power replenishment and the frequency regulation command of the fast energy storage unit.

[0211] The frequency regulation command of the fast energy storage unit is superimposed with the power replenishment to form the target power of the fast energy storage unit. At the same time, the battery energy storage unit provides continuous frequency regulation support and compensates for the power changes caused by the power replenishment of the fast energy storage unit, so that the total output of the hybrid energy storage system still meets the frequency regulation requirements.

[0212] 5) When the SOC of the fast energy storage unit returns to the median range or after a new impulsive or oscillatory frequency regulation event occurs, exit the return-to-center control.

[0213] For example, when the frequency regulation event enters the recovery phase, the total frequency regulation demand has dropped to 1MW, and the SOC of the fast energy storage unit is 35%, which is lower than the median range of 45% to 55%. At this time, the hybrid energy storage system generates 0.8MW of replenishment power to replenish the fast energy storage unit. At the same time, the battery energy storage unit undertakes the remaining frequency regulation support and replenishment compensation, so that the hybrid energy storage system still operates according to the 1MW frequency regulation demand. When the SOC of the fast energy storage unit rises to around 50%, the return-to-center control stops and the normal standby state is restored.

[0214] Some embodiments of this application also provide a computer-readable medium having computer-readable instructions stored thereon, which can be executed by a processor to implement the aforementioned power allocation method for hybrid energy storage.

[0215] Some embodiments of this application also provide an electronic device, which includes a memory for storing computer program instructions and a processor for executing the computer program instructions, wherein when the computer program instructions are executed by the processor, the electronic device performs the aforementioned hybrid energy storage power distribution method.

[0216] The power allocation scheme for hybrid energy storage provided in this application has the following advantages:

[0217] 1. Introduce the future evolution trend of frequency modulation events as an independent control input into the hybrid energy storage power distribution link to realize the transformation from "current state driven" to "event evolution driven".

[0218] 2. Construct frequency regulation event types such as impact type, continuous deviation type, oscillation type and recovery type to match the responsibilities of fast and slow energy storage units with the frequency regulation event types.

[0219] 3. Construct a battery life risk index, a rapid energy storage energy depletion risk index, and a mode switching jitter risk index, and use them as a unified correction quantity for power pre-allocation and real-time correction.

[0220] 4. A two-stage control architecture of "power pre-allocation + real-time correction + rebalancing" is adopted to ensure the feasibility of the project while taking into account frequency regulation performance, continuous support capability and energy storage equipment life.

[0221] 5. Flexible control is achieved through parameters such as dynamic power boundary, dynamic dead zone, dynamic hold time, and dynamic recovery time, thereby improving the adaptability under complex frequency modulation conditions.

[0222] In summary, the solution provided in this application can collect grid frequency change data and operating status data of multiple energy storage units, predict grid frequency regulation demand in the future time period based on the collected data, determine the frequency regulation event type based on the prediction results and event classification rules, determine the risk index of the energy storage unit based on the operating status data, prediction results, and frequency regulation event type, determine the pre-allocated power of the energy storage unit based on the frequency regulation event type and risk index, and correct the pre-allocated power based on the constraints of the energy storage unit to determine the final power command. This allows for early identification of the persistence, reversal, and recovery characteristics of frequency regulation events, improving the foresight of hybrid energy storage power allocation; prioritizing the use of fast energy storage under impact conditions, taking over battery energy storage for continuous support as early as possible under persistent deviation conditions, and suppressing frequent battery reversal under oscillation conditions to reduce the impact on energy storage equipment; quickly correcting the allocation results when the energy storage equipment reaches power, energy, or risk boundaries, improving frequency regulation tracking performance and control stability; and performing fast energy storage replenishment and capacity rebalancing control during the recovery phase to improve the available regulation capacity in subsequent frequency regulation events and reduce the performance degradation caused by long-term biased operation of fast energy storage.

[0223] It should be noted that this application can be implemented in software and / or a combination of software and hardware, for example, using an application-specific integrated circuit (ASIC), a general-purpose computer, or any other similar hardware device. In one embodiment, the software program of this application can be executed by a processor to implement the steps or functions described above. Similarly, the software program of this application (including related data structures) can be stored in a computer-readable recording medium, such as RAM memory, a magnetic or optical drive, a floppy disk, or similar devices. Furthermore, some steps or functions of this application can be implemented in hardware, for example, as circuitry that cooperates with a processor to perform the various steps or functions.

[0224] In a typical configuration of this application, both the terminal and the network device include one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0225] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0226] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include non-transitory computer-readable media, such as modulated data signals and carrier waves.

[0227] Furthermore, a portion of this application can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to this application through the operation of the computer. The program instructions invoking the methods of this application may be stored in a fixed or removable recording medium, and / or transmitted via a data stream in a broadcast or other signal carrying medium, and / or stored in the working memory of a computer device operating according to the program instructions. Here, one embodiment of this application includes a device comprising a memory for storing computer program instructions and a processor for executing the program instructions, wherein, when the computer program instructions are executed by the processor, the device is triggered to run methods and / or technical solutions based on the foregoing embodiments of this application.

[0228] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from the spirit or essential characteristics of this application. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of this application is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within this application. No reference numerals in the claims should be construed as limiting the scope of the claims. Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices recited in the apparatus claims may also be implemented by a single unit or device in software or hardware. The terms "first," "second," etc., are used to indicate names and do not indicate any particular order.

