Multi-scale self-adaptive cooperative network construction type energy storage frequency modulation control system and method

Abstract

The invention provides a multi-scale self-adaptive cooperative network construction type energy storage frequency modulation control system and method, and the system comprises a constant-voltage energy storage power supply which is used for providing DC electric energy; the three-phase inverter circuit is used for converting the direct-current electric energy into alternating-current electric energy and outputting the alternating-current electric energy to a power grid or absorbing the electric energy; the power calculator is used for calculating active power and reactive power output to the power grid by the three-phase inverter circuit; the battery SOH dynamic evaluation controller is used for outputting an SOH attenuation rate signal; the active loop controller is used for generating multi-scale adaptive frequency modulation data and sending the multi-scale adaptive frequency modulation data to the pulse adjustment controller; the reactive power ring controller is used for generating reactive power regulation data according to the power grid voltage deviation signal and sending the reactive power regulation data to the pulse regulation controller; and the pulse regulation controller is used for outputting a final pulse regulation instruction and a multi-scale cooperative control command to the three-phase inverter circuit. The oscillation risk is reduced, and long-term stable operation of the power system is guaranteed.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system control, and particularly relates to a multi-scale adaptive collaborative grid-forming energy storage frequency modulation control system and method. Background Art

[0002] With the continuous transformation of the global energy structure, the proportion of new energy power generation (such as wind power, photovoltaic power, etc.) in the power system is increasing day by day. Since such new energy units often have intermittency and volatility, their output power is prone to rapid and unpredictable changes affected by environmental factors. When a large amount of such energy is connected to the power grid, the following technical problems are likely to occur:

[0003] 1. Frequency Fluctuation and Equipment Safety

[0004] The short-term fluctuations in the output of new energy units will cause frequent and large-amplitude changes in the power grid frequency. Frequency instability not only affects the normal operation of conventional units and energy storage devices, but also may cause misoperation of power grid relay protection, and even trigger large-scale safety risks.

[0005] 2. Limitations of Traditional Frequency Modulation Methods

[0006] Existing frequency modulation methods often adopt fixed-parameter control strategies, and use the same set of control parameters for power grid frequency deviations of different degrees or different rates. On the one hand, the linkage between primary frequency modulation and secondary frequency modulation is insufficient, and it is difficult to balance between short-term rapid response and long-term deep regulation; on the other hand, when the power grid frequency fluctuates rapidly and greatly, the fixed-parameter strategy may lead to low regulation accuracy and efficiency, and cannot meet the dynamic requirements of the system.

[0007] 3. Energy Storage Battery Health and Economy Issues

[0008] The deep charge and discharge of the energy storage system during frequency modulation will exacerbate the attenuation of the battery. Especially in the working conditions with severe frequency fluctuations, traditional methods often ignore the comprehensive consideration of the state of health (SOH) of the energy storage battery, resulting in a shortened service life of the energy storage system and an increased operating cost. How to balance power grid frequency stability and battery life management has become a problem that is difficult to effectively balance in the existing technology.

[0009] 4. Large-Scale Grid Connection Oscillation Risk

[0010] As the penetration rate of new energy increases, the inertia of the power system gradually decreases, and additional oscillation modes are likely to occur. If the frequency deviation cannot be controlled in a timely and effective manner, the system oscillation risk will further increase, reducing the overall power quality and aggravating equipment wear.

[0011] In view of the above existing technical problems, due to the lack of comprehensive coordination of power grid frequency stability and energy storage battery life under multiple time scales, the frequency modulation control method of traditional grid-forming energy storage converters has obvious deficiencies in terms of rapidity, accuracy, and economy. How to reduce the energy storage attenuation cost while ensuring the frequency modulation effect and further improve the safety and reliability of large-scale new energy grid connection still urgently needs to be improved and perfected. It is precisely in response to these problems that the industry expects an advanced frequency modulation control technology that can take into account multi-scale frequency modulation accuracy and energy storage life management and improve the safety margin of the power system. Summary of the Invention

[0012] The purpose of the present invention is to solve the deficiencies existing in the above background technology, and provide a grid-forming energy storage frequency modulation control system and method with multi-scale adaptive coordination, comprehensively considering the frequency modulation economy in the long time period and the power grid frequency stability in the short time period, reducing the oscillation risk, and ensuring the long-term stable operation of the power system.

[0013] The technical solution adopted by the present invention is: a grid-forming energy storage frequency modulation control system based on multi-scale adaptive coordination, including:

[0014] A constant voltage energy storage power supply, whose output terminal is connected to the DC input terminal of the three-phase inverter circuit, and is used to provide DC electrical energy;

[0015] A three-phase inverter circuit, whose AC output terminal is connected to the power grid through a filter inductor, a filter capacitor, and an equivalent power grid impedance in sequence, and is used to convert DC electrical energy into AC electrical energy and output or absorb electrical energy to the power grid;

[0016] A power calculator, electrically connected to the output side of the three-phase inverter circuit, and is used to calculate the active power and reactive power output by the three-phase inverter circuit to the power grid;

[0017] A battery SOH dynamic evaluation controller, which is used to monitor and evaluate the health state of the constant voltage energy storage power supply and output an SOH attenuation rate signal;

[0018] An active loop controller, which is used to generate multi-scale adaptive frequency modulation data based on the SOH prediction signal, the attenuation rate signal, and the power grid frequency information and send it to the pulse regulation controller;

[0019] A reactive loop controller, which is used to generate reactive power regulation data according to the power grid voltage deviation signal and send it to the pulse regulation controller;

[0020] A pulse regulation controller, which is used to receive the control data output by the active loop controller and the reactive loop controller, perform fast pulse regulation in combination with the transient changes of the power grid, and output the final pulse regulation instruction and multi-scale coordination control command to the three-phase inverter circuit to achieve stable regulation of the power grid frequency and voltage.

[0021] In the above technical solution, the battery SOH dynamic evaluation controller includes:

[0022] A battery temperature comparator for comparing the detected battery temperature with a preset temperature threshold and outputting a temperature comparison signal;

[0023] An SOH prediction controller for estimating the battery SOH value according to the temperature comparison signal and the operating parameters of the battery;

[0024] A differential controller for performing a differential operation on the SOH value and outputting an SOH attenuation rate prediction signal;

[0025] Wherein, the SOH attenuation rate prediction signal is sent to the active power loop controller, so that the frequency modulation control process can be adaptively adjusted according to the battery health state.

[0026] In the above technical solution, the SOH prediction controller includes:

[0027] A battery normal operation response trigger module for receiving and identifying the battery normal operation response signal generated by the battery temperature comparator, and starting the LSTM prediction algorithm module to perform SOH prediction after confirming that the battery operating temperature is in a normal state;

[0028] An LSTM prediction algorithm module for obtaining the real-time SOH state and historical SOH state data of the battery, and co-processing the data through a trained long short-term memory neural network to output the predicted battery SOH value;

[0029] Wherein, the battery temperature comparator compares the detected battery temperature with a preset temperature threshold and selects and outputs a battery normal operation response signal according to the comparison result.

[0030] In the above technical solution, the active power loop controller includes:

[0031] An adaptive model selection controller for outputting a corresponding frequency modulation model or mode according to the frequency information of the power grid and the SOH attenuation rate signal;

[0032] An active power loop coordinated frequency modulation controller for allocating and coordinating the output weights of primary frequency modulation and secondary frequency modulation according to the output of the adaptive model selection controller;

[0033] An active power loop primary frequency modulation circuit for receiving a primary frequency modulation control signal and realizing short-time fast frequency modulation support;

[0034] An active power loop secondary frequency modulation circuit for compensating the residual deviation of primary frequency modulation on a longer time scale;

[0035] Among them, the active loop controller sends the integrated multi-scale frequency modulation data to the pulse regulation controller.

[0036] In the above technical solution, the adaptive model selection controller includes:

[0037] A variance calculator, configured to subtract the grid frequency signal from the reference grid frequency signal and perform differential processing to obtain a frequency error variance signal;

[0038] A state latch, configured to perform threshold comparison and latching on the SOH attenuation rate prediction signal to obtain an SOH attenuation rate signal;

[0039] A first divider, configured to divide the frequency error variance signal by the SOH attenuation rate signal to obtain a frequency modulation-battery life impact signal;

[0040] A first adaptive controller, configured to compare the frequency modulation-battery life impact signal with a preset threshold to determine a frequency modulation output mode control signal, and output the frequency modulation output mode control signal to the active loop coordinated frequency modulation controller to select modes such as primary frequency modulation priority or primary and secondary frequency modulation coordination.

[0041] In the above technical solution, the active loop coordinated frequency modulation controller includes:

[0042] A first subtractor, configured to subtract the grid frequency signal from the reference frequency signal to obtain a frequency error signal;

[0043] A differentiator, configured to perform differential calculation on the frequency error signal to generate a frequency error change rate signal;

[0044] A second adaptive controller, configured to perform adaptive comparison processing on the frequency error signal and the frequency error change rate signal according to the frequency modulation output mode selection signal, and output a coordinated frequency modulation control signal;

[0045] A fuzzy controller, configured to perform fuzzy operation processing on the coordinated frequency modulation control signal and the SOH attenuation rate signal according to a preset fuzzy set and fuzzy rules to obtain an output weight ratio signal for primary and secondary frequency modulation;

[0046] The active loop coordinated frequency modulation controller sends the output weight ratio signal to the active loop primary frequency modulation circuit and the secondary frequency modulation circuit, thereby realizing multi-scale coordinated frequency modulation.

