A method for adaptive control of energy storage battery considering SOH dynamic constraint

Abstract

This invention discloses an adaptive control method for energy storage batteries that takes into account dynamic constraints of State of Health (SOH). The method includes: acquiring the SOH of the energy storage battery in real time; establishing a mapping relationship between SOH and the battery's ohmic internal resistance and polarization characteristics; introducing a neural network to compensate for voltage prediction errors under aging conditions; evaluating the maximum power support capability that the energy storage system can provide to the AC side under the current SOH state; determining the critical energy boundary for system transient synchronization and stability and the safe adjustment range of control parameters; designing a dual self-adjusting adaptive actuator constrained by the energy boundary, adaptively adjusting the virtual inertia and damping coefficient under both physical and stability constraints; and introducing a health correction mechanism based on SOH to dynamically limit the control parameters. This invention achieves optimized adaptive support capability of the energy storage battery throughout its entire lifespan while ensuring safe operation on the DC side and transient synchronization and stability of the system.

Description

Technical Field

[0001] This invention relates to new energy power generation and energy storage control technology, specifically to an adaptive control method for energy storage batteries that takes into account dynamic constraints of SOH (State of Health). Background Technology

[0002] With the energy structure shifting towards low-carbon and clean energy, new energy power generation technologies, represented by wind and solar power, have developed rapidly. However, the output power of new energy sources has significant randomness and intermittency. After large-scale grid connection, the power system exhibits obvious characteristics of "low inertia and weak support," posing a severe challenge to the safe and stable operation of the power grid. Therefore, grid-based energy storage technology, with its ability to simulate synchronous generators and actively support grid frequency and voltage, has become a core component in building new power systems.

[0003] Improving the penetration rate of new energy sources and ensuring their reliable access are the core objectives of grid-based energy storage technology. Grid-based energy storage not only meets the system's requirements for inertia and damping, but also actively provides instantaneous power support when the grid experiences transient fluctuations. The active support performance of the energy storage converter is mainly affected by its control strategy and the physical energy state of the DC side. During grid-connected operation, the converter needs to provide virtual inertia as much as possible to suppress sudden frequency changes while meeting power command tracking requirements. However, in practical applications, energy storage systems face multiple limitations: on the one hand, the new energy grid experiences frequent and severe disturbances, requiring the energy storage converter to frequently participate in high-power deep compensation; on the other hand, the energy storage battery inevitably experiences health degradation during long-term operation.

[0004] The changes in electrochemical characteristics caused by battery aging significantly reduce its DC-side real power support capability, making the DC bus voltage extremely vulnerable during transient processes. Current research, both domestically and internationally, largely focuses on improving AC-side algorithms, often neglecting the dynamic constraints of physical limits on control parameters after battery health declines. Blindly implementing strong inertia support in the later stages of battery aging can easily lead to DC voltage collapse and converter transient instability. Therefore, establishing an advanced adaptive control strategy that considers dynamic constraints based on the evolution mechanism of battery state of health (SOH) to ensure the converter can safely and effectively provide grid-connected support throughout its entire lifespan has become a crucial research area in the field of energy storage control for new power systems. Summary of the Invention

[0005] Purpose of the invention: The purpose of this invention is to provide an adaptive control method for energy storage batteries that takes into account dynamic constraints of state of health (SOH). This method can solve the problem that existing grid-type control technologies ignore the decline in DC-side support capacity caused by the degradation of battery health status, which easily leads to DC voltage collapse and system synchronization instability during transient processes in new energy power grids.

[0006] Technical solution: The present invention provides an adaptive control method for energy storage batteries that takes into account dynamic constraints of State of Health (SOH), comprising:

[0007] Obtain real-time health status (SOH) parameters of the energy storage battery and generate equivalent physical parameters related to the battery aging status.

[0008] The real-time ohmic internal resistance and polarization resistance under the current state of harmonics (SOH) of the battery are calculated based on equivalent physical parameters. The calculated values ​​are input into a pre-constructed RC equivalent model of the battery using a mechanism and data fusion approach. The model outputs the battery output voltage. A neural network error compensation module is introduced to correct the nonlinear error in the battery output voltage under aging conditions, thus obtaining the battery terminal voltage.

[0009] Based on the battery's terminal voltage, real-time ohmic internal resistance, and polarization resistance under the current state of harmonics (SOH), combined with the dynamic balance equation of the DC bus voltage, the maximum output power that the grid-connected energy storage system can provide to the AC side under the current healthy state is calculated.

