Energy storage frequency modulation adaptive control method and system

Abstract

The invention relates to the technical field of self-adaptive control systems, and discloses an energy storage frequency modulation self-adaptive control method and system, and the method comprises the steps: collecting a power grid frequency signal in real time, carrying out the filtering and noise reduction processing, and dividing according to a preset time period to generate a historical frequency sequence; calculating an energy imbalance index based on the statistical characteristics of the frequency sequence of each time period, and dynamically generating a time-phased SOC reference value sequence; in the frequency dead zone, the actual SOC is made to return to the reference value of the current time period in a low-power fine tuning mode; and performing closed-loop correction on the next-day SOC reference value sequence according to the deviation between the frequency modulation theoretical response energy and the actual response energy. The problems of fixed SOC setting, response imbalance and limited frequency modulation capability in the prior art are solved, and the purposes of widening the frequency modulation range, improving the response precision and prolonging the service life of equipment are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of general control or regulation systems, in particular adaptive control systems. BACKGROUND

[0002] The stability of power grid frequency is one of the basic indicators for ensuring the safe and reliable operation of a power system, and directly reflects the balance between supply and demand of active power in the system. When the frequency deviates from the rated value, it may cause fluctuations in the power system, and even lead to large-scale power outages in severe cases. Therefore, frequency regulation resources with efficient and accurate regulation capabilities are crucial for power grid operation. In recent years, with the large-scale access of high-penetration renewable energy sources such as wind power and photovoltaic power, the system inertia has decreased significantly, resulting in more frequent and severe fluctuations in power grid frequency. Frequency regulation resources need to have higher response speed, regulation accuracy and flexibility. However, traditional frequency regulation methods mainly rely on thermal power units, which have slow response speed and limited regulation accuracy, making it difficult to meet the demand for fast dynamic regulation in new power systems.

[0003] Energy storage technology has gradually become an ideal auxiliary frequency regulation method due to its fast response and flexible control advantages. Lithium-ion batteries are widely used in practical engineering due to their high energy density and relatively low cost. However, in the context of high-frequency and short-time response frequency regulation, lithium batteries have safety hazards, accelerated degradation of battery cells, and low utilization, which restricts their application range. To overcome the above limitations, some high-power energy storage technologies (such as nickel-hydrogen batteries, supercapacitors, capacitor batteries, and flywheel energy storage) have shown great potential in primary frequency regulation and fast power support scenarios due to their fast response, high power density, high charging and discharging efficiency, and long cycle life. However, in actual operation, such high-power energy storage devices still face challenges such as low energy per unit, rapid changes in state of charge (SOC), and the like. In particular, in the case of long-term high or low power grid frequency, if the SOC reference value is not set reasonably, the energy storage system may enter the SOC protection zone and lose its regulation capability, thereby weakening the frequency response capability. The common frequency regulation control strategy currently uses a fixed SOC setting.

[0004] For example, the Chinese patent with publication number CN111049167B discloses a kind of energy storage system power control method and system, provide the following technical scheme, the present application provides a kind of energy storage system power control method and system, comprising: obtaining grid frequency voltage index, initial power scheduling instruction, the application scene that can participate in scheduling and each application scene priority;When the grid frequency voltage index is out of limit, the energy storage system is based on the application scene priority and the state of energy storage system and carries out power regulation;Otherwise, the energy storage system is based on the initial power scheduling instruction issued by grid and carries out power regulation, enhance the controllability of energy storage system and the flexible access of energy storage system, by the priority setting of application scene, multiple types multiple capacity energy storage battery is orderly in various operating scenarios of grid, enhance the anti-disturbance operating ability of grid.But the above-mentioned one kind of energy storage system power control method and system, by obtaining grid frequency and initial power instruction realizes power regulation, but not fully consider the statistical characteristics and future trend of frequency, prone to conservative SOC control strategy, thereby limit the frequency modulation capability of energy storage system.Therefore, it is urgent to propose a kind of adaptive control method that fuses grid frequency prediction and SOC dynamic regulation, can dynamically adjust SOC reference value according to frequency variation trend, widen adjustable power range, improve frequency modulation precision while taking into account the service life of energy storage equipment, further release the frequency modulation potential of high-power energy storage system.

[0005] Therefore, compared with the prior art, the present application is based on the first set of projects "frequency modulation device based on high-power nickel-hydrogen battery energy storage", which is to realize the first 6MW / 0.6MWh fire storage system in China. The core technology is 10C high-rate charging and discharging (breakthrough of domestic 5Ah / 5C bottleneck), 100ms millisecond level response and 600,000 times of shallow charge and discharge cycle life. SUMMARY

[0006] The present application solves the problems of SOC fixed setting, response imbalance and limited frequency modulation capability in the prior art, and proposes an adaptive control method and system for energy storage frequency modulation, achieving the purposes of widening frequency modulation range, improving response accuracy and prolonging equipment life.

[0007] Further, the present application is adapted to a method that can dynamically adjust the SOC reference under the frequency variation trend, solves the problem of mismatch between SOC trajectory and frequency variation trend in the prior art, to meet the demand of high-power energy storage system (such as nickel-hydrogen battery or super capacitor) for fast frequency modulation of power grid, realize the adaptive response of energy storage system to power grid frequency, improve the frequency modulation accuracy, system disturbance resistance and equipment service life.

[0008] To achieve the above-mentioned purposes, the present application adopts the following technical scheme: An adaptive control method for energy storage frequency modulation, comprising: Real-time acquisition of power grid frequency signal, filtering and noise reduction processing and dividing into historical frequency sequence according to preset time period; Based on the statistical characteristics of each period frequency sequence, the energy imbalance index is calculated, and the time-division SOC reference value sequence is dynamically generated; In the frequency dead zone interval, the actual SOC is returned to the current period reference value in a low-power fine-tuning manner; According to the deviation of the actual response energy from the response energy, the next day SOC reference value sequence is closed-loop corrected.

[0009] Breakthrough fixed SOC setting limit, realize adaptive management of energy storage SOC through dynamic reference value, dead zone fine-tuning and closed-loop correction, significantly improve the frequency regulation accuracy, response range and system reliability.

[0010] An adaptive control system for energy storage frequency regulation, comprising: A frequency sampling module connected to the frequency processing module, which acquires power grid frequency signals in real time; The frequency processing module performs filtering, smoothing and time period division operations; The SOC reference generation module receives data from the frequency processing module and dynamically outputs time-division SOC reference values to the SOC callback control module; The SOC callback control module triggers low-power SOC fine-tuning in the dead zone frequency interval; The power control module is connected to the SOC callback control module and executes frequency regulation power instructions and SOC callback instructions; The feedback evaluation module is connected to the power control module and analyzes the closed-loop optimization of the SOC reference value through the energy deviation.