[0047] In the above technical solution, the active loop primary frequency modulation circuit includes:

[0048] A second subtractor, configured to compare the grid voltage frequency with the frequency target signal and output a grid voltage frequency error signal;

[0049] A third adaptive controller, configured to calculate an optimal droop coefficient signal based on an output power ratio signal and a grid voltage frequency error signal;

[0050] A multiplier, configured to calculate a primary frequency regulation feedback signal based on the optimal droop coefficient signal and the grid voltage frequency error signal; and send it to the secondary frequency regulation circuit of the active power loop.

[0051] In the above technical solution, the secondary frequency regulation circuit of the active power loop includes:

[0052] A second divider, configured to divide the primary frequency regulation feedback signal by the grid frequency error signal to obtain an adaptive droop parameter signal;

[0053] An adaptive compensator, configured to perform compensation calculation based on the frequency modulation output ratio signal and the adaptive droop parameter signal, and output an adaptive PI control parameter signal;

[0054] An integrator, configured to perform proportional-integral operation on the frequency error signal according to the adaptive PI control parameter signal to generate a compensated active power reference value;

[0055] An adder, configured to perform frequency compensation and integral operation on the compensated active power reference value and the primary frequency regulation feedback signal to obtain an adaptive frequency regulation feedback signal;

[0056] Wherein, the adaptive frequency regulation feedback signal further compensates for the residual deviation of the primary frequency regulation on a relatively long time scale, and is integrated through the active power loop controller, and then a phase angle reference signal is obtained and output to the pulse regulation controller.

[0057] In the above technical solution, the active power loop controller further includes:

[0058] An active power loop subtractor, configured to subtract the adaptive frequency regulation feedback signal from the set value of the active power at the grid connection point to obtain an active power reference value;

[0059] An active power loop adder, configured to add the active power reference value and the active power at the grid connection point to obtain an error active power signal;

[0060] An active power loop integrator, configured to calculate and process the error active power signal into a frequency signal, and obtain a phase angle reference signal after integral calculation.

[0061] The present invention further provides a multi-scale adaptive collaborative grid-forming energy storage frequency modulation control method, which is implemented based on the system described in the above technical solution, and includes the following steps:

[0062] A constant-voltage energy storage power supply, whose output terminal is connected to the DC input terminal of the three-phase inverter circuit to provide DC electrical energy;

[0063] Three-phase inverter circuit, whose AC output terminal is connected to the power grid through a filter inductor, a filter capacitor and an equivalent impedance of the power grid in sequence, converts DC electrical energy into AC electrical energy and outputs or absorbs electrical energy to the power grid;

[0064] Power calculator, electrically connected to the output side of the three-phase inverter circuit, calculates the active power and reactive power output by the three-phase inverter circuit to the power grid;

[0065] Battery SOH dynamic evaluation controller, monitors and evaluates the health state of the constant-voltage energy storage power supply and outputs an SOH decay rate signal;

[0066] Active loop controller, generates multi-scale adaptive frequency modulation data based on the SOH decay rate signal and grid frequency information and sends it to the pulse regulation controller;

[0067] Reactive loop controller, generates reactive power regulation data according to the grid voltage deviation signal and sends it to the pulse regulation controller;

[0068] Pulse regulation controller, receives the control data output by the active loop controller and the reactive loop controller, combines the transient changes of the power grid to perform fast pulse regulation, and outputs the final pulse regulation instruction and multi-scale collaborative control command to the three-phase inverter circuit to achieve stable regulation of the grid frequency and voltage.

[0069] The beneficial effects of the present invention are: by integrating the constant-voltage energy storage power supply, three-phase inverter circuit, power calculator, battery SOH dynamic evaluation controller, active / reactive loop controller and pulse regulation controller in the same system, the present invention realizes the unified management and collaborative regulation of active and reactive power. The pulse regulation controller forms a complement with the active / reactive loop in the short-term fast regulation and long-term stable regulation, which can improve the comprehensive control accuracy, response speed and stability of the system for frequency and voltage. The introduction of the battery SOH dynamic evaluation controller enables the consideration of battery life and economy during the frequency modulation process, providing guarantee for the long-term operation of large-scale energy storage systems.

[0070] Furthermore, the battery temperature comparator of the present invention enables the system to capture the battery temperature change in real time and compare it with the threshold value, timely identify high temperature or abnormal conditions, and protect the battery safety. Through the cooperation of the SOH prediction controller and the differential controller, it is possible to make a trend or decay rate prediction of the battery health state, providing key support for subsequent adaptive frequency modulation. The output SOH decay rate prediction signal can provide real-time battery condition feedback for the active loop controller, enabling the frequency modulation strategy to maintain a good frequency modulation effect while taking into account the battery life.

[0071] Furthermore, by introducing the LSTM algorithm into the SOH prediction controller, the present invention can perform high-precision time-series analysis on the current and historical state data of the battery, improving the prediction accuracy. The battery normal working response trigger module ensures that the LSTM prediction is only executed within the temperature safety range, enhancing the effectiveness of the algorithm and the battery safety. A more intelligent state of health monitoring method can provide a reliable basis for subsequent frequency regulation decisions, balancing the rapid regulation demand and battery life management.

[0072] Furthermore, through the hierarchical design of the adaptive model selection controller, the active power loop cooperative frequency regulation controller, and the primary / secondary frequency regulation circuit, the present invention enables the frequency regulation to be allocated to different subsystems with fast time and long time, making it more targeted. The primary and secondary frequency regulations cooperate with each other to ensure rapid suppression in the case of large frequency deviations and fine compensation in the case of small deviations and steady-state stages, significantly improving the frequency regulation stability. The adaptive model selection controller can dynamically switch the frequency regulation mode according to the frequency and SOH signals, better balancing the frequency regulation demand and battery health consumption.

[0073] Furthermore, the present invention improves the regulation accuracy through a variance calculator and a latch, performs variance and differential processing on the grid frequency deviation, and latches the threshold of the SOH decay rate prediction signal, which can accurately identify large deviations or critical states and provide timely and accurate signal basis for the frequency regulation strategy. By dividing the frequency error variance signal by the SOH decay rate signal, the present invention quantifies the "frequency regulation-battery life impact signal", which can intuitively measure the balance point between the frequency regulation gain and the energy storage decay cost. The present invention compares the impact signal with a preset threshold and outputs a frequency regulation output mode control signal, enabling flexible switching between primary frequency regulation or primary and secondary cooperative modes, and realizing the overall consideration of efficiency and life.

[0074] Furthermore, through the first subtractor and differentiator, the present invention can continuously monitor the grid frequency deviation and its change rate in real time, providing refined input for subsequent adaptive control. The second adaptive controller performs adaptive comparison processing according to information such as the "frequency regulation output mode selection signal", error signal, and error change rate, enabling the system to have better adaptive capabilities when dealing with different fluctuation scenarios. The present invention uses fuzzy rules to comprehensively calculate the cooperative frequency regulation signal and generate the output weight ratio of primary and secondary frequency regulations to achieve seamless connection of multi-scale cooperative frequency regulation.

[0075] Furthermore, for the invention's re-comparison of the grid voltage frequency, the second subtractor focuses more on the difference between the grid voltage frequency and the target frequency, providing a more direct error signal input. The third adaptive controller combines the "output power weight ratio signal" and the "grid voltage frequency error signal" to dynamically calculate the optimal droop coefficient and achieve efficient and fast frequency modulation response. The invention simplifies the control link through a multiplier. Multiplying the droop coefficient by the frequency error signal to obtain the primary frequency modulation feedback signal, which has a clear structure and is easy to implement, providing a clear input for the subsequent processing of the secondary frequency modulation circuit.

[0076] Furthermore, the secondary frequency modulation of the invention is more refined. Through the cooperation of the second divider, the adaptive compensator, the integrator, and the adder, it realizes further fine correction of the primary frequency modulation feedback signal and compensates for the residual deviation that cannot be solved by the primary frequency modulation. The invention adaptively adjusts the proportional-integral control parameters under large / small frequency conversion conditions to achieve deep steady-state correction of the frequency and fully protect the battery life. The generated adaptive frequency modulation feedback signal corrects the residual deviation of the primary frequency modulation on a long time scale, making the frequency deviation tend to zero and maintaining stability, enhancing the steady-state performance of the system operation.

[0077] Furthermore, the invention realizes more advanced active power management. By respectively operating on the adaptive frequency modulation feedback signal and the active power at the grid connection point through the active loop subtractor and the adder, the system active power reference value and the error signal can be further obtained. After integrating the error active power signal, the phase angle reference signal is obtained, enabling the system to balance the frequency modulation demand and the battery characteristics at the global level and optimizing the grid frequency support effect. This additional circuit further improves the end link of the active loop control to ensure accurate monitoring and correction of the system output frequency after multi-stage compensation.