[0010] A nonlinear state equation for the power angle-frequency of a grid-connected energy storage system is constructed. Using the energy function method, the critical energy and critical energy stability boundary for maintaining transient synchronization stability of the grid-connected energy storage system are determined based on the upper limit of the maximum output power. The critical energy stability boundary is then used as the safe adjustment range of the control parameters of the grid-connected energy storage system.

[0011] The charging and discharging operation mode of the energy storage system is identified in real time, thereby identifying potential transient instability risks and generating transient risk assessment results; and by adjusting the final adaptive virtual reactance, transient limiting of the output current is achieved, thereby protecting the hardware safety of aging batteries.

[0012] Based on the transient risk assessment results, an exponential correction operator reflecting the physical health of the battery is introduced. A dual self-adjusting adaptive actuator is used to saturate and limit the virtual inertia and virtual damping respectively within the safe adjustment range, and output the final control parameters. Based on the final control parameters, a control command for the grid-type converter is generated within the safe adjustment range. This control command performs adaptive adjustment of the virtual inertia coefficient and virtual damping coefficient according to the frequency fluctuation characteristics to achieve adaptive support for the power grid.

[0013] Furthermore, the calculation of the battery's real-time ohmic internal resistance and polarization resistance under the current state of harmonic equilibrium (SOH) based on equivalent physical parameters includes:

[0014] Considering the decrease in conductivity due to battery degradation, a linear mapping relationship between the degree of aging and the physical ohmic internal resistance is established:

[0015] ;

[0016] In the formula, For real-time ohmic internal resistance; The initial ohmic internal resistance; Internal resistance at the end of the lifespan; This refers to the real-time health status of the energy storage battery.

[0017] To simulate the nonlinear characteristic of a sharp increase in polarization resistance during the later stages of aging, the following formula is used:

[0018] ;

[0019] In the formula, For the present The polarization internal resistance is below; The initial polarization resistance, This is the aging sensitivity coefficient.

[0020] Furthermore, the neural network error compensation module corrects the nonlinear error in the battery output voltage under aging conditions to obtain the battery terminal voltage, including:

[0021] A feedforward neural network is introduced to predict polarization residual error, and a data-model fusion-driven terminal voltage prediction equation is constructed. The expression for polarization residual error is a complex function of SOH, charging and discharging current, temperature, and resting time.

[0022] ;

[0023] In the formula, Calculate the difference between the actual measured voltage and the voltage from the physical model; This is a nonlinear mapping function constructed by a feedforward neural network; This refers to the charging and discharging current. For temperature; This refers to the settling time;

[0024] The expression for the data-model fusion-driven terminal voltage prediction equation is as follows:

[0025] ;

[0026] In the formula, Terminal voltage; This is the open-circuit voltage; This refers to the charging and discharging current. This is the polarization voltage.

[0027] Furthermore, the expression for the dynamic balance equation of the DC bus voltage is as follows:

[0028] ;

[0029] in, For DC bus capacitors; This is the DC bus voltage; This represents the battery's output power under its current healthy condition. It provides DC output active power; It outputs active power for AC.

[0030] Furthermore, the maximum output power that the grid-connected energy storage system can provide to the AC side under its current healthy state is calculated using the following formula:

[0031] ;

[0032] in, The maximum power provided by the battery under the current health condition is the upper limit of the maximum output power that the grid-connected energy storage system can provide to the AC side under the current health condition. This is the DC bus voltage; This is the minimum permissible operating voltage threshold for the DC bus. To account for the battery's equivalent total internal resistance after SOH, its value is equal to the sum of the battery's real-time ohmic internal resistance and the polarization internal resistance under the current SOH.

[0033] Furthermore, the power angle-frequency nonlinear state equation of the grid-connected energy storage system is constructed as follows:

[0034] A DC voltage compensation circuit is introduced to correct the grid-type angular frequency in real time.

[0035] ;

[0036] In the formula, Angular frequency; The reference angular frequency; This is the droop coefficient; To output active power in AC mode; The active power setpoint; DC voltage compensation gain; This is the DC bus voltage; This is the reference value for the DC bus voltage;

[0037] Define the power angle-frequency nonlinear state equation of a grid-connected energy storage system:

[0038] ;

[0039] In the formula, δ is the power angle; Angular frequency; The rated angular frequency of the power grid; It is the equivalent inertia coefficient; To simulate the mechanical power input to a generator; This is the equivalent damping coefficient; It outputs active power for AC.