[0011] Modular design has clear division of labor and clear data flow closed loop, and efficient cooperation realizes the whole adaptive control strategy, which is convenient for engineering deployment.

[0012] As preferred, the filtering and noise reduction processing and dividing into historical frequency sequence according to preset time period includes: using band-pass filtering and weighted moving average algorithm for noise reduction and smoothing processing, dividing the whole day frequency into N time periods for constructing historical frequency sequence, and extracting the dynamic change characteristics of frequency.

[0013] Improve the quality of frequency signal and identify the time period mode, lay the foundation for subsequent accurate analysis of power grid frequency dynamic characteristics and time-division SOC setting.

[0014] As preferred, the statistical features based on the frequency sequence of each period of time include: counting the number of low frequencies and the number of high frequencies and the number of dead zone frequencies in each period of time; wherein the number of low frequencies refers to the number of times when the frequency is less than the rated value minus the dead zone threshold, and the number of high frequencies refers to the number of times when the frequency is greater than the rated value plus the dead zone threshold; calculating the low-frequency cumulative action amount and the high-frequency cumulative action amount; wherein the low-frequency cumulative action amount is calculated by multiplying the integral value of the frequency negative offset by a first proportional coefficient, and the high-frequency cumulative action amount is calculated by multiplying the integral value of the frequency positive offset by a first proportional coefficient.

[0015] Quantifying the high and low frequency offset activity and the cumulative energy action amount overcomes the shortcomings of instantaneous offset judgment and provides key data for evaluating the energy imbalance of the power grid in the period.

[0016] As preferred, the energy imbalance index is specifically the difference between the low-frequency cumulative action amount and the high-frequency cumulative action amount divided by the sum of the two.

[0017] In the existing technology, the frequency offset is usually only judged by the instantaneous offset amplitude to determine the high / low frequency regulation demand, and the persistence and strength accumulation effect of the frequency offset in the whole period of time are ignored. The present application first proposes to construct the ratio of the high and low frequency energy storage charge and discharge power offset difference to the total power offset difference based on the frequency integral action amount, and define it as an energy imbalance index. This index takes into account the up and down frequency directionality, amplitude and cumulative continuous action amount, and can greatly reflect the SOC bias direction and cumulative offset amount of the energy storage system in the period of time, providing a quantitative basis for subsequent SOC reference value setting, and overcoming the problem that the traditional method cannot describe the accurate SOC in the dynamic process of different periods of time.

[0018] As preferred, the dynamically generated SOC reference value includes: mapping the energy imbalance index to the SOC offset amount, specifically the SOC offset amount is equal to the sensitivity coefficient multiplied by the energy imbalance index; wherein the typical value of the sensitivity coefficient is 0.1-0.3; and the time-division SOC reference value is the initial reference value plus the SOC offset amount, and the upper and lower limits are set so that it is within the range of 20%-80%.

[0019] In the traditional energy storage frequency modulation system, the SOC reference setting generally adopts a fixed median value (such as 50%) or is only based on experience adjustment. The present application maps the energy imbalance degree of the frequency behavior to the SOC reference offset amount through a continuous adjustable SOC dynamic correction mechanism, and introduces a sensitivity adjustment coefficient to realize flexible adjustment. This continuous mapping logic of the grid frequency behavior to the SOC reference supports the smooth evolution of the SOC reference value with the frequency offset trend, avoiding the limitations of experience value setting. At the same time, the set coefficient can also be seasonally self-adaptive, and has model generalization ability.

[0020] As preferred, the dynamically generated time-sharing period SOC reference value sequence further comprises constructing a dynamic correction model in combination with the past 7-day historical frequency trend, specifically: obtaining the historical average value of the energy imbalance index of the same period in the previous 6 days; multiplying the difference between the current period and the historical average value by the sensitivity coefficient to obtain the dynamic adjustment amount of the SOC reference value; superimposing the adjustment amount on the SOC reference value of the previous period to generate the corrected SOC reference value.

[0021] The application further introduces a mean sliding window mechanism of cross-period historical frequency behavior trend to realize dynamic correction of the SOC reference, which is equivalent to introducing a double structure of inertia term and correction term, and embeds trend memory in the SOC regulation process, avoids over-reaction of the system to occasional frequency abnormalities, and improves the stability and anti-disturbance ability of the overall SOC regulation.

[0022] As preferred, the low-power fine-tuning mode specifically comprises: calculating the absolute deviation amount of the real-time SOC and the reference value of the current period; multiplying the absolute deviation amount by a second proportional coefficient to generate a callback power; terminating the callback operation when the absolute deviation amount is less than or equal to a preset threshold; wherein the callback power is limited to 5% to 10% of the rated power, and the second proportional coefficient is set according to the ratio of the sum of high-frequency times and low-frequency times to the dead zone frequency times multiplied by the first proportional coefficient.

[0023] In the dead zone period, the SOC regression target value is safely and efficiently fine-tuned, and the callback speed is intelligently set according to the frequency adjustment activity.

[0024] As preferred, the closed-loop correction next-day SOC reference value sequence specifically comprises: after the daily frequency adjustment task is completed, the system re-evaluates the difference between the frequency data of each period and the actual response energy and the theoretical value, calculates the deviation and performs feedback correction with a third proportional coefficient to dynamically update the next-day SOC reference value, so that the control strategy forms a closed-loop iterative optimization.

[0025] The strategy forms a closed loop, continuously optimizes the accuracy of the SOC reference setting through deviation feedback, and relieves the accumulation of SOC deviation.

[0026] As preferred, the system re-evaluates the difference between the frequency data of each period and the actual response energy and the theoretical value, which includes: calculating the theoretical response energy and the actual response energy of each period, wherein the theoretical response energy is a first proportional coefficient multiplied by the frequency offset integral value, and the actual response energy is the integral value of the actual power with respect to time; the calculation of the deviation and the feedback correction with the third proportional coefficient includes: dividing the difference between the theoretical response energy and the actual response energy by the theoretical response energy to obtain an energy deviation coefficient, and multiplying the energy deviation coefficient by the third proportional coefficient to generate a correction amount of the SOC reference value, wherein the typical value of the third proportional coefficient is 1%-5%; the dynamic update of the next day SOC reference value is to superimpose the correction amount on the initial SOC reference value of the next day.