[0078] Furthermore, the invention connects the aforementioned system hardware / control strategies in series in the form of method steps, facilitating step-by-step deployment and implementation in actual engineering. Through the sequence and data flow description, each controller, energy storage power supply, inverter and other modules cooperate on different time scales to ensure the efficient connection of the stable regulation of frequency and voltage. The invention realizes flexible engineering applications for the system energy flow, control flow, and signal flow, providing an operable guiding specification for different power sites. Description of the Drawings

[0079] Figure 1 is the topological structure diagram of the invention;

[0080] Figure 2 is the structure diagram of the battery health state dynamic assessment controller;

[0081] Figure 3 is the structure diagram of the active loop controller;

[0082] Figure 4It is the structure diagram of the active loop adaptive mode selection controller;

[0083] Figure 5 It is the structure diagram of the active loop coordinated frequency modulation controller;

[0084] Figure 6 It is the structure diagram of the principle of the active loop primary frequency modulation circuit;

[0085] Figure 7 It is the structure diagram of the principle of the active loop secondary frequency modulation circuit;

[0086] Figure 8 It is the comparative analysis diagram of traditional control and multi-scale adaptive coordinated frequency modulation control.

[0087] Among them, 1 - constant voltage energy storage power supply, 2 - three-phase inverter circuit, 3 - filter inductor, 4 - filter capacitor, 5 - grid equivalent impedance, 6 - grid, 7 - battery SOH dynamic evaluation controller, 8 - pulse adjustment controller, 9 - active loop controller, 10 - reactive loop controller, 11 - power calculator; 71 - battery temperature comparator, 72 - SOH prediction controller, 73 - differential controller; 91 - adaptive model selection controller, 92 - coordinated frequency modulation controller, 93 - active loop primary frequency modulation circuit, 94 - active loop secondary frequency modulation circuit, 95 - active loop subtractor, 96 - active loop adder, 97 - active loop integrator; 911 - variance calculator, 912 - state latch, 913 - first divider, 914 - first adaptive controller; 921 - first subtractor, 922 - differentiator, 923 - second adaptive controller, 924 - fuzzy controller; 931 - second subtractor, 932 - third adaptive controller, 933 - multiplier; 941 - second divider, 942 - adaptive compensator, 943 - integrator, 944 - adder. Specific implementation manners

[0088] The following further elaborates on the present invention in detail in conjunction with the accompanying drawings and specific embodiments, which is convenient for clearly understanding the present invention, but they do not constitute a limitation to the present invention.

[0089] Embodiment 1

[0090] As Figure 1 shown, the present invention provides a grid-forming energy storage frequency modulation control system based on multi-scale adaptation and coordination, including:

[0091] A constant voltage energy storage power supply 1, whose output terminal is connected to the DC input terminal of a three-phase inverter circuit 2, for providing DC electrical energy;

[0092] A three-phase inverter circuit 2, whose AC output terminal is connected to the grid 6 through a filter inductor 3, a filter capacitor 4, and a grid equivalent impedance 5 in sequence, for converting DC electrical energy into AC electrical energy and outputting or absorbing electrical energy to the grid;

[0093] A power calculator 11, electrically connected to the output side of the three-phase inverter circuit 2, for calculating the active power and reactive power output by the three-phase inverter circuit 2 to the power grid 6;

[0094] A battery SOH dynamic evaluation controller 7, for monitoring and evaluating the health state of the constant-voltage energy storage power supply 1 and outputting an SOH decay rate signal;

[0095] An active loop controller 9, for generating multi-scale adaptive frequency modulation data based on the SOH decay rate signal and the power grid frequency information and sending it to the pulse regulation controller 8;

[0096] A reactive loop controller 10, for generating reactive power regulation data according to the power grid voltage deviation signal and sending it to the pulse regulation controller 8;

[0097] A pulse regulation controller 8, for receiving the control data output by the active loop controller 9 and the reactive loop controller 10, performing fast pulse regulation in combination with the transient changes of the power grid 6, and outputting the final pulse regulation instruction and multi-scale collaborative control command to the three-phase inverter circuit 2 to achieve stable regulation of the power grid frequency and voltage.

[0098] Specifically, as Figure 2 shown, the battery SOH dynamic evaluation controller 7 includes a battery temperature comparator 71, an SOH prediction controller 72, and a differential controller 73, for dynamically monitoring and evaluating the health state of the energy storage battery, and feeding back the evaluation result to the active loop controller 9 for frequency modulation decision-making, so as to realize the collaborative optimization of the energy storage battery life and the power grid frequency regulation.

[0099] The battery temperature comparator 71 compares the collected battery signal with a set threshold, detects the working temperature of the battery (i.e., the constant-voltage energy storage power supply 1), and generates a battery normal working response signal 710 after processing to trigger the SOH prediction controller. The SOH prediction controller 72 performs collaborative processing on the collected battery SOH state and the historical SOH state, and obtains an SOH prediction signal 720 through the trained LSTM algorithm. The differential controller 73 performs differential calculation processing on the SOH prediction signal 720 transmitted by the SOH prediction controller to obtain an SOH decay rate prediction signal 730, and transmits it to the active loop controller 9 to participate in the frequency modulation decision-making.

[0100] Preferably, the battery temperature comparator 71 is connected to a battery temperature sensor for real-time collection and monitoring of the working temperature of the battery. The battery temperature comparator 71 compares the detected temperature value with a preset temperature threshold:

[0101] If the temperature is within the safe range, a battery normal operation response signal 710 is generated to trigger the normal operation of the SOH prediction controller;

[0102] When the temperature exceeds the limit or is abnormal, protection measures can be triggered to directly discard the currently collected battery SOH status data and transmit a set over-threshold signal to the SOH prediction controller. At the same time, the weight of the historical SOH data in the SOH prediction module is increased to ensure that the most conservative frequency modulation control method is directly adopted after the data is transmitted to the active power loop regulator.

[0103] In this embodiment, the battery temperature comparator 71 can use the temperature measurement module integrated in the battery management system (BMS) to output the temperature signal in digital or analog form and make real-time comparison with the preset threshold (such as formulated according to battery specifications or safety standards). If it is detected that the temperature is in the normal range, a battery normal operation response signal 710 is output as the trigger condition for starting the next SOH prediction.

[0104] After the battery normal operation response signal 710 is triggered, the SOH prediction controller 72 obtains the real-time SOH status (such as remaining capacity, internal resistance, etc.) and historical SOH status data of the battery from the BMS or the local data recording unit.

[0105] The SOH prediction controller 72 internally executes an algorithm module based on the long short-term memory (LSTM) neural network to co-process the real-time and historical data to predict the health trend of the battery under the current working conditions. Specifically:

[0106] The LSTM algorithm module loads a neural network model trained for the target battery. The model inputs include voltage, current, and historical SOH trajectories, and the output is the predicted SOH value at the next moment. By using multiple LSTM units to extract and learn features from time series data, the aging law and health decay curve of the battery can be captured more accurately. The prediction result is output in the form of an SOH prediction signal 720 and can be further fed back to the system monitoring platform or fused with the frequency modulation loop information. The final SOH prediction signal (usually 0 - 1 or 0% - 100%) is calculated as shown in the following formula:

[0107] SOH t = σ(W y ·h t + b y )

[0108] where, f t is the forget gate, C t is the updated memory cell, h t is the current hidden state, and the specific calculation process is as shown in the following formula:

[0109] f t= σ(W f · [h t-1 ,x t + b f )

[0110]

[0111] h t = o t ⊙ tanh(C t )

[0112] On the premise of ensuring temperature safety, the SOH prediction controller 72 can perform real-time dynamic evaluation on the battery health state, providing accurate health information support for subsequent frequency modulation decision-making.

[0113] The differential controller 73 receives the SOH prediction signal 720 output by the SOH prediction controller 72 and performs differential operation on this signal to obtain key parameters such as the battery SOH attenuation rate.

[0114] In this embodiment, the differential controller 73 performs first-order differential processing on the SOH prediction value to obtain the SOH attenuation rate prediction signal 730, which is used to characterize the downward trend of the battery health degree per unit time. In this way, the accelerated deterioration phenomenon of the battery health state can be detected in time, and a warning can be issued or the frequency modulation strategy can be adjusted when necessary.

[0115] The SOH attenuation rate prediction signal 730 is finally transmitted to the active power loop controller 9; the active power loop controller 9 dynamically adjusts the primary frequency modulation droop coefficient, secondary frequency modulation compensation coefficient, etc. in combination with this attenuation rate information and the grid frequency deviation situation, effectively reducing the economic loss caused by battery attenuation while meeting the rapid grid frequency regulation requirements.

[0116] When the normal output signal of the battery temperature comparator 71 triggers the SOH prediction controller 72 and the SOH attenuation rate prediction signal 730 is obtained through the differential controller 73, this signal is input to the active power loop primary frequency modulation or secondary frequency modulation circuit of the active power loop controller 9.

[0117] Through this signal 730, the frequency modulation strategies such as the droop parameter and the adaptive compensation gain can be appropriately adjusted according to the real-time change of the battery health state, avoiding high-intensity discharge when the battery is overly attenuated, so as to take into account the life management of the energy storage battery while ensuring the stability of the grid frequency.