[0040] Furthermore, the critical energy and critical energy stability boundary for the grid-connected energy storage system to maintain transient synchronization are calculated using the following formulas:

[0041] ;

[0042] in, This is the energy function value; For follow The critical energy that is modified due to degradation; To simulate the mechanical power input to the generator; δ is the power angle; This refers to the angular frequency deviation. The initial stable power angle of the grid-connected energy storage system before the occurrence of disturbance; This represents the active power amplitude of the AC side power angle characteristic curve.

[0043] Furthermore, the real-time identification of the charging and discharging operation mode of the energy storage system, thereby identifying potential transient instability risks and generating transient risk assessment results, includes:

[0044] Define operating mode factors :

[0045] ;

[0046] In the formula: A value of 1 represents the discharge mode, and a value of -1 represents the charging mode. To output active power for AC; to assess transient instability risk through quantitative acceleration area:

[0047] ;

[0048] In the formula, To accelerate the area; The initial phase angle; This is the fault clearing angle; To simulate the mechanical power input of a generator; simultaneously, to reduce the transient current limit based on the degree of aging to protect the battery hardware:

[0049] ;

[0050] In the formula, This is the transient current limiting value; γ is the rated current; γ is the sensitivity coefficient; This refers to the real-time health status of the energy storage battery.

[0051] ;

[0052] In the formula, For the final adaptive virtual reactance; Flag is determined by Triggered in conjunction with current over-limit conditions; The virtual reactance is adaptively capped; This is the adaptive lower limit for virtual reactance.

[0053] Furthermore, based on the transient risk assessment results, an exponential correction operator reflecting the physical health of the battery is introduced. A dual self-adjusting adaptive actuator is used to saturate and limit the virtual inertia and virtual damping within the safe adjustment range, respectively, outputting the final control parameters, including:

[0054] Calculate the basic adaptive inertia based on power grid frequency disturbances. :

[0055] ;

[0056] In the formula, It is the static inertia; The first adjustable gain; This is the second adjustable gain; This is for frequency deviation; The frequency change rate is used; an exponential correction operator reflecting the physical health of the battery is introduced. :

[0057] ;

[0058] The virtual inertia is saturated and limited by combining the critical energy stability boundary, and the final adaptive inertia is output. for:

[0059] ;

[0060] In the formula, It is a saturation limiting function; This is the minimum limit for virtual inertia; The virtual inertia upper limit is set; online adjustment of the damping coefficient is performed synchronously.

[0061] ;

[0062] In the formula, This is the final adaptive damping coefficient; The basic damping coefficient; The gain is adjusted for damping. This is for frequency deviation; The initial resistance; The real-time ohmic internal resistance.

[0063] Furthermore, the control commands of the grid-type converter are summarized in the phase angle closed-loop output equation:

[0064] ;

[0065] In the formula, Phase angle; The reference angular frequency; Given power; For the Laplace operator; To output active power in AC mode; This is the final adaptive damping coefficient; For the final adaptive inertia; ω is the angular frequency.

[0066] Beneficial effects: Compared with the prior art, the significant technical effects of the present invention are as follows:

[0067] (1) Deep perception of aging mechanism: In step S2, by dynamically mapping physical impedance through SOH, the limitation of treating energy storage as an ideal power source is broken, and the control system can accurately perceive the battery life decay and the actual power limit.

[0068] (2) Significantly improve DC safety: In steps S2 and S3, DC voltage compensation and power limit are established based on physical constraints, which effectively prevents the DC voltage from collapsing due to "power over-draining" under strong inertia support conditions of aging batteries.

[0069] (3) Establish dynamic stability boundary: In step S4, the SOH constraint is transformed into a dynamic stability boundary using the energy function to ensure that the adaptive algorithm is always within the physical safety red line throughout the battery's entire life cycle.

[0070] (4) Collaborative support and asset protection: In step S6, a health correction operator is introduced to dynamically adjust the support strength, which effectively extends the service life of energy storage assets while meeting the active support needs of the power grid. Attached Figure Description

[0071] Figure 1 This is a schematic diagram of the process of the present invention;

[0072] Figure 2 This is a schematic diagram of the first-order RC circuit model in this invention;

[0073] Figure 3 This is a flowchart of the model-neural network data fusion process in this invention;

[0074] Figure 4 This is a schematic diagram of the grid-type energy storage converter in this invention;

[0075] Figure 5 This is the overall control diagram of VSG in this invention;

[0076] Figure 6 This is a schematic diagram of the structure of the dual self-adjusting adaptive actuator in this invention. Detailed Implementation

[0077] The technical solution of the present invention will now be described in detail with reference to specific embodiments and accompanying drawings.