[0027] The energy deviation calculation and closed-loop correction algorithm are definitely defined, small step and progressive optimization are realized, and the closed-loop stability is ensured.

[0028] Compared with the prior art, the application has the beneficial effects that.

[0029] 1. The application generates an adaptive SOC reference value sequence by analyzing the dynamic characteristics of the power grid frequency in real time, and the energy imbalance index accurately quantifies the direction and degree of power grid demand, combined with the historical trend prediction correction mechanism, so that the energy storage system is always in the best response state. Significantly improve the frequency regulation accuracy and grid matching, effectively avoid SOC protective failure.

[0030] 2. The application adopts a low-power intelligent fine-tuning strategy during the frequency stabilization period, efficiently repairs the SOC deviation and optimizes the next round response preparation, establishes a daily closed-loop evaluation system, and continuously optimizes the SOC reference setting through the progressive feedback correction of the theoretical response and the actual energy deviation. This mechanism suppresses the SOC deviation accumulation problem from the source, balances the frequency regulation performance and device durability, and prolongs the service life of the high-power energy storage system.

[0031] 3. The modular system architecture of the application decomposes the complex control logic into six cooperative units, forming a complete data closed loop of sampling, processing, decision-making, execution, and feedback. The responsibilities of each module are clear and the interface is clear, especially suitable for embedded controller deployment. This design makes the dynamic SOC management strategy have strong engineering landing nature, and provides a reliable solution for the rapid frequency regulation demand of new power systems. BRIEF DESCRIPTION OF DRAWINGS

[0032] Figure 1 The method flowchart of an embodiment of the energy storage frequency regulation adaptive control method and system of the application.

[0033] Figure 2 The SOC reference value and feedback correction comparison chart of embodiment one of the energy storage frequency regulation adaptive control method and system of the application.

[0034] Figure 3 Figure 2 is a comparison chart of the SOC reference value of each time period and the feedback correction for the second embodiment of the energy storage frequency modulation adaptive control method and system of the present application. DETAILED DESCRIPTION

[0035] In order to make the purpose, technical solutions and advantages of the present disclosure clearer, the following will further describe the embodiments of the present disclosure in detail with reference to the drawings. The proportions of the components are not drawn according to the true proportions, and the proportions and sizes shown in the drawings should not limit the essential technical solutions of the present application. These embodiments do not describe all the details, and the present application is not limited to the specific embodiments described.

[0036] Referring to Figures 1-3 Figure 1 shows an energy storage frequency modulation adaptive control method, which comprises: Real-time acquisition of power grid frequency signals, filtering and noise reduction processing, and division into historical frequency sequences according to a preset time period; Based on the statistical characteristics of the frequency sequence of each time period, the energy imbalance index is calculated, and the time period SOC reference value sequence is dynamically generated; In the frequency dead zone interval, the actual SOC is returned to the current time period reference value in a low-power fine-tuning manner; According to the deviation of the actual response energy from the response energy according to the frequency modulation theory, the next day SOC reference value sequence is closed-loop corrected.

[0037] An energy storage frequency modulation adaptive control system comprises: A frequency sampling module connected to a frequency processing module for real-time acquisition of power grid frequency signals; The frequency processing module performs filtering, smoothing and time period division operations; The SOC reference generation module receives data from the frequency processing module and dynamically outputs the time period SOC reference value to the SOC callback control module; The SOC callback control module triggers low-power SOC fine-tuning in the dead zone frequency interval; The power control module is connected to the SOC callback control module and executes the frequency modulation power instruction and the SOC callback instruction; The feedback evaluation module is connected to the power control module and analyzes the closed-loop optimization of the SOC reference value through the energy deviation.

[0038] The application discloses a frequency modulation adaptive control method, device and system for a high-power energy storage system, extracts frequency characteristics in different time periods through real-time sampling and filtering, generates a dynamic SOC reference value, and combines a dead zone adjustment and an energy deviation feedback closed-loop correction mechanism to realize adaptive management of the energy storage system SOC and improvement of frequency modulation capability. The method overcomes the problem of response imbalance of a traditional SOC fixed setting, has adaptive, trend tracking and closed-loop adjustment capabilities in different time periods, is suitable for high-rate energy storage devices such as nickel-hydrogen batteries and supercapacitors, can significantly enhance the frequency response accuracy and grid frequency matching, and is suitable for fast frequency modulation application scenarios in new power systems.

[0039] As Figure 1 shown in an embodiment, Figure 1 is a method flowchart of an embodiment of the energy storage frequency modulation adaptive control method and system. First, a grid frequency signal is collected in real time, band-pass filtering and weighted sliding mean algorithm are used for noise reduction and smoothing processing, the frequency throughout the day is divided into N time periods in units of hours, a historical frequency sequence is constructed to extract the dynamic change characteristics of the frequency. Then, an energy imbalance index is calculated based on the statistical characteristics of the frequency sequence in each period, which includes the number of low frequencies, the number of high frequencies and the number of dead zone frequencies in each period (wherein the number of low frequencies refers to the number of times that the frequency is less than the rated value minus the dead zone threshold, and the number of high frequencies refers to the number of times that the frequency is greater than the rated value plus the dead zone threshold); the low-frequency cumulative action amount and the high-frequency cumulative action amount are calculated (wherein the low-frequency cumulative action amount is calculated by multiplying the integral value of the frequency negative deviation by a first proportional coefficient, and the high-frequency cumulative action amount is calculated by multiplying the integral value of the frequency positive deviation by a first proportional coefficient); the energy imbalance index is specifically the difference between the low-frequency cumulative action amount and the high-frequency cumulative action amount divided by the sum of the two.

[0040] Then, according to the energy imbalance index, a time-division SOC reference value sequence is dynamically generated. The specific process is to map the energy imbalance index to an SOC offset (the SOC offset is equal to the sensitivity coefficient multiplied by the energy imbalance index, and the typical value of the sensitivity coefficient is 0.1-0.3); the time-division SOC reference value is the initial reference value plus the SOC offset, and the upper and lower limits are set to be within the range of 20%-80%. The dynamic generation process also combines the past 7-day historical frequency trend to construct a dynamic correction model: the historical average value of the energy imbalance index of the same period in the previous 6 days is obtained; the difference between the current period index and the historical average value is multiplied by the sensitivity coefficient to obtain the dynamic adjustment amount of the SOC reference value; the adjustment amount is superimposed on the SOC reference value of the previous period to generate the corrected SOC reference value. Subsequently, within the frequency dead zone interval, a low-power fine-tuning method is used to make the actual SOC return to the current period reference value, which specifically includes calculating the absolute deviation amount of the real-time SOC and the current period reference value; the absolute deviation amount is multiplied by a second proportional coefficient to generate a callback power (the callback power is limited to the range of 5% to 10% of the rated power, and the second proportional coefficient is set according to the ratio of the sum of the high frequency times and the low frequency times to the dead zone frequency times multiplied by the first proportional coefficient); when the absolute deviation amount is less than or equal to a preset threshold, the callback operation is terminated.