[0118] The battery SOH dynamic evaluation controller 7 shown in this embodiment can, on the basis of ensuring the safety of the battery temperature, perform deep learning and prediction on the current and historical SOH data of the battery through the LSTM algorithm, and use a differentiator to obtain the SOH decay rate information in real time. This information and the power grid frequency signal are input into the active power loop controller 9 together, forming an adaptive control closed loop that takes into account both the battery life and the power grid frequency regulation requirements, greatly improving the accuracy and economy of multi-scale frequency regulation, and avoiding the problems of excessive battery consumption or insufficient frequency regulation performance caused by the fixed parameter strategy.

[0119] Specifically, as Figure 3 shown, the active power loop controller 9 includes an adaptive model selection controller 91, a coordinated frequency regulation controller 92, an active power loop primary frequency regulation circuit 93, an active power loop secondary frequency regulation circuit 94, an active power loop subtractor 95, an active power loop adder 96, and an active power loop integrator 97.

[0120] The adaptive model selection controller 91 makes a coordinated processing decision on the frequency signal transmitted by the power calculator 11 and the SOH decay rate prediction signal transmitted by the battery SOH dynamic evaluation controller 7, compares it with the set threshold, and adaptively calculates the frequency regulation output mode signal 910. The coordinated frequency regulation controller 92 splits the transmitted frequency regulation output mode signal 910 to obtain the primary and secondary frequency regulation output ratio signals 920. The active power loop primary frequency regulation circuit 93 multiplies and processes the transmitted angular frequency signal according to the frequency regulation output ratio signal 920 to obtain the primary frequency regulation feedback signal 930. The active power loop secondary frequency regulation circuit 94 performs frequency compensation and integration operations on the primary frequency regulation feedback signal 930 to obtain the active power compensation value 940. Then, the active power compensation value 940 is subtracted from the grid-connected active power set value by the active power loop subtractor 95 to obtain the active power reference value 950. The active power reference value 950 is added to the grid-connected active power by the active power loop adder 96 to obtain the error active power signal 960. The active power loop integrator 97 calculates and processes the transmitted error active power signal into a frequency signal, and after integral calculation, obtains the phase angle reference signal 970, and then outputs the phase angle reference signal 970 to the pulse regulation controller 8.

[0121] Preferably, the adaptive model selection controller 91 obtains the power grid frequency signal from the power calculator 11, and this frequency signal can be compared with the reference frequency to obtain the real-time frequency deviation value (or further calculate derivative quantities such as the frequency change rate and variance according to requirements).

[0122] At the same time, the adaptive model selection controller 91 also receives the SOH decay rate prediction signal output by the battery SOH dynamic evaluation controller 7. This signal represents the healthy decay speed of the battery at the current operating point and can be regarded as an important indicator for measuring the risk of battery life consumption.

[0123] To balance grid frequency stability and the lifespan of energy storage batteries, in this embodiment, the adaptive model selection controller 91 sets several thresholds or rules internally to distinguish different operating conditions. For example:

[0124] When the gap between the grid frequency and the reference frequency is large, it means a stronger frequency modulation effort is required; if the gap is small, a mild strategy can be adopted.

[0125] If the detected battery decay rate is at a high level, the frequency modulation behavior of over-discharge / charge should be restricted to avoid severely shortening the battery lifespan.

[0126] The adaptive model selection controller 91 comprehensively compares the above two types of information and may adopt the following (or more) judgment methods:

[0127] If the frequency deviation is large and the SOH decay rate is low, indicating that the system urgently needs high-power frequency modulation and the battery health status allows, then "strong primary frequency modulation" or "primary-secondary collaborative amplitude increase mode" is preferentially selected;

[0128] If the frequency deviation is small or the SOH decay rate is high, then "low-intensity primary frequency modulation" or "mainly secondary compensation" is preferred to reduce battery wear;

[0129] If both are in the medium range, the economy and real-time demand can be comprehensively evaluated, and the "adaptive primary-secondary hybrid mode" can be adopted.

[0130] One or more preset thresholds can be set inside the adaptive model selection controller: for example, several regions (sub-spaces) are divided in the multi-dimensional (frequency deviation, decay rate) space. The adaptive model selection controller 91 determines which frequency modulation mode to adopt according to the regions where the "deviation amount" and "decay rate" are located, and generates the corresponding mode index or control coefficient.

[0131] In a more advanced scenario, fuzzy logic, neural network or other adaptive algorithms can be combined to perform non-linear mapping on the frequency deviation and SOH decay rate, dynamically adjust the thresholds or control gains, and make the control strategy more intelligent.

[0132] The adaptive model selection controller 91 encapsulates the result after the above collaborative processing into the "frequency modulation output mode signal 910" and sends it to the collaborative frequency modulation controller 92 or other subsequent links.

[0133] The different modes of the frequency modulation output mode signal 910 may correspond to:

[0134] Primary frequency modulation priority: focusing on fast response and high-power output;

[0135] Secondary frequency modulation assistance: performing long-term scale compensation on the basis of primary frequency modulation;

[0136] Primary-secondary coordination: Give full play to the advantages of both simultaneously;

[0137] Battery SOH protection mode: When the attenuation rate is higher than the set threshold, limit the frequency regulation output or compensate with other resources to ensure the battery life.

[0138] With the dynamic changes of grid conditions, load demands, battery attenuation rates and other conditions, the adaptive model selection controller 91 can switch the optimal frequency regulation mode in real time, enabling the system to quickly respond to large fluctuations and avoid excessive battery consumption under stable conditions. Try to maintain frequency stability under large load shocks, reduce the high-intensity call on the battery when unnecessary, extend the battery life and reduce the system operation cost.

[0139] The "threshold comparison" and "coordination processing" processes of this embodiment can be extended or parameter-tuned according to different battery types, different grid levels or richer monitoring data, such as adding battery SOC (state) constraints, temperature constraints, environmental parameters, etc., to further enhance the system's adaptive ability.

[0140] Through the above process, the adaptive model selection controller 91 makes a coordinated judgment on the frequency signal and the SOH attenuation rate signal, and provides a reasonable mode instruction for the active power regulation of the primary frequency regulation, secondary frequency regulation and even the subsequent pulse regulation controller, so as to balance the multi-scale frequency regulation effect and the energy storage battery life management.

[0141] Preferably, the coordinated frequency regulation controller 92 first receives the "frequency regulation output mode signal 910" output from the adaptive model selection controller 91. This signal has comprehensively considered the frequency information of the power calculator 11 and the SOH attenuation rate signal of the battery SOH dynamic evaluation controller 7, and determines the frequency regulation mode that the system should adopt (such as "high-intensity primary frequency regulation first", "primary-secondary coordination", "secondary frequency regulation compensation mainly", etc.) through the adaptive judgment of preset thresholds or rules.

[0142] A set of "mode-ratio" mapping relationships (such as look-up tables, fuzzy rules or adaptive algorithms) can be stored or accessed inside the coordinated frequency regulation controller 92. According to the frequency regulation mode indicated by the frequency regulation output mode signal 910, the distribution weights or output ratios of primary and secondary frequency regulations are thus parsed. For example:

[0143] Mode A (primary frequency regulation first): When the grid frequency deviation is large and the battery SOH state is good, give priority to strengthening the primary frequency regulation power output. The weight of primary frequency regulation may be allocated 80% - 90%, and only 10% - 20% is left for secondary frequency regulation for subsequent refined compensation;

[0144] Mode B (primary and secondary coordinated operation): When the grid frequency deviation is in the medium range or the battery SOH has certain limitations, the primary frequency regulation and the secondary frequency regulation are coordinated with a relatively balanced output ratio (such as 50%:50%);

[0145] Mode C (secondary frequency regulation dominant): In the scenario where the frequency deviation is small but long-term stability needs to be maintained, the primary frequency regulation maintains a low gain, allowing the secondary frequency regulation to perform the main power compensation to reduce the rapid impact on the battery health and protect the energy storage life.

[0146] After completing the mode analysis, the coordinated frequency regulation controller 92 integrates the calculated or looked-up primary and secondary frequency regulation distribution coefficients into the output ratio signal 920, and transmits it to the active power loop primary frequency regulation circuit 93 and the active power loop secondary frequency regulation circuit 94 respectively. The output ratio signal 920 can adopt various forms:

[0147] Scalar coefficient: Such as the primary frequency regulation weight and the secondary frequency regulation weight, and ensure that the sum of the two is 1;

[0148] Independent reference value: Such as separately giving the primary frequency regulation reference and the secondary frequency regulation reference;

[0149] Logic identifier: Indicates which frequency regulation method dominates at the current moment, or the proportion in which the two cooperate to output.

[0150] If the grid frequency deviation or the battery SOH attenuation rate signal changes significantly, the adaptive model selection controller 91 can output a new frequency regulation output mode signal 910 in real time; the coordinated frequency regulation controller 92 then re-analyzes the mode and generates an updated distribution weight to ensure that the system can be switched to a more appropriate frequency regulation ratio in a timely manner.