[0078] like Figure 1 As shown, the present invention provides an adaptive control method for energy storage batteries that considers dynamic constraints on State of Health (SOH). Its core logic lies in transforming the microscopic aging mechanism of the battery into macroscopic grid support boundary constraints, mapping the real-time SOH of the energy storage battery into a physically feasible range of terminal voltage and power, providing a constraint basis for subsequent control parameter adjustment; further transforming the physical constraints into a transient stability domain of the system; and generating restricted control commands within the transient stability domain. The method includes the following steps:

[0079] S1. Obtain the real-time State of Health (SOH) parameters of the energy storage battery and generate equivalent physical parameters related to the battery's aging state. Details are as follows:

[0080] like Figure 3 As shown, the focus is first on the deep perception of the physical characteristics of the energy storage unit. The energy storage management system acquires the real-time State of Health (SOH) of the energy storage battery and uses it as the core variable for evaluating the evolution of physical impedance. During operation, the battery management system first acquires the current SOH and historical SOH evolution information of the battery, and simultaneously collects characteristic data such as battery operating current and voltage.

[0081] Equivalent physical parameters include initial ohmic internal resistance Internal resistance at the end of life Real-time health status of energy storage batteries This is obtained in real time.

[0082] S2. Calculate the real-time ohmic internal resistance and polarization resistance under the current state of harmonics (SOH) of the battery based on equivalent physical parameters; adopt a mechanism and data fusion approach, input the calculated values ​​into a pre-constructed battery RC equivalent model, and the model outputs the battery output voltage; introduce a neural network error compensation module to correct the nonlinear error in the battery output voltage under aging conditions, and obtain the battery terminal voltage.

[0083] like Figure 3 As shown, the specific implementation process of step S2 is as follows:

[0084] S2.1 Calculate the real-time ohmic internal resistance and polarization resistance of the battery under the current state of harmonic equilibrium (SOH) based on the equivalent physical parameters, as follows:

[0085] This invention constructs a first-order RC equivalent physical model of the battery based on State of Health (SOH) sensing. By mapping the SOH to the model parameters, the battery terminal voltage is predicted, thereby obtaining the physically feasible operating range of the battery under its current health condition. Considering the decrease in conductivity due to battery degradation, a linear mapping relationship is established between the degree of aging and the physical ohmic internal resistance.

[0086] (1)

[0087] In the formula, For real-time ohmic internal resistance; The initial ohmic internal resistance; Internal resistance at the end of the lifespan; This refers to the real-time health status of the energy storage battery.

[0088] In addition to ohmic losses, the aging process also significantly exacerbates the polarization phenomenon in the electrochemical reaction process, which not only changes the ohmic internal resistance but also significantly affects the charge transfer resistance. To simulate the nonlinear characteristics of the sharp increase in polarization internal resistance in the later stage of aging, the following formula is used:

[0089] (2)

[0090] In the formula, For the present The polarization internal resistance is below; The initial polarization resistance, This is the aging sensitivity coefficient.

[0091] S2.2. By adopting a mechanism and data fusion approach, the calculated values ​​are input into a pre-constructed battery RC equivalent model, and the model outputs the battery output voltage, as follows:

[0092] like Figure 2 As shown, based on the first-order RC equivalent circuit model, the formula can be obtained:

[0093] (3)

[0094] In the formula, This refers to the battery output voltage. This is the open-circuit voltage; It is a series resistor; Polarization resistor; Polarizing capacitor; This is the polarization voltage across the polarization resistor; This is the current flowing through the battery.

[0095] S2.3. A neural network error compensation module is introduced to correct the nonlinear error in the battery output voltage under aging conditions, and the battery terminal voltage is obtained as follows:

[0096] Because the polarization phenomenon after battery aging is highly nonlinear, the prediction accuracy of traditional physical models drops significantly when the state of harmonic equilibrium (SOH) is below 80%. To further eliminate the nonlinear residuals that traditional physical models cannot capture under aging conditions, this invention introduces a neural network error compensation module to ensure maximum power output under the current healthy state. The effectiveness of the assessment throughout the battery's entire lifespan. For example... Figure 3 As shown, the output of the first-order RC equivalent circuit model is corrected, and a feedforward neural network is introduced to predict the polarization residual error. A data-model fusion-driven terminal voltage prediction equation is constructed. The expression for this error term (i.e., polarization residual error) is a complex function of SOH, charging and discharging current, temperature, and resting time.