[0041] Finally, according to the deviation of the frequency response energy and the actual response energy, the next day SOC reference value sequence is closed-loop corrected: after the daily frequency modulation task is completed, the system re-evaluates the frequency data of each period, calculates the difference between the theoretical response energy (the first proportional coefficient multiplied by the frequency offset integral value) and the actual response energy (the integral value of the actual power with respect to time) of each period, and divides the theoretical response energy to obtain the energy deviation coefficient; the energy deviation coefficient is multiplied by a third proportional coefficient (the typical value is 1%-5%) to generate a correction amount of the SOC reference value; the next day SOC reference value is dynamically updated by superimposing the correction amount on the initial SOC reference value of the next day, so that the entire control strategy forms a closed-loop iterative optimization.

[0042] In another embodiment, the present application discloses a high-power energy storage frequency modulation control method, device and system that combines power grid frequency characteristic analysis and energy storage state of charge (SOC) dynamic management strategy, aiming to solve the problem that the fixed or empirical SOC setting method is often used in existing high-power energy storage frequency modulation control, resulting in mismatch between the SOC trajectory of the energy storage system and the trend of the power grid frequency, thereby affecting the release of the frequency modulation capability of the energy storage device, shortening the service life of the system, and even causing the risk of SOC over-limit operation.

[0043] The method mainly includes the following four steps: S1, frequency sampling and processing: the real-time frequency of the power grid is sampled, and band-pass filtering (such as Butterworth filter) and weighted sliding mean algorithm are used for noise reduction and smoothing processing, and historical frequency sequence {X1, X2, …, X N} is constructed by dividing by hour, and the dynamic change characteristics of the frequency are extracted. The original frequency signal is processed using a band-pass filter to retain the effective frequency modulation band (function represented as bandpassfilter(f(t))); sliding average calculation is performed on the filtered data to weaken high-frequency jitter, and the formula is that the current time smoothing value is equal to the average value of the filtered values of the past m sampling points.

[0044] The historical frequency time sequence is {f(t-nΔt1), …, f(t)}, where Δt1 is the sampling period, preferably 50 ms, and n is the length of the historical sequence, preferably 24 hours in length.

[0045] The frequency of the whole day is divided into N time periods by hour, which facilitates frequency pattern recognition and parameter training by time period, and the whole day frequency sequence X can be represented as: {X1, X2, …, X N}.

[0046] S2, SOC reference adjustment: frequency offset statistics are performed on the frequency sequence of the divided time period, high and low frequency integral action is calculated, a frequency imbalance index is constructed, and a current time period SOC reference value is generated, and the whole day SOC reference sequence {SOC_base,1, SOC_base,2, …, SOC_base,N} is formed by analogy.

[0047] The frequency offset classification and statistics method is: analyze the frequency sequence of each time period, count the number of low frequencies Nlow (satisfying f(t)<f0-Δfdead), the number of high frequencies Nhigh (satisfying f(t)>f0+Δfdead), and the number of dead zone frequencies Ndead (satisfying f0-Δfdead≤f(t)≤f0+Δfdead).

[0048] The low frequency cumulative action is equal to the coefficient k0 multiplied by the cumulative sum of all low frequency deviations (f0-f(t)) multiplied by the sampling period Δt1 in this period; the high frequency integral action is equal to the coefficient k0 multiplied by the cumulative sum of all high frequency deviations (f(t)-f0) multiplied by the sampling period Δt1 in this period.

[0049] The energy imbalance index ηi is equal to the difference between the low-frequency cumulative action amount and the high-frequency cumulative action amount divided by the sum of the two. In the existing technology, the frequency deviation is usually only judged by the instantaneous deviation amplitude to determine the high / low frequency adjustment demand, and the accumulation effect of the continuity and intensity of the frequency deviation in the whole period is ignored. The present application first proposes to construct the ratio of the high and low frequency energy storage charging and discharging power deviation difference to the total power deviation difference based on the frequency integral action amount, which is defined as the energy imbalance index ηi. The index takes into account the up and down frequency directionality, amplitude and cumulative continuous action amount, and can greatly reflect the SOC bias direction and cumulative deviation of the energy storage system in the period, providing a quantitative basis for setting the subsequent SOC reference value, and overcoming the problem that the traditional method cannot accurately describe the SOC deviation dynamic process in different periods.

[0050] The time-segmented SOC reference value SOC_base,i is equal to the reference value SOC_base plus the deviation βi, where βi is the product of ksoc and ηi, and ksoc is the SOC bias sensitivity coefficient, and the typical value is 0.1-0.3. In the traditional energy storage frequency modulation system, the SOC reference setting generally adopts a fixed median value (such as 50%) or is only based on experience adjustment. The present application maps the energy imbalance degree (ηi) of the frequency behavior to the SOC reference deviation (βi) through a continuous adjustable SOC dynamic correction mechanism, and introduces a sensitivity adjustment coefficient ksoc to realize adjustment flexibility. This continuous mapping logic of grid frequency behavior to SOC reference supports the smooth evolution of the SOC reference value with the frequency deviation trend, avoiding the limitations of empirical value setting, and the set ksoc coefficient can also be seasonally self-adaptive, with model generalization ability.

[0051] To prevent over-adjustment of SOC, the upper and lower limits of SOC are set to SOCmin to SOCmax, and the typical value is 20%-80%. In the traditional SOC control, although there is often a way to set the upper and lower limits, the present application combines the dynamic SOC reference adjustment mechanism with the limit value strategy, so that the SOC reference adjustment process is always limited to the battery health working interval, ensuring the frequency modulation response performance while maintaining the life of the battery cell.