[0151] The coordinated frequency regulation controller 92 dynamically allocates the primary and secondary frequency regulations according to the actual working conditions, so that it can quickly suppress the frequency fluctuation in the large deviation scenario, and can also reduce the excessive impact on the energy storage battery in the small deviation or long-term operation scenario. By real-time sensing the SOH attenuation rate, the system reduces the unnecessary consumption of the battery life while ensuring efficient frequency regulation, and improves the overall frequency regulation economic benefit. The active power loop primary frequency regulation circuit 93 processes the fast and short-term fluctuations, and the active power loop secondary frequency regulation circuit 94 performs the long-term compensation. The two work together with adjustable weights, which can better match the multi-scale regulation requirements of the power grid.

[0152] Preferably, as Figure 4 shown, the active power loop adaptive mode selection controller 91 includes a variance calculator 911, a state latch 912, a first divider 913, and a first adaptive controller 914.

[0153] The variance calculator 911 receives the grid frequency signal and the reference grid frequency signal from the power calculator 11.

[0154] First, subtract the two to obtain the frequency deviation Δf, and then perform a differential operation or other high-order operations on this deviation signal to obtain its dynamic characteristics.

[0155] In this embodiment, in order to more accurately describe the dispersion degree or volatility of the large / small frequency difference, similar variance or square operations can be performed on Δf and its change rate to obtain the frequency error variance signal 9110. This frequency error variance signal 9110 can better reflect the degree and duration of the large-scale fluctuation of the grid frequency than a single frequency error, and is an important basis for subsequent judgment of the frequency modulation intensity and battery attenuation risk.

[0156] The status latch 912 receives the SOH attenuation rate prediction signal from the battery SOH dynamic evaluation controller 7. Before input, it can be screened by threshold comparison to determine whether to latch.

[0157] When it is detected that the SOH attenuation rate prediction signal exceeds a certain preset threshold (for example, indicating that the battery attenuation speed is too fast or the life consumption is excessive), the status latch 912 can latch it as the SOH attenuation rate signal 9120 and hold it until the next update or manual reset. This latching process can avoid jitter in the frequency modulation mode decision caused by the rapid change of the SOH information, and at the same time facilitate specific mode protection in special situations (such as abnormal battery temperature).

[0158] The status latch was previously in the latched state. After the SOH attenuation rate prediction signal is continuously lower than the lower limit of the threshold hysteresis and meets the shortest timing condition, it will be automatically or manually confirmed to be reset, and the power output and droop parameters will be gradually restored according to the preset soft recovery curve to avoid stress on the battery cells caused by large power mutations.

[0159] When the SOH attenuation rate prediction signal does not exceed the preset threshold and the status latch was not previously in the latched state, there is no need to start the latching mechanism; the system will perform regular sampling and processing on this signal with "real-time update", and send the real-time SOH attenuation rate prediction signal as the SOH attenuation rate signal 9120 to the subsequent module, giving sufficient flexibility to the subsequent frequency modulation control loop.

[0160] The first divider 913 is used to divide the frequency error variance signal 9110 by the SOH attenuation rate signal 9120 to obtain the frequency modulation-battery life impact signal 9130 as shown in the following formula:

[0161]

[0162] where δ is the frequency modulation-battery life impact signal 9130, σ 2 is the frequency error variance signal 9110, SOH d is the SOH attenuation rate signal 9120

[0163] The design idea of this embodiment is as follows: the larger the frequency error variance value, the more severe the power grid fluctuation is characterized, and the higher the demand for frequency modulation; while the larger the SOH attenuation rate value, the more significant the battery health loss is characterized, and it is not suitable for high-intensity frequency modulation output.

[0164] Through a division operation, a comprehensive index 9130 can be obtained, which intuitively reflects "the degree of battery life loss that may be caused by large frequency differences". If the result is large, it means that the demand for frequency modulation is relatively urgent and the battery attenuation rate is also low (or vice versa), which is beneficial to increasing the frequency modulation intensity; if the result is small, it may represent a high battery attenuation rate or an insignificant frequency difference, and the frequency modulation intensity needs to be restricted.

[0165] The first adaptive controller 914 receives the frequency modulation-battery life impact signal 9130 and compares it with the preset threshold or multi-threshold rule inside or performs fuzzy processing.

[0166] When the frequency modulation-battery life impact signal 9130 is higher than a certain threshold (or falls within a certain sub-interval), it can be determined that the system is beneficial to enhancing frequency modulation and outputs a "high-intensity frequency modulation mode signal"; when 9130 is low, it tends to reduce the frequency modulation intensity or select a "primary-secondary collaborative compensation" mode to relieve the battery burden.

[0167] The first adaptive controller 914 generates a frequency modulation output mode control signal 910 based on the above judgment result. This mode control signal 910 is sent to the coordinated frequency modulation controller 92 or other control modules for further splitting or distributing primary and secondary frequency modulation powers.

[0168] If the battery attenuation rate surges, it means that the energy storage system is in a large health risk area, and the first adaptive controller 914 can forcibly reduce the frequency modulation power output or activate the corresponding protection logic; if the power grid frequency variance increases significantly and the battery is still in a relatively healthy state, the primary frequency modulation intensity is increased or a high-gain mode is executed to quickly stabilize the frequency. The specific switching mode is as follows:

[0169]

[0170] where θ0 and θ1 are the set upper and lower limit thresholds respectively

[0171] In this embodiment, through the division combination of the frequency error variance signal 9110 and the SOH attenuation rate signal 9120, the first adaptive controller 914 can timely select an efficient frequency modulation mode or a conservative frequency modulation mode, taking into account grid safety and battery service life. The status latch 912 performs a latch process after comparing with key thresholds, which can effectively reduce the frequent mode switching caused by small fluctuations of the SOH signal and improve the stability of system regulation. When facing large-scale or rapid frequency changes, if the battery SOH attenuation rate is within the safe range at this time, the first adaptive controller 914 will output a more powerful frequency modulation output mode control signal 910, and the primary frequency modulation circuit and the secondary frequency modulation circuit cooperate with each other to significantly shorten the frequency convergence time and improve the system inertia support.

[0172] Through the cooperation of the above-mentioned modules, while capturing the severity of the grid frequency difference and the state of the battery attenuation rate, the adaptive model selection controller 91 performs comprehensive quantitative calculations on the two, and outputs a more targeted frequency modulation output mode control signal 910, providing technical support for large-scale energy storage applications to balance economy and stability in multi-scale frequency modulation.

[0173] Preferably, as Figure 5 shown, the active loop collaborative frequency modulation controller 92 includes a first subtractor 921, a differentiator 922, a second adaptive controller 923, and a fuzzy controller 924.

[0174] The first subtractor 921 receives the actual grid frequency signal and the reference frequency signal, performs a subtraction operation, and obtains a frequency error signal 9210.

[0175] In this embodiment, the first subtractor 921 can be implemented by using a hardware differential circuit or a software operation method, and the frequency error value is updated in real time to reflect the current operating deviation degree of the grid. This frequency error signal 9210 is one of the reference inputs for subsequent fast and long-term scale frequency modulation decisions.

[0176] The differentiator 922 performs a differentiation process on the frequency error signal 9210 and outputs a frequency error change rate signal 9220, which can evaluate the dynamic characteristics such as the speed and acceleration of the grid frequency change.

[0177] When the grid frequency error soars or drops at a high rate, the frequency error change rate signal 9220 will present a large value, providing a warning message of "short-term overshoot or large-rate fluctuation" for the subsequent control module.

[0178] The second adaptive controller 923 receives the aforementioned frequency error signal 9210 and frequency error change rate signal 9220, and combines the "frequency modulation output mode selection signal" of the upper-level adaptive model selection controller 91 to perform threshold comparison subspace division.

[0179] In this embodiment, the second adaptive controller 923 will enter the corresponding control subspace according to the mode category indicated by the frequency modulation output mode selection signal. For example:

[0180] High-intensity frequency modulation subspace: When both the frequency error and the change rate are large and the battery SOH allows, a larger gain is adopted to achieve rapid frequency stabilization;

[0181] Medium-intensity coordinated frequency modulation subspace: When the frequency deviation is medium or the battery attenuation rate is high, primary and secondary frequency modulation share the load together to balance the response speed and battery life;

[0182] Low-intensity frequency stabilization subspace: When the frequency deviation is small or the battery health condition is poor, the controller 923 outputs a relatively smooth frequency modulation command.

[0183] In this embodiment, within different subspaces, the second adaptive controller 923 can select different control gain mapping values based on the frequency error signal 9210 and the change rate signal 9220 and the thresholds determined by experimental experience, and calculate the coordinated frequency modulation control signal 9230 as shown in Table 1:

[0184] Table 1 Coordinated frequency modulation control signal output mechanism table

[0185]

[0186] The coordinated frequency modulation control signal 9230 is a mapping value, which reflects the comprehensive adjustment intensity of the current frequency difference, and then is handed over to the fuzzy controller 924 for more refined fuzzy decision-making.

[0187] The fuzzy controller 924 receives the coordinated frequency modulation control signal 9230 output by the second adaptive controller 923 and the SOH attenuation rate signal 730, and performs fuzzy inference operations on them according to the preset fuzzy sets and fuzzy rules, and outputs the output weight ratio signal 920 of the two frequency modulations.