[0097] (4)

[0098] In the formula, Calculate the difference between the actual measured voltage and the voltage from the physical model; This is a nonlinear mapping function constructed by a feedforward neural network; This refers to the charging and discharging current. For temperature; This refers to the settling time.

[0099] The expression for the data-model fusion-driven terminal voltage prediction equation is as follows:

[0100] (5)

[0101] In the formula, Terminal voltage; This is the open-circuit voltage; This refers to the charging and discharging current. This is the polarization voltage. Terminal voltage. The constraint value under the current healthy state serves as the physical constraint boundary for the battery terminal voltage.

[0102] Furthermore, from equation (3), the polarization voltage follows the following dynamic equation:

[0103] (6)

[0104] In the formula: U P (0) represents the initial value of the polarization voltage in the model, and τ is the time constant.

[0105] S3. Based on the battery terminal voltage, the battery real-time ohmic internal resistance, and the polarization resistance under the current SOH, combined with the DC bus voltage dynamic balance equation, calculate the maximum output power limit that the grid-connected energy storage system can provide to the AC side under the current healthy state.

[0106] like Figure 3 As shown, the specific implementation process of step S3 is as follows:

[0107] After completing the physical characteristic perception, the system enters the DC-side support capability assessment stage based on the physical equivalent circuit model. This stage focuses on the real-time energy extraction characteristics of the grid control system on the DC side and establishes the DC bus voltage. The dynamic equilibrium equation.

[0108] Since the VSC itself is a switching unit, its own losses can be ignored. The DC input power and AC output power are the same, satisfying the power balance relationship. Furthermore, if the DC / DC converter's own losses are ignored, the DC-side input active power includes the output power of the energy storage system through the DC / DC converter and the power of the DC capacitor, that is:

[0109] (7)

[0110] in, For DC bus capacitors; This is the DC bus voltage; This refers to the battery's output power. It provides DC output active power; It outputs active power for AC.

[0111] Therefore, by incorporating the characteristics of the aged battery into the DC-side dynamic model of the converter, the limiting mechanism of SOH on the grid support capability is analyzed, and the dynamic balance equation of the DC bus voltage is established, the expression of which is as follows:

[0112] (8)

[0113] in, For DC bus capacitors; This is the DC bus voltage; This represents the battery's output power under its current healthy condition. It provides DC output active power; It outputs active power for AC.

[0114] like Figure 4 As shown, the grid-connected energy storage system provides active power support to the AC side through the converter, and its power output is essentially limited by the energy supply capacity of the DC-side batteries and bus. Based on the principle of dynamic balance of DC bus voltage, the equivalent internal resistance considering the state of harmonics (SOH) is introduced into the DC-side power model to derive the maximum upper limit of output power that the system can maintain DC voltage stability under the current aging condition.

[0115] Considering the internal resistance voltage drop caused by battery aging, the actual output power constraint of the battery is:

[0116] (9)

[0117] in, This represents the battery's output power under its current healthy condition. Terminal voltage; This represents the real-time ohmic internal resistance of the battery.

[0118] Therefore, the maximum output power that the DC side can maintain voltage stability under the current healthy state can be accurately derived. In other words, the maximum output power that the grid-connected energy storage system can provide to the AC side under the current healthy state can be calculated using the following formula:

[0119] (10)

[0120] in, The maximum power provided by the battery under the current health condition is the upper limit of the maximum output power that the grid-connected energy storage system can provide to the AC side under the current health condition. This is the DC bus voltage; This is the minimum permissible operating voltage threshold for the DC bus. To account for the battery's equivalent total internal resistance after SOH, its value is equal to the sum of the battery's real-time ohmic internal resistance and the polarization internal resistance under the current SOH. The upper limit decreases dynamically with decreasing SOH, serving as an important physical constraint for subsequent stability analysis and control parameter adjustment. When evaluating DC-side support capability, considering that energy loss due to battery aging is composed of both ohmic and polarization losses, the mapped value is... and Perform linear superposition to synthesize the equivalent total internal resistance The equivalent total internal resistance This accurately reflects the voltage drop characteristics of an aged battery under transient high-power output. Furthermore, The value can be determined according to The voltage drop residual obtained from the mapping is corrected to improve the calculation accuracy of the power support boundary under aging conditions.