[0052] A dynamic correction model is constructed by combining the frequency trend of the past 7 days: the current period SOC reference value is equal to the reference value of the previous period plus ksoc multiplied by [η i (t) - Past 7-day η iThe difference between the current period and the historical average value is multiplied by the sensitivity coefficient to obtain a dynamic adjustment amount of the SOC reference value, and the adjustment amount is superimposed on the SOC reference value of the previous period to generate a corrected SOC reference value. The application further introduces a mean sliding window mechanism of cross-period historical frequency behavior trend to realize dynamic correction of the SOC reference. This method is equivalent to introducing a double structure of inertia term and correction term, embedding trend memory in the SOC adjustment process, avoiding excessive response of the system to occasional frequency anomalies, and improving the stability and anti-disturbance ability of the overall SOC adjustment.

[0053] S3, SOC callback during dead zone period: when the frequency is in the dead zone interval, the energy storage device is controlled in a low-power mode to adjust the SOC according to the current period SOC reference value, so that the actual SOC gradually returns to the target reference value.

[0054] SOC deviation calculation: ΔSOC(t) is equal to SOC_base,i minus the current actual SOC(t), the callback power P_recovery(t) is equal to kr multiplied by ΔSOC(t), and is limited in the range of [-P_recovery,max, P_recovery,max] by the saturation function sat, where kr is a proportional coefficient. When |ΔSOC(t)|≤ΔSOC_thresh, the callback is terminated. kr can be set according to the ratio of the sum of high frequency times and low frequency times to the dead zone frequency times multiplied by the first proportional coefficient k0. P_recovery,max is generally set to 5%-10% of the rated power.

[0055] S4, frequency adjustment execution deviation feedback: after the daily frequency adjustment task is completed, the system re-evaluates the difference between the frequency data of each period and the actual response energy and the theoretical value, calculates the deviation and feedback correction with a proportional coefficient kf, and dynamically updates the next day SOC_base, so that the control strategy forms a closed-loop iterative optimization.

[0056] The sampling period is Δt2, the frequency adjustment theoretical action amount E_theory,i is equal to k0 multiplied by Σ(f(t)-f0) multiplied by Δt1; the actual response energy E_actual,i is equal to the actual power P_real,i(t) multiplied by the cumulative sum of the sampling period Δt2. The energy deviation coefficient δi is equal to (E_theory,i-E_actual,i) divided by E_theory,i.

[0057] The corrected next day SOC_base,i(t+1)' is equal to the original reference value SOC_base,i(t+1) plus kf multiplied by δi, where kf is typically 1%-5%.

[0058] The method can be embedded in an electronic control unit or an industrial controller with computing capability, and all logical steps are realized through software.

[0059] The application also comprises an energy storage frequency regulation adaptive control system, which comprises a frequency sampling module, a frequency processing module, an SOC reference generation module, an SOC callback control module, a power control module and a feedback evaluation module, and each module cooperates to realize the above control method.

[0060] The application also comprises an energy storage frequency regulation device, which comprises an adaptive control system, a PCS, a step-up transformer, a BMS, a nickel-hydrogen battery system, an energy management system and the like.

[0061] Compared with the prior art, the application has the following beneficial effects: The time-period-based SOC reference control effectively broadens the frequency regulation adjustable range of the energy storage system, significantly reduces the frequency regulation failure caused by the SOC limit, improves the forward-looking and robustness of the control strategy by combining the adaptive mechanism of frequency trend correction, strengthens the SOC repair capability of the system during the frequency regulation idle period by introducing the SOC callback strategy in the dead zone, improves the preparation state of the secondary frequency regulation, improves the SOC control accuracy by the feedback closed-loop correction mechanism, alleviates the SOC deviation accumulation problem, and prolongs the service life of the energy storage equipment.

[0062] As shown in an embodiment, Figure 2 As shown in an embodiment, Figure 2 FIG. 1 is a comparison diagram of the SOC reference values and feedback correction of each time period of an embodiment one of the energy storage frequency regulation adaptive control method and system of the application. In this embodiment one, the project is deployed at a 6kV substation in Jiaquan, Ningxia, the energy storage system adopts a nickel-hydrogen battery module totaling 540 groups, each group is composed of 20 single cells in series, the single module rated voltage is 24V, the capacity is 50Ah, the total system configuration is 10MW / 0.6MWh, and supports 10C rate charging and discharging capability. The system is composed of an adaptive control system, a PCS, a step-up transformer, a BMS, a nickel-hydrogen battery system, an energy management system and the like. The adaptive control system is composed of an embedded industrial controller based on ARM Cortex-A72 architecture as the core unit, running Linux system and integrating six function modules of frequency sampling, data preprocessing, trend identification, SOC reference generation, control execution and feedback correction.

[0063] The adaptive control system executes the steps as follows: S1: Frequency sampling and processing: Real-time sampling of grid frequency f(t), sampling period Δt1=50ms, sample length n=1,728,000 points (24 hours), construct historical frequency time series {f(t-nΔt1),...,f(t)}. Filter processing uses a third-order Butterworth band-pass filter, with a bandwidth of 49.8Hz to 50.2Hz, expressed as Butterworth_bandpass(f(t), ω_low=49.8Hz, ω_high=50.2Hz), which can filter out low-frequency baseline drift and high-frequency sampling noise, ensuring that the frequency signal retains the main dynamic frequency modulation components.

[0064] Smooth processing uses a 5-point weighted moving average algorithm: Specifically, the current time smooth value is equal to the average of the past m sample point filter values Σ_{k=0}^{4}f_filtered(t-kΔt1), where m=5, k is the weight coefficient, and the mean weight k=1 is used; this method is used to further weaken high-frequency jitter and improve the smoothness of frequency trend analysis. To facilitate frequency feature pattern recognition and SOC reference generation, the processed frequency sequence is divided into N=4 segments according to actual control needs, i.e.: X1: 0:00-8:00, X2: 8:00-12:00, X3: 12:00-18:00, X4: 18:00-24:00, construct sequence {X1, X2, …, X4} as the basis input for subsequent frequency feature recognition and SOC reference generation.