[0188] In this embodiment, in order to make the primary frequency modulation and the secondary frequency modulation work together on multiple time scales, the coordinated frequency modulation control signal and the corresponding threshold difference of the coordinated frequency modulation control signal are respectively set: including PB (positive big), PS (positive small), ZO (zero), NS (negative small) and NB (negative big), and their values can be set according to actual needs. The numerical values corresponding to the high, medium and low gears of the frequency modulation weight can also be set according to actual needs.

[0189] The fuzzy rule setting of this embodiment is specifically shown in Table 2.

[0190] Table 2 Fuzzy rule base

[0191]

[0192] Among them, the following membership functions are used to map each input and output parameter:

[0193] Control signal: Gaussian membership functions are used for PB and NB, and triangular membership functions are used for PS, ZO, and NS.

[0194] SOH prediction signal: Trapezoidal membership function.

[0195] Weight output signal: Gaussian membership function.

[0196] Through the non-linear mapping of the fuzzy controller 924, it can flexibly handle various working conditions: both taking into account fast response in large disturbance scenarios and taking into account battery life and frequency accuracy in small disturbance or long-term regulation scenarios.

[0197] The finally generated output weight ratio signal 920 will be respectively allocated to the primary frequency modulation circuit 93 of the active power loop and the secondary frequency modulation circuit 94 of the active power loop, indicating the proportion of frequency modulation power that should be borne at the current moment. The frequency modulation ratio signal 920 is composed of the primary frequency modulation weight signal k0 delivered to the primary frequency modulation circuit and the secondary frequency modulation weight signal k1 delivered to the secondary frequency modulation circuit. The relationship between the two is shown in the following formula:

[0198] k0 = 1 - k1

[0199] The primary frequency modulation circuit 93 of the active power loop mainly focuses on quickly suppressing large fluctuations in a short time, while the secondary frequency modulation circuit 94 of the active power loop corrects the residual deviation on a longer time scale, improves the steady-state performance of the system, and realizes multi-scale adaptive collaborative frequency modulation.

[0200] In this embodiment, the second adaptive controller 923 performs subspace division in combination with the frequency modulation output mode selection signal, so that the control gain can be flexibly adjusted according to the scale of the frequency error and the battery state. The fuzzy controller 924 performs non-linear mapping on the collaborative frequency modulation control signal to generate a reasonable weight distribution for primary / secondary frequency modulation, avoiding the difficulty of a single fixed parameter in taking into account various working conditions. Through comprehensive consideration of multiple scenarios such as large deviations, high-speed changes, small deviations, and long-term compensation, accurate steady-state control of the grid frequency and long-life operation of the energy storage system are realized.

[0201] Generally speaking, in this embodiment, the active power loop collaborative frequency modulation controller 92 can perform hierarchical and multi-scale processing on the grid frequency deviation signal by the mutual cooperation of the first subtractor 921, the differentiator 922, the second adaptive controller 923, and the fuzzy controller 924, deeply integrate the frequency modulation output with the health management of the energy storage battery, effectively extend the battery service life while ensuring fast frequency stabilization, and improve the safety and economy of the large power grid system.

[0202] Specifically, asFigure 6 As shown in Figure 6 , the active loop primary frequency modulation circuit 93 includes a second subtractor 931, a third adaptive controller 932, and a multiplier 933.

[0203] The second subtractor 931 subtracts the grid voltage frequency from the target frequency and outputs a grid voltage frequency error signal 9310; this error signal 9310 is used to measure the current deviation amplitude of the grid and lays the foundation for subsequent judgment of "large frequency deviation" and "small frequency deviation" and calculation of the adaptive droop coefficient.

[0204] The third adaptive controller 932 determines the current frequency modulation mode (such as high-intensity primary frequency modulation priority, primary-secondary coordination, etc.) according to the grid voltage frequency error signal 9310, and further calculates the droop coefficient in the "large frequency deviation" or "small frequency deviation" region based on the frequency modulation ratio signal 920.

[0205] If the grid voltage frequency error signal 9310 exceeds the set threshold, the system determines that it enters the "large deviation subspace"; if the grid voltage frequency error signal 9310 △f ac is lower than the threshold, it is determined as the "small deviation subspace" to call different droop calculation functions respectively.

[0206] When the grid voltage frequency error signal 9310 is large and the ratio signal 920 shows high-intensity primary frequency modulation priority at the same time, the third adaptive controller 932 selects the Sigmoid function to calculate the droop coefficient to achieve fast response:

[0207]

[0208] where K P，max is the maximum value of the droop coefficient in the basic stable state of the system (determined by specific system parameters, generally taken as 1), Δf is the grid frequency deviation, Δf0 is the offset threshold (generally 0.5Hz), and α is the parameter for adjusting the steepness of the curve

[0209]

[0210] where k0 is the weight reference coefficient, K p is the finally used droop coefficient.

[0211] When the absolute value of the grid voltage frequency error signal 9310 is larger, the Sigmoid function tends to the saturation region, and a higher droop coefficient can be output to amplify the primary frequency modulation force and significantly suppress the frequency fluctuation in a short time; when the grid voltage frequency error signal 9310 returns to the middle and small intervals, the Sigmoid value converges to the middle value, which helps to avoid over-regulation.

[0212] When the grid voltage frequency error signal 9310 is within a small range or the frequency modulation ratio signal 920 indicates primary-secondary coordination, the system pays more attention to battery life and fine-tuning accuracy, and a Gaussian function can be used to calculate the droop coefficient:

[0213]

[0214] Among them, K P,base is the maximum value of the droop coefficient in the basic stable state of the system (determined by specific system parameters, generally taken as 1), Δf is the grid frequency deviation, μ is the mean value (generally taken as the target frequency point 0Hz), and β is used to control the width of the Gaussian curve (generally taken from 0.1 to 0.3)

[0215]

[0216] When the frequency difference is small, the Gaussian function is close to 1, and the frequency modulation force is relatively mild, reducing the large impact on the battery;

[0217] If the frequency difference increases, the output of the Gaussian function drops rapidly, which can avoid excessive discharging / charging within a range of fluctuations that are not yet severe, taking into account both response and life.

[0218] When selecting the above function and calculating the parameters, the third adaptive controller 932 can also fine-tune parameters such as α, β, k0, etc. with reference to the battery SOH attenuation rate or the state of the energy storage capacity. For example: when the battery SOH attenuation rate is high, slightly reduce α or β to reduce the overall frequency modulation force; when the battery health condition is excellent, α or β can be increased to enhance the frequency modulation ability.

[0219] In the large / small deviation subspaces, the ratio signal 920 will also perform more refined tuning on the key parameters of the Sigmoid or Gaussian function:

[0220] Increase or decrease the steepness of α: when the ratio signal 920 shows "primary frequency modulation is absolutely prioritized", α can be moderately increased to make the Sigmoid function steeper and amplify the frequency modulation force under large deviations; if "primary frequency modulation only makes a small contribution", then slightly reduce α to make the Sigmoid curve flatter.

[0221] Adjust the width of β: when the ratio signal 920 is medium and primary frequency modulation needs to participate flexibly within a certain range, β can be slightly increased to keep the Gaussian function stably output within the small deviation range; if the ratio signal 920 is low, then reduce β to further reduce the primary frequency modulation intensity.

[0222] Dynamic change of the reference coefficient k0: if the system detects that the battery SOH attenuation rate is high or the number of grid fluctuations is too frequent, the ratio signal 920 may indicate conservative output of primary frequency modulation. At this time, k0 can also be adjusted downwards to overall reduce the level of the droop coefficient.

[0223] The multiplier 933 obtains the optimal droop coefficient signal 9320 output from the third adaptive controller 932 and the grid voltage frequency error signal 9310, and performs a multiplication operation to generate a primary frequency modulation feedback signal 930 as shown in the following formula:

[0224] ΔP primary =-K P ·Δf

[0225] In this embodiment, the Sigmoid function and the Gaussian function are respectively adopted for large deviations and small deviations to adaptively adjust the droop coefficient, avoiding the problems of efficiency or life loss caused by the "one-size-fits-all" approach of traditional fixed droop coefficients for large / small disturbances. At large frequency deviations, the Sigmoid curve can respond quickly and significantly reduce frequency mutations; at small frequency deviations, the Gaussian curve ensures fine adjustment and reduces the impact on the battery. During the frequency modulation process, the frequency modulation strength can be adaptively limited or enhanced according to the battery SOH decay rate or capacity state, making the system operation more economical and sustainable. The primary frequency modulation loop realizes short-term fast frequency stabilization and small-amplitude fine control in a Sigmoid / Gaussian manner, and also provides a more ideal initial condition for the subsequent secondary frequency modulation circuit, realizing coordinated frequency modulation on multiple time scales.

[0226] Through the above technical solutions, the active loop primary frequency modulation circuit 93 can quickly suppress the grid frequency fluctuation under large deviations and maintain frequency stability in a gentle manner under small deviations, greatly improving the comprehensive performance of system regulation and the service life of energy storage batteries, and meeting the high requirements for frequency modulation efficiency and economy in the large-scale new energy grid-connected environment.

[0227] Specifically, as Figure 7 shown, the active loop secondary frequency modulation circuit 94 includes a second divider 941, an adaptive compensator 942, an integrator 943, and an adder 944.

[0228] The second divider 941 receives the primary frequency modulation feedback signal 930, performs reverse processing on the signal, and outputs an adaptive droop parameter signal 9410 through division by the grid voltage frequency error signal.