Claims

1. A power distribution method for hybrid energy storage, characterized in that, The method includes: According to a preset sampling period, power grid frequency change data and operating status data of multiple energy storage units are collected. The power grid frequency change data includes power grid frequency deviation, frequency change rate and automatic generation control command trajectory. The operating status data includes SOC, SOH, terminal voltage, current, temperature, internal resistance, available charging and discharging power and remaining available energy. Based on the collected power grid frequency change data and operating status data, the power grid frequency regulation demand within a preset time period is predicted. Based on the prediction results and preset event classification rules, the frequency regulation event type is determined. The prediction results include future total frequency regulation demand power sequence data, predicted maximum amplitude, symbol duration, number of symbol inversions, symbol inversion density, predicted remaining duration, recovery phase slope, and whether the predicted end power enters the dead zone. The frequency regulation event types include impulsive events, persistent deviation events, oscillation events, and recovery events. Based on the operating status data, the prediction results, and the frequency modulation event type, a number of risk indices for the energy storage units are determined, wherein the risk indices include a battery life risk index, a rapid energy storage energy depletion risk index, and a mode switching vibration risk index. The pre-allocated power of multiple energy storage units is determined based on the frequency regulation event type and the risk index; The pre-allocated power is corrected and the final power command is determined based on the preset constraints of the multiple energy storage units. Based on the collected power grid frequency change data and operating status data, the power grid frequency regulation demand within a preset future time period is predicted, including: constructing historical sliding window data based on the power grid frequency change data and operating status data; calculating total frequency regulation demand power sequence data based on the power grid frequency change data; predicting future total frequency regulation demand sequence data using a preset time series prediction algorithm based on the historical sliding window data and the total frequency regulation demand power sequence data; extracting event evolution feature data from the future total frequency regulation demand sequence data; and outputting the future total frequency regulation demand sequence data and the event evolution feature data as the prediction result. Based on the prediction results and preset event classification rules, the frequency modulation event type is determined, including: when the predicted maximum amplitude is greater than a preset impact threshold, the predicted remaining duration is less than a preset remaining duration threshold, and the symbol inversion density is less than a preset symbol density threshold, the frequency modulation event type is determined as an impact event; when the predicted symbol duration is greater than a preset symbol duration threshold and the symbol inversion density is less than a preset symbol density threshold, the frequency modulation event type is determined as a persistent deviation event; when the predicted number of symbol inversions is greater than a preset symbol inversion threshold or the symbol inversion density is greater than a preset symbol density threshold, the frequency modulation event type is determined as an oscillation event; when the recovery phase slope is less than zero and the predicted end power enters the dead zone, the frequency modulation event type is determined as a recovery event. Based on the operating status data, the prediction results, and the frequency modulation event type, risk indices for multiple energy storage units are determined, including: determining a battery life risk index based on the ratio of the available charge / discharge power or current of the battery energy storage unit to a preset rated value, the current change amplitude at adjacent sampling times, the number of power reversals, the temperature rise trend, the internal resistance change trend, and the SOC safety margin, wherein the energy storage unit includes a battery energy storage unit and a fast energy storage unit; determining a fast energy storage energy depletion risk index based on the degree of SOC deviation from the median, the instantaneous power ratio, the predicted remaining duration, the set maximum duration, and the remaining available energy margin of the fast energy storage unit; and determining a mode switching jitter risk index based on the sign reversal density, the actual power switching frequency, and the hold-up duration margin in the prediction results corresponding to the energy storage unit. Determining the pre-allocated power of multiple energy storage units based on the frequency regulation event type and the risk index includes: determining the basic share of rapid energy storage based on the frequency regulation event type; determining the target share of rapid energy storage based on the risk index and the basic share of rapid energy storage; determining the pre-allocated power of the battery and the pre-allocated power of rapid energy storage based on the target share of rapid energy storage and the future total frequency regulation demand power; and determining and outputting preset constraints based on SOC, SOH, temperature, rated power, remaining available energy, and ramp rate, wherein the constraints include SOC boundary constraints, SOH boundary constraints, power boundary constraints, ramp rate boundary constraints, temperature rise boundary constraints, minimum hold time constraints, and compensation window constraints.

2. The method according to claim 1, characterized in that, The pre-allocated power is corrected and the final power command is determined based on preset constraints of multiple energy storage units, including: The dynamic available power range is determined based on the rated power of the energy storage unit, the SOC boundary constraints, the SOH boundary constraints, the temperature rise boundary constraints, the ramp rate boundary constraints, and the remaining available energy. The pre-allocated power is limited according to the dynamic available power range; Determine the power deficit between the total future frequency regulation demand and the pre-allocated power after limiting; Execute the aforementioned gradient rate boundary constraints; Implement minimum hold time constraints and hysteresis constraints; Determine the final power of the battery and the final power of the fast energy storage that meet the constraints.

3. The method according to claim 1, characterized in that, The method also includes: When the frequency regulation event ends or enters the recovery phase, the fast energy storage unit is restored to its capacity and controlled to return to SOC, and the battery energy storage unit provides continuous frequency regulation support.

4. The method according to claim 3, characterized in that, When the frequency regulation event ends or enters the recovery phase, the fast energy storage unit is subjected to capacity restoration and SOC return control, including: Determine whether the FM event has ended or entered the recovery phase; When the SOC of the fast energy storage unit deviates from the median range, the SOC deviation is determined; The compensation power is determined based on the SOC deviation. The target power of the fast energy storage unit is determined based on the replenishment power and the frequency modulation command of the fast energy storage unit. When the SOC of the fast energy storage unit returns to the median range, or after a new impulsive frequency modulation event or oscillatory frequency modulation event occurs, the return-to-center control is exited.

5. A computer-readable medium having stored thereon computer-readable instructions that can be executed by a processor to implement the method as described in any one of claims 1 to 4.

6. An electronic device comprising a memory for storing computer program instructions and a processor for executing the computer program instructions, wherein, When the computer program instructions are executed by the processor, the electronic device performs the method as described in any one of claims 1 to 4.

Images