[0065] S2: SOC reference adjustment. According to the four time periods divided in step S1, the system calculates and dynamically generates the SOC reference value for each period based on the statistical characteristics of the frequency sequence in each period and the energy imbalance index ηi, the calculation steps are as follows: Period X1 (0:00-8:00): Nlow=320, Nhigh=318, Ndead=62; Plow=29.4, Phigh=29.1, η1=(Plow-Phigh) / (Plow+Phigh)=0.0051, ksoc=0.25, β1=0.0013, SOC_base,1=50%+0.13%≈50.13%≈50%; Period X2 (8:00-12:00): Nlow=510, Nhigh=90, Ndead=40; Plow=52.7, Phigh=13.2, η2=0.6, ksoc=0.25, β2=0.2, SOC_base,2=50%+20%=70%; Period X3 (12:00-18:00): Nlow=120, Nhigh=480, Ndead=60; Plow=11.5, Phigh=41.2, η3=-0.564, ksoc=0.25, β3=-0.197, SOC_base,3=50%-19.7%=30.3%; Period X4 (18:00-24:00): Nlow=430, Nhigh=90, Ndead=60; Plow=44.8, Phigh=16.5, η4=0.461, ksoc=0.25, β4=0.1, SOC_base,4=50%+10%=60%.

[0066] The above SOC base values are automatically generated by the system according to real-time frequency, and are updated rolling during the day, and can be further dynamically predicted and corrected by a 7-day frequency trend correction model, with the current day ηi(t) and the average of the previous 6 days ηi, the historical average value of ηi is as follows: X1: η1(t)=0.0051, avg=0.006, Δη=-0.0009, β1=-0.0002, The corrected SOC_base,1=50%-0.02%=49.98%.

[0067] X2: η2(t)=0.60, avg=0.52, Δη=+0.08, β2=+0.02, The corrected SOC_base,2=70%+2%=72.00%.

[0068] X3: η3(t)= -0.564, avg= -0.48, Δη= -0.084, β3= -0.021, The corrected SOC_base,3=30.3%-2.1%=28.20%.

[0069] X4: η4(t)=0.461, avg=0.44, Δη=+0.021, β4=+0.005, The corrected SOC_base,4=60%+0.5%=60.50%.

[0070] S3: Dead zone period SOC callback: The SOC_base,1, SOC_base,2, SOC_base,3, SOC_base,4 values of the previous day are set as the new SOC_base,i target values of the day according to the X1, X2, X3, X4 time period, and when the frequency of each time period is between f0-Δfdead≤f(t)≤f0+Δfdead, the real-time SOC value is adjusted in a low-power manner. The specific adjustment control is as follows: X1 (0:00-8:00): SOC_base=49.98%, current detection value 47.1%, ΔSOC=2.9%; set kr=2700, P_recovery=sat(kr×ΔSOC)=78.3kW, adjustment time about 85s; X2 (8:00-12:00): SOC_base=72.00%, current value 67.6%, ΔSOC=2.4%; set kr=2700, P_recovery=64.8kW, adjustment time about 95s; X3 (12:00-18:00): SOC_base=28.20%, current value 32.4%, ΔSOC=–2.4%; set kr=2700, P_recovery=–64.8kW, discharge adjustment time about 92s; X4 (18:00-24:00): SOC_base=60.50%, current value 57.9%, ΔSOC=2.1%; set kr=2700, P_recovery=56.7kW, adjustment time about 85s.

[0071] The control strategy dynamically limits the amplitude according to the real-time ΔSOC, ensures that P_recovery does not exceed 10% of the rated value (i.e. 100kW), and automatically terminates when |ΔSOC|≤0.5%.

[0072] S4 Frequency adjustment execution deviation feedback: After completing the daily frequency adjustment task, the system recalculates the frequency data of each time period of the current day to generate the initial SOC_base,i(t+1) of the next day. To better demonstrate the correction process, the following lists the initial reference values SOC_base,i(t+1) generated by the frequency data analysis of the current day: SOC_base,1(t+1) =50.2%; SOC_base,2(t+1) = 68.1%; SOC_base,3(t+1) = 34.0%; SOC_base,4(t+1) = 61.2%.

[0073] Then, according to the difference between the theoretical response energy Etheory,i and the actual response energy Eactual,i, the energy deviation coefficient δi is calculated, and the closed-loop SOC reference is generated through proportional feedback correction: The next day SOC_base,i(t+1)' after correction is equal to the original reference value SOC_base,i(t+1) plus kf multiplied by δi.

[0074] Where kf=0.05 is the updated fixed feedback sensitivity coefficient. Example calculations are as follows: X1: Etheory=0.46MWh, Eactual=0.437MWh, δ1=-0.05, SOC_base,1(t+1)' = 50.2%-0.25%=49.95%; X2: Etheory=0.38MWh, Eactual=0.367MWh, δ2=-0.034, SOC_base,2(t+1)' = 68.1%-0.17%=67.93%; X3: Etheory=0.32MWh, Eactual=0.325MWh, δ3=+0.0156, SOC_base,3(t+1)' =34.0%+0.078%=34.078%; X4: Etheory=0.52MWh, Eactual=0.497MWh, δ4=-0.0442, SOC_base,4(t+1)' =61.2%-0.221%=60.979%.

[0075] Figure 2 For the comparison of the SOC reference values before and after feedback correction in the six time periods in Embodiment Two, the dynamic correction ability and adaptability of the system are clearly embodied.

[0076] As shown in an embodiment, Figure 3 As shown in an embodiment, Figure 3 FIG. 2 is a comparison chart of the SOC reference values before and after feedback correction in Embodiment Two of the energy storage frequency modulation adaptive control method and system. In Embodiment Two, the system is deployed at a 220kV substation in Jiangsu to support the load side, the energy storage system uses super capacitor technology, adopts a multi-cluster grouping architecture design, and has a total configuration capacity of 6MW / 0.4MWh. Each cluster of super capacitor modules has a rated voltage of 48V, and the energy capacity of a single cluster is about 0.8kWh. The system includes about 500 groups of series-parallel modules. The system is composed of an adaptive control system, a bidirectional PCS, an isolation booster transformer, a BMS, a super capacitor module, an electric energy management system, and an upper computer dispatch interface module. The adaptive control system is composed of an embedded industrial controller based on ARM Cortex-A72 architecture as the core unit, running the Linux system and integrating six functional modules of frequency sampling, data preprocessing, trend identification, SOC reference generation, control execution, and feedback correction.