[0229] In this embodiment, the primary frequency modulation feedback signal 930 usually contains short-term fast response information "after multiplying by the droop coefficient"; by performing division processing on it, the "actual contribution level of the droop parameter" or "the potential for continued compensation" can be relatively extracted.

[0230] When the primary frequency modulation feedback signal 930 is large, the output adaptive droop parameter signal 9410 of the second divider 941 is often high, indicating that the primary frequency modulation plays a large role at a high rate; if the signal 930 is small, it means that the primary frequency modulation gain is limited and secondary frequency modulation is required to further increase the compensation.

[0231] The adaptive compensator 942 receives two types of key inputs:

[0232] The adaptive droop parameter signal 9410, the quantized information about the primary frequency regulation intensity transmitted from the second divider 941;

[0233] The frequency modulation output ratio signal 920, which is used to distinguish between "large frequency conversion cases" and "small frequency conversion cases" or other modes, indicates the power distribution shares that primary frequency regulation and secondary frequency regulation should bear respectively at the current moment, so as to divide the subspaces.

[0234] When the frequency modulation output ratio signal 920 shows that secondary frequency regulation accounts for a higher proportion or "secondary frequency regulation needs to play a major compensation role", it indicates that the system may be in a "large frequency conversion scenario": If the primary frequency regulation feedback signal 930 is also large, it often means that there are obvious deviations in the power grid, and secondary frequency regulation needs to accelerate convergence in the short term; if primary frequency regulation has quickly suppressed most of the deviations, secondary frequency regulation can decide to retain a higher gain according to the frequency modulation output ratio signal 920, so as to take into account the battery life while continuing to eliminate the residual deviations.

[0235] On the contrary, when the frequency modulation output ratio signal 920 tends to "primary frequency regulation dominates" or the proportion of "primary and secondary cooperation" is in the middle, secondary frequency regulation is more inclined to make fine corrections to small deviations in the long term, and slow down the energy storage attenuation problem caused by over-discharge / charge.

[0236] According to the proportion of secondary frequency regulation indicated by the frequency modulation output ratio signal 920, and in combination with the analysis of the "primary frequency regulation feedback signal 930", the adaptive compensator (942) can divide secondary frequency regulation into a "large frequency conversion subspace" or a "small frequency conversion subspace":

[0237] Large frequency conversion subspace: When the weight of secondary frequency regulation in the frequency modulation output ratio signal 920 is high, and the primary frequency regulation feedback signal 930 indicates that there is still a large frequency difference that has not been completely suppressed at present, the adaptive compensator 942 will correspondingly increase the gain of the adaptive PI control parameter signal 9420, prompting secondary frequency regulation to perform more powerful corrective actions in a short time;

[0238] Small frequency conversion subspace: If the frequency modulation output ratio signal 920 does not assign too large a share to secondary frequency regulation (or primary frequency regulation has significantly reduced the frequency difference), the adaptive compensator 942 will relatively reduce the adaptive PI control parameter signal 9420, and adjust the frequency smoothly with a longer time constant to reduce large-scale operations of the energy storage.

[0239] Optimal control principle: Within each subspace, the frequency modulation output ratio signal 920 can further refine the upper and lower limits or gain adjustment rules of the adaptive PI control parameter signal 9420. For example:

[0240] If the frequency modulation output ratio signal 920 determines from the outside that the "battery SOH state is good" in the large frequency conversion mode, it is allowed to increase the gain of the adaptive PI control parameter signal 9420 and shorten the integral time constant in the short term;

[0241] If it is detected that the frequency modulation output ratio signal 920 indicates that "the battery attenuation rate is high and the discharge depth needs to be controlled", even in the large frequency conversion scenario, the compensation strength of the adaptive PI control parameter signal 9420 will be moderately slowed down to prevent irreversible damage to the battery.

[0242] In this embodiment, the adaptive compensator 942 divides the sub-space according to the frequency modulation output ratio signal 920 and in combination with the SOH signal priority principle as shown in Table 3.

[0243] Table 3 SOH signal priority principle

[0244]

[0245] Perform compensation calculation according to the transmitted adaptive droop parameter signal 9410. While considering the droop strength used for primary frequency modulation, it also takes into account the current power grid fluctuation size and battery health economy, and processes the adaptive PI control parameter signal 9420 (including the proportional coefficient K I Yes, integral coefficient T I ) as shown in the following formula:

[0246] K I = K I,base ·k1·(1 - α·max(0, 0.8 - SOH))

[0247] K I，base = K base - K p

[0248]

[0249] T I，base = T base

[0250] Among them, K base 、T base are the proportional gain and integral time reference value respectively, α is the SOH attenuation influence coefficient (generally 0.3), β is the integral time compensation coefficient (generally 0.1), and k1 is the secondary frequency modulation weight factor; SOH represents the SOH prediction signal 720.

[0251] Secondly, ensure the basic response performance of the system through the upper and lower limit threshold comparison as shown in the following formula:

[0252] K I ≥0.2·KI，base

[0253] T I ≥3·T I，base

[0254] The former prevents response slowness, and the latter prevents slow integration.

[0255] The integrator 943 performs proportional-integral operation on the adaptive PI control parameter signal 9420 and applies it to the grid frequency error signal to obtain the compensated active power reference value 9430:

[0256]

[0257] In this embodiment, the output of the integrator 943 can not only correct the remaining frequency deviation after primary frequency modulation but also maintain the convergence of the frequency to the reference value on a long-time scale.

[0258] Under different working conditions, since the PI parameters are dynamically adjusted according to the grid state and battery life requirements, the secondary frequency modulation can flexibly balance between "accelerating convergence" and "reducing battery load".

[0259] The adder 944 adds the compensated active power reference value 9430 and the primary frequency modulation feedback signal 930 output by the active power loop primary frequency modulation circuit 93 to generate the adaptive frequency modulation feedback signal 940.

[0260] The primary frequency modulation has achieved short-time rapid suppression at the front end, while the secondary frequency modulation deeply corrects the remaining or persistent frequency deviation; the adder 944 enables the final active power output to have stronger resistance to high-frequency / low-frequency changes.

[0261] The adaptive frequency modulation feedback signal 940 is usually sent to the active power loop controller (9) or directly connected to subsequent links (such as grid-connected power controller, phase angle reference signal generation unit, etc.) to complete the closed-loop regulation of active power on multiple time scales.

[0262] Compared with the short-time strong response of the primary frequency modulation in this embodiment, the secondary frequency modulation further accurately compensates for the frequency deviation under the action of the integrator 943 to ensure that the power system maintains the target frequency for a long time. The adaptive compensator 942 dynamically distinguishes between large / small frequency conversion scenarios according to the "frequency modulation output ratio signal 920" and the "primary frequency modulation feedback signal 930", and adjusts the PI parameters accordingly according to the environment and battery life requirements. Increasing the gain for large frequency conversion can quickly restore the system frequency, and reducing the gain for small frequency conversion can avoid excessive battery loss and frequent fluctuations of the frequency. On the premise that the primary frequency modulation has adaptively adjusted the droop parameters, the secondary frequency modulation can use a more gentle and detailed compensation strategy to reduce the additional burden on the battery and achieve the comprehensive optimization of frequency modulation benefits and life costs.

[0263] In summary, based on the adaptive droop parameter adjustment of the primary frequency modulation circuit, the active loop secondary frequency modulation circuit 94 described in this embodiment incorporates large / small frequency conversion conditions into the overall frequency modulation model through the organic cooperation of the second divider 941, the adaptive compensator 942, the integrator 943, and the adder 944, dynamically adjusts the proportional-integral control parameters, and further extends the service life of the energy storage device while satisfying the fast and stable frequency. This technical solution is particularly applicable to large-scale new energy grid-connected scenarios, providing high-precision and low-latency frequency support and ensuring the continuous and stable operation of the power system.

[0264] As Figure 8 shown, compared with the traditional control method, the multi-scale adaptive collaborative frequency modulation control proposed by the present invention can provide greater frequency support when disturbed, making the system frequency converge faster and improving the stability of the system.

[0265] Embodiment 2

[0266] The present invention also provides a multi-scale adaptive collaborative grid-forming energy storage frequency modulation control method, which is implemented based on the system described in the above technical solution, and includes the following steps:

[0267] A constant-voltage energy storage power supply, whose output terminal is connected to the DC input terminal of the three-phase inverter circuit, providing DC electrical energy;

[0268] A three-phase inverter circuit, whose AC output terminal is connected to the power grid through a filter inductor, a filter capacitor, and an equivalent impedance of the power grid in sequence, converting DC electrical energy into AC electrical energy and outputting or absorbing electrical energy to the power grid;

[0269] A power calculator, electrically connected to the output side of the three-phase inverter circuit, calculating the active power and reactive power output by the three-phase inverter circuit to the power grid;

[0270] A battery SOH dynamic evaluation controller, monitoring and evaluating the health state of the constant-voltage energy storage power supply and outputting an SOH decay rate signal;

[0271] An active loop controller, generating multi-scale adaptive frequency modulation data based on the SOH decay rate signal and the power grid frequency information and sending it to the pulse regulation controller;

[0272] A reactive loop controller, generating reactive power regulation data according to the power grid voltage deviation signal and sending it to the pulse regulation controller;

[0273] A pulse regulation controller, receiving the control data output by the active loop controller and the reactive loop controller, performing fast pulse regulation in combination with the transient changes of the power grid, and outputting the final pulse regulation instruction and multi-scale collaborative control command to the three-phase inverter circuit to achieve stable regulation of the power grid frequency and voltage.

[0274] The content not described in detail in this specification belongs to the prior art well-known to those of ordinary skill in the art.

Claims

1. A grid-forming energy storage frequency modulation control system based on multi-scale adaptive collaboration, characterized in that: Comprising: A constant - voltage energy - storage power supply, whose output terminal is connected to the DC input terminal of the three - phase inverter circuit, for providing DC electrical energy; A three - phase inverter circuit, whose AC output terminal is connected to the power grid through a filter inductor, a filter capacitor and an equivalent impedance of the power grid in sequence, for converting DC electrical energy into AC electrical energy and outputting or absorbing electrical energy to the power grid; A power calculator, electrically connected to the output side of the three - phase inverter circuit, for calculating the active power and reactive power output by the three - phase inverter circuit to the power grid; A battery SOH dynamic evaluation controller, for monitoring and evaluating the health state of the constant - voltage energy - storage power supply and outputting an SOH decay rate signal; An active - loop controller, for generating multi - scale adaptive frequency - modulation data based on the SOH decay rate signal and grid frequency information and sending it to the pulse regulation controller; A reactive - loop controller, for generating reactive - power regulation data according to the grid voltage deviation signal and sending it to the pulse regulation controller; A pulse regulation controller, for receiving the control data output by the active - loop controller and the reactive - loop controller, performing fast pulse regulation in combination with grid transient changes, and outputting the final pulse regulation instruction and multi - scale collaborative control command to the three - phase inverter circuit to achieve stable regulation of the grid frequency and voltage.

2. The multi-scale adaptive collaborative network-constructing energy storage frequency modulation control system according to claim 1, wherein The battery SOH dynamic evaluation controller includes: A battery temperature comparator, for comparing the detected battery temperature with a preset temperature threshold and outputting a temperature comparison signal; An SOH prediction controller, for estimating the battery SOH value according to the temperature comparison signal and the operating parameters of the battery; A differential controller, for performing differential operation on the SOH value and outputting an SOH decay rate prediction signal; Wherein, the SOH decay rate prediction signal is sent to the active - loop controller, so that the frequency - modulation control process can be adaptively adjusted according to the battery health state.

3. The multi-scale adaptive collaborative network-constructing energy storage frequency modulation control system according to claim 2, wherein The SOH prediction controller includes: A battery normal - operation response trigger module, for receiving and identifying the battery normal - operation response signal generated by the battery temperature comparator, and starting the LSTM prediction algorithm module to perform SOH prediction after confirming that the battery operating temperature is in the normal state; An LSTM prediction algorithm module, for obtaining the real - time SOH state and historical SOH state data of the battery, and cooperatively processing the data through a trained long - short - term memory neural network to output the predicted battery SOH value; Wherein, the battery temperature comparator compares the detected battery temperature with the preset temperature threshold and selects to output the battery normal - operation response signal according to the comparison result.

4. The multi-scale adaptive collaborative network-constructing energy storage frequency modulation control system according to claim 1, wherein The active - loop controller includes: An adaptive model selection controller, for outputting a corresponding frequency - modulation model or mode according to the grid frequency information and the SOH decay rate signal; An active - loop collaborative frequency - modulation controller, for allocating and coordinating primary frequency modulation and secondary frequency modulation according to the output of the adaptive model selection controller; An active - loop primary frequency - modulation circuit, for receiving a primary frequency - modulation control signal and realizing short - time fast frequency - modulation support; An active - loop secondary frequency - modulation circuit, for compensating the residual deviation of primary frequency modulation on a longer - time scale; Among them, the active loop controller sends the integrated multi-scale frequency modulation data to the pulse regulation controller.

5. The multi-scale adaptive collaborative network-constructing energy storage frequency modulation control system according to claim 4, wherein The adaptive model selection controller includes: A variance calculator, which subtracts the grid frequency signal from the reference grid frequency signal and performs differential processing to obtain a frequency error variance signal; A state latch, which performs threshold comparison and latching on the SOH attenuation rate prediction signal to obtain an SOH attenuation rate signal; A first divider, which divides the frequency error variance signal by the SOH attenuation rate signal to obtain a frequency modulation-battery life impact signal; A first adaptive controller, which compares the frequency modulation-battery life impact signal with a preset threshold to determine a frequency modulation output mode control signal, and outputs the frequency modulation output mode control signal to the active loop coordinated frequency modulation controller to select modes such as primary frequency modulation priority or primary and secondary frequency modulation coordination.

6. The multi-scale adaptive collaborative network-constructing energy storage frequency modulation control system according to claim 5, wherein The active loop coordinated frequency modulation controller includes: A first subtractor, which subtracts the grid frequency signal from the reference frequency signal to obtain a frequency error signal; A differentiator, which performs differential calculation on the frequency error signal to generate a frequency error change rate signal; A second adaptive controller, which adaptively compares the frequency error signal and the frequency error change rate signal according to the frequency modulation output mode selection signal, and outputs a coordinated frequency modulation control signal; A fuzzy controller, which performs fuzzy operation processing on the coordinated frequency modulation control signal and the SOH attenuation rate signal according to a preset fuzzy set and fuzzy rules to obtain a primary and secondary frequency modulation output weight ratio signal; The active loop coordinated frequency modulation controller sends the output weight ratio signal to the active loop primary frequency modulation circuit and the secondary frequency modulation circuit, thereby realizing multi-scale coordinated frequency modulation.

7. The multi-scale adaptive collaborative network-constructing energy storage frequency modulation control system according to claim 6, wherein The active loop primary frequency modulation circuit includes: A second subtractor, which compares the grid voltage frequency with the frequency target signal and outputs a grid voltage frequency error signal; A third adaptive controller, which calculates an optimal droop coefficient signal according to the output weight ratio signal and the grid voltage frequency error signal; A multiplier, which calculates a primary frequency modulation feedback signal according to the optimal droop coefficient signal and the grid voltage frequency error signal; and sends it to the active loop secondary frequency modulation circuit.

8. The multi-scale adaptive collaborative network-constructing energy storage frequency modulation control system according to claim 7, wherein The active loop secondary frequency modulation circuit includes: A second divider, which divides the primary frequency modulation feedback signal by the grid frequency error signal to obtain an adaptive droop parameter signal; An adaptive compensator, which performs compensation calculation according to the frequency modulation output ratio signal and the adaptive droop parameter signal, and outputs an adaptive PI control parameter signal; An integrator, which performs proportional integral operation on the frequency error signal according to the adaptive PI control parameter signal to generate a compensated active power reference value; An adder, which performs frequency compensation and integral operation on the compensated active power reference value and the primary frequency modulation feedback signal to obtain an adaptive frequency modulation feedback signal; Among them, the adaptive frequency modulation feedback signal further compensates for the primary frequency modulation residual deviation on a relatively long time scale, and is integrated by the active loop controller, and then a phase angle reference signal is obtained and output to the pulse regulation controller.

9. The multi-scale adaptive collaborative network-constructing energy storage frequency modulation control system according to claim 8, characterized in that The active loop controller further includes: An active loop subtractor is used to subtract the adaptive frequency modulation feedback signal from the set value of the active power at the point of common coupling to obtain the active power reference value; An active loop adder is used to add the active power reference value to the active power at the point of common coupling to obtain the error active power signal; An active loop integrator is used to calculate and process the error active power signal into a frequency signal, and a phase angle reference signal is obtained after integral calculation.

10. A multi-scale adaptive collaborative grid-forming energy storage frequency modulation control method, characterized in that, Implemented based on the system described in any one of claims 1-9, including the following steps: A constant-voltage energy storage power supply, whose output terminal is connected to the DC input terminal of the three-phase inverter circuit, providing DC electrical energy; A three-phase inverter circuit, whose AC output terminal is connected to the power grid through a filter inductor, a filter capacitor and an equivalent impedance of the power grid in sequence, converting DC electrical energy into AC electrical energy and outputting or absorbing electrical energy to the power grid; A power calculator, electrically connected to the output side of the three-phase inverter circuit, calculating the active power and reactive power output by the three-phase inverter circuit to the power grid; A battery SOH dynamic evaluation controller monitors and evaluates the health state of the constant-voltage energy storage power supply and outputs an SOH attenuation rate signal; An active loop controller generates multi-scale adaptive frequency modulation data based on the SOH prediction state signal, the SOH attenuation rate signal and the power grid frequency information and sends it to the pulse regulation controller; A reactive loop controller generates reactive power regulation data according to the power grid voltage deviation signal and sends it to the pulse regulation controller; A pulse regulation controller receives the control data output by the active loop controller and the reactive loop controller, combines the transient changes of the power grid for fast pulse regulation, and outputs the final pulse regulation instruction and multi-scale collaborative control command to the three-phase inverter circuit to achieve stable regulation of the power grid frequency and voltage.

Images