[0077] The adaptive control system performs the following steps: S1: Frequency sampling and processing: Real-time sampling of the grid frequency f(t), sampling period At1=50ms, sample length n=1,728,000 points (24 hours), construct the historical frequency time series {f(t-nAt1),...,f(t)}. Filter processing uses a third-order Butterworth band-pass filter, with a bandwidth of 49.8Hz to 50.2Hz, The expression is Butterworth_bandpass(f(t), ω_low=49.8Hz, ω_high=50.2Hz), which can filter out low-frequency baseline drift and high-frequency sampling noise, ensuring that the frequency signal retains the main dynamic frequency modulation component. The smoothing process uses a 5-point weighted moving average algorithm, which is: The current time smoothed value is equal to the average value of the past m sampling point filtered values Σ_{k=0}^{4}f_filtered(t-kAt1), where m=5, k is the weight coefficient, and the mean weight k=1 is used; this method is used to further weaken high-frequency jitter and improve the smoothness of frequency trend analysis. To facilitate frequency feature pattern recognition and SOC reference generation, the processed frequency sequence is divided into N=6 segments according to actual control needs, i.e.: X1: 0:00-4:00, X2: 4:00-8:00, X3: 8:00-12:00, X4: 12:00-16:00, X5: 16:00-20:00, X6: 20:00-24:00, construct the sequence {X1, X2, …, X6} as the basis input for subsequent frequency feature recognition and SOC reference generation.

[0078] S2: SOC reference adjustment. According to the six time periods divided in step S1, the system calculates and dynamically generates the SOC reference value for each period according to the statistical characteristics of the frequency sequence in each period and the energy imbalance index ηi, the calculation steps are as follows: Period X1 (0:00-4:00): Nlow=230, Nhigh=200, Ndead=70; Plow=21.5, Phigh=20.6, η1=(Plow–Phigh) / (Plow+Phigh)=0.0212, ksoc=0.25, β1=0.0053, SOC_base,1=50%+0.53%=50.53%; Period X2 (4:00-8:00): Nlow=260, Nhigh=120, Ndead=60; Plow=26.8, Phigh=14.3, η2=0.304, β2=0.076, SOC_base,2=50%+7.6%=57.6%; Period X3 (8:00–12:00): Nlow=400, Nhigh=80, Ndead=40; Plow=41.0, Phigh=9.2, η3=0.634, β3=0.159, SOC_base,3=50%+15.9%=65.9%; Period X4 (12:00–16:00): Nlow=110, Nhigh=360, Ndead=30; Plow=11.5, Phigh=38.2, η4=–0.537, β4=–0.134, SOC_base,4=50%–13.4%=36.6%; Period X5 (16:00–20:00): Nlow=310, Nhigh=150, Ndead=40; Plow=33.5, Phigh=17.4, η5=0.316, β5=0.079, SOC_base,5=50%+7.9%=57.9%; Period X6 (20:00–24:00): Nlow=150, Nhigh=330, Ndead=60; Plow=16.4, Phigh=34.8, η6=–0.359, β6=–0.090, SOC_base,6=50%–9.0%=41.0%.

[0079] The above SOC benchmark value is automatically generated by the system according to the real-time frequency and is updated daily. It can be further dynamically predicted and corrected through the 7-day frequency trend correction model. It is corrected by the current day ηi(t) and the average value of ηi of the previous 6 days average(ηi). The historical average value of ηi is as follows: X1: η1(t)=0.0212, avg=0.020, Δη=+0.0012, β1=+0.0003, After correction, SOC_base,1=50.53%+0.03%=50.56%; X2: η2(t)=0.304, avg=0.310, Δη=–0.006, β2=–0.0015, After correction, SOC_base,2=57.6%–0.15%=57.45%; X3: η3(t) = 0.634, avg = 0.620, Δη = +0.014, β3 = +0.0035, Corrected SOC_base,3 = 65.9% + 0.35% = 66.25%; X4: η4(t) = -0.537, avg = -0.520, Δη = -0.017, β4 = -0.0043, Corrected SOC_base,4 = 36.6% - 0.43% = 36.17%; X5: η5(t) = 0.316, avg = 0.300, Δη = +0.016, β5 = +0.004, Corrected SOC_base,5 = 57.9% + 0.4% = 58.3%; X6: η6(t) = -0.359, avg = -0.340, Δη = -0.019, β6 = -0.0048, Corrected SOC_base,6 = 41.0% - 0.48% = 40.52%.

[0080] S3: Deadband period SOC callback: The previous day SOC_base,1, SOC_base,2,... SOC_base,6 values are used as the SOC reference value of each time period, when the frequency f(t) is in the deadband interval [f0-Δf_dead, f0+Δf_dead], the energy storage system adjusts the real-time SOC in low power mode, so that the SOC gradually approaches SOC_base,i, as follows: X1 (0:00-4:00): SOC_base = 50.56%, current detection value 48.1%, ΔSOC = 2.46%; Set kr = 2700, P_recovery = 66.42kW, callback time ≈ 80s; X2 (4:00-8:00): SOC_base = 57.45%, current value 54.6%, ΔSOC = 2.85%; P_recovery = 76.95kW, time ≈ 90s; X3 (8:00-12:00): SOC_base = 66.25%, current value 61.4%, ΔSOC = 4.85%; P_recovery = 130.95kW (limited to 100kW), callback time ≈ 120s; X4 (12:00-16:00): SOC_base=36.17%, current value 38.6%, ΔSOC= -2.43%; P_recovery= -65.61kW, time ≈ 90s; X5 (16:00-20:00): SOC_base=58.3%, current value 56.2%, ΔSOC=2.1%; P_recovery=56.7kW, time ≈ 85s; X6 (20:00-24:00): SOC_base=40.52%, current value 43.4%, ΔSOC= -2.88%; P_recovery= -77.76kW, time ≈ 95s.

[0081] The control strategy is automatically terminated when |ΔSOC|≤0.5%, and P_recovery cannot exceed 10% of the rated power.

[0082] S4: Frequency adjustment execution deviation feedback: After completing the daily frequency adjustment task, the system recalculates and generates the initial SOC_base,i(t+1) for the next day. The following is a hypothetical calculation: SOC_base,1(t+1)=50.3%; SOC_base,2(t+1)=56.9%; SOC_base,3(t+1)=67.2%; SOC_base,4(t+1)=35.8%; SOC_base,5(t+1)=58.9%; SOC_base,6(t+1)=41.0%.

[0083] The deviation δi between the theoretical energy E_theory and the actual energy E_actual is corrected: The corrected next day SOC_base,i(t+1)' is equal to the original baseline value SOC_base,i(t+1) plus kf multiplied by δi, kf=0.05.

[0084] X1: E_theory=0.32MWh, E_actual=0.304MWh, δ1= -0.05, SOC_base,1= 50.3% - 0.25%= 50.05%; X2: E_theory=0.26MWh, E_actual=0.254MWh, δ2= -0.023, SOC_base,2= 56.9% - 0.12%= 56.78%; X3: E_theory=0.31MWh, E_actual=0.319MWh, δ3 = +0.029, SOC_base,3 = 67.2% + 0.15% = 67.35%; X4: E_theory = 0.22 MWh, E_actual = 0.213 MWh, δ4 = -0.0318, SOC_base,4 = 35.8% - 0.16% = 35.64%; X5: E_theory = 0.28 MWh, E_actual = 0.27 MWh, δ5 = -0.032, SOC_base,5 = 58.9% - 0.16% = 58.74%; X6: E_theory = 0.24 MWh, E_actual = 0.247 MWh, δ6 = +0.029, SOC_base,6 = 41.0% + 0.15% = 41.15%.

[0085] Figure 3 The comparative changes of the SOC base values of the six time periods in Embodiment Two before and after the feedback correction are shown, which clearly shows the dynamic correction ability and adaptability of the system.

[0086] The present application is not limited to the above-mentioned embodiments, and any changes in shape or material composition are allowed, as long as the structure design provided by the present application is adopted. Any deformation of the present application shall be considered within the protection scope of the present application.

Claims

1. A method for adaptive control of energy storage frequency modulation, characterized in that: include: Collect grid frequency signals in real time, perform filtering and noise reduction processing, and generate historical frequency sequences according to preset time periods; Calculate the energy imbalance index based on the statistical characteristics of the frequency series of each time period and dynamically generate the SOC benchmark value series for each time period; In the frequency dead zone, the actual SOC is returned to the current period reference value by low-power fine-tuning; According to the deviation between the theoretical response energy and the actual response energy of frequency modulation, the next day's SOC reference value sequence is corrected in a closed loop.

2. The energy storage frequency modulation adaptive control method according to claim 1, characterized in that: The statistical features based on the frequency sequence of each time period include: counting the number of low frequencies, high frequencies, and dead zone frequencies in each time period; wherein the low frequency number refers to the number of times the frequency is less than the rated value minus the dead zone threshold, and the high frequency number refers to the number of times the frequency is greater than the rated value plus the dead zone threshold; Calculate the low-frequency cumulative action amount and the high-frequency cumulative action amount; wherein, the low-frequency cumulative action amount is calculated by multiplying the first proportional coefficient by the integral value of the negative frequency offset, and the high-frequency cumulative action amount is calculated by multiplying the first proportional coefficient by the integral value of the positive frequency offset.

3. The energy storage frequency modulation adaptive control method according to claim 2, characterized in that: The energy imbalance indicator is specifically a difference between the low-frequency cumulative motion amount and the high-frequency cumulative motion amount divided by the sum of the two.

4. The energy storage frequency modulation adaptive control method according to claim 1, 2 or 3, characterized in that: The dynamic generation of the SOC reference value includes: mapping the energy imbalance indicator to an SOC offset, specifically, the SOC offset is equal to the sensitivity coefficient multiplied by the energy imbalance indicator; wherein the sensitivity coefficient is typically 0.1-0.3; the time period SOC reference value is the initial reference value plus the SOC offset, and upper and lower limits are set to be within the range of 20%-80%.

5. The energy storage frequency modulation adaptive control method according to claim 4, characterized in that: The dynamic generation of the time-divided SOC reference value sequence also includes building a dynamic correction model based on the historical frequency trend of the past 7 days, specifically: Get the historical average value of the energy imbalance indicator for the same period of the previous 6 days; Multiplying the difference between the current period and the historical average by the sensitivity coefficient to obtain a dynamic adjustment amount of the SOC reference value; The adjustment amount is added to the SOC reference value of the previous period to generate a corrected SOC reference value.

6. The energy storage frequency modulation adaptive control method according to claim 5, characterized in that: The low power fine-tuning method specifically includes: Calculate the absolute deviation between the real-time SOC and the current period benchmark value; Multiplying the absolute deviation by a second proportional coefficient to generate a callback power; When the absolute deviation is less than or equal to the preset threshold, the callback operation is terminated; The callback power is limited to 5% to 10% of the rated power, and the second proportional coefficient is set according to the ratio of the sum of the high frequency times and the low frequency times to the dead zone frequency times multiplied by the first proportional coefficient.

7. The energy storage frequency modulation adaptive control method according to claim 5 or 6, characterized in that: The closed-loop correction of the next-day SOC baseline value sequence specifically includes: after the daily frequency modulation task is completed, the system re-evaluates the frequency data of each time period and the difference between the actual response energy and the theoretical value, calculates the deviation and performs feedback correction with a third proportional coefficient, dynamically updates the next-day SOC baseline value, and forms a closed-loop iterative optimization of the control strategy.

8. The energy storage frequency modulation adaptive control method according to claim 7, characterized in that: The system re-evaluates the frequency data of each time period and the difference between the actual response energy and the theoretical value, including: calculating the theoretical response energy and the actual response energy of each time period, wherein the theoretical response energy is a first proportional coefficient multiplied by the frequency offset integral value, and the actual response energy is an integral value of the actual power over time; Calculating the deviation and performing feedback correction using the third proportional coefficient includes: dividing the difference between the theoretical response energy and the actual response energy by the theoretical response energy to obtain an energy deviation coefficient, and multiplying the energy deviation coefficient by the third proportional coefficient to generate a correction amount for the SOC reference value, wherein a typical value of the third proportional coefficient is 1%-5%; The dynamically updating next-day SOC reference value is to add the correction amount to the next-day initial SOC reference value.

9. The energy storage frequency modulation adaptive control method according to claim 8, characterized in that: The filtering and noise reduction processing and generating a historical frequency sequence by pre-set time periods include: using bandpass filtering and a weighted sliding mean algorithm to perform noise reduction and smoothing processing, dividing the frequency throughout the day into N time periods by hour to construct a historical frequency sequence, and extracting the dynamic change characteristics of the frequency.

10. An energy storage frequency modulation adaptive control system, adopting an energy storage frequency modulation adaptive control method according to any one of claims 1 to 9, characterized in that: include: The frequency sampling module is connected to the frequency processing module to collect the grid frequency signal in real time; Frequency processing module, which performs filtering, smoothing and time segmentation operations; The SOC benchmark generation module receives data from the frequency processing module and dynamically outputs the time-segment SOC benchmark value to the SOC callback control module; SOC callback control module triggers low-power SOC fine-tuning in the dead-band frequency range; The power control module is connected to the SOC callback control module to execute the frequency modulation power command and the SOC callback command; The feedback evaluation module is connected to the power control module and optimizes the SOC baseline value through closed-loop energy deviation analysis.

Images