[0121] S4. Construct the power angle-frequency nonlinear state equation of the grid-connected energy storage system. Using the energy function method, determine the critical energy and critical energy stability boundary for the grid-connected energy storage system to maintain transient synchronization and stability based on the upper limit of the maximum output power. Use the critical energy stability boundary as the safe adjustment range of the control parameters of the grid-connected energy storage system.

[0122] As the battery's state of equilibrium (SOH) decreases, its internal resistance increases and its effective capacity decreases, directly causing the DC voltage to drop to the critical value more easily under transient disturbances (i.e., ...). This shrinks the range of energy release on the DC side, thus causing changes in the transient stability energy boundary of the grid-connected energy storage system. Under different SOH conditions, the allowable stable operating region of the system varies significantly. For example... Figure 5 As shown, this invention establishes the power angle-frequency nonlinear state equation of the grid-connected energy storage system based on the VSG control structure, and constructs an energy function on this basis to describe the degree of energy shift of the system during transient disturbances.

[0123] The specific implementation process of step S4 is as follows:

[0124] S4.1 Construct the power angle-frequency nonlinear state equation of the grid-connected energy storage system. The construction process is as follows:

[0125] To prevent excessive active support on the AC side from causing the DC side to collapse due to over-power extraction, a DC voltage compensation circuit is introduced to correct the grid angular frequency in real time.

[0126] (11)

[0127] In the formula, Angular frequency; The reference angular frequency; This is the droop coefficient; To output active power in AC mode; The active power setpoint; DC voltage compensation gain; This is the DC bus voltage; This is the reference value for the DC bus voltage.

[0128] Define the power angle-frequency nonlinear state equation of a grid-connected energy storage system:

[0129] (12)

[0130] In the formula, δ is the power angle; Angular frequency; The rated angular frequency of the power grid; It is the equivalent inertia coefficient; To simulate the mechanical power input to a generator; This is the equivalent damping coefficient; It outputs active power for AC.

[0131] S4.2. Using the energy function method, determine the critical energy and critical energy stability boundary for the grid-connected energy storage system to maintain transient synchronization and stability based on the upper limit of the maximum output power. Use the critical energy stability boundary as the safe adjustment range for the control parameters of the grid-connected energy storage system. Specifically:

[0132] Construct a scalar function that reflects the degree of energy shift in the system:

[0133] (13)

[0134] In the formula, This is the energy function value; This represents the angular frequency deviation.

[0135] Consider virtual impedance X v The equation for the active power output affected by this is:

[0136] (14)

[0137] In the formula, To output active power in AC mode; This represents the amplitude of the internal potential. This refers to the voltage at the grid connection point. For line reactance; The virtual reactance is taken into account for SOH feedback; δ is the power angle.

[0138] Based on the physical support limit, the critical stability energy boundary under the current SOH is determined as follows: The critical energy and critical energy stability boundary for the grid-connected energy storage system to maintain transient synchronization stability are calculated using the following formulas:

[0139] (15)

[0140] in, This is the energy function value; For follow The critical energy that is modified due to degradation; To simulate the mechanical power input to the generator; δ is the power angle; This refers to the angular frequency deviation. The initial stable power angle of the grid-connected energy storage system before the occurrence of disturbance; This represents the active power amplitude of the AC-side power angle characteristic curve. Combining this with the maximum power support capability under the aforementioned SOH constraint, the physical-side support limit is mapped to the system's critical stability energy boundary. This boundary establishes the safe adjustment red line for control parameters to maintain synchronous stability under transient disturbances, thus clarifying the safe adjustment range of control parameters such as virtual inertia and virtual damping during transient processes.

[0141] S5 identifies the charging and discharging operation mode of the energy storage system in real time, thereby identifying potential transient instability risks and generating transient risk assessment results; and by adjusting the final adaptive virtual reactance, it achieves transient limiting of the output current, thereby protecting the hardware safety of aging batteries.

[0142] The transient risk assessment results determine the focus of stability control. To address the differences in instability risk under different operating conditions, the system performs real-time operating mode identification: identifying the real-time operating condition of the energy storage system and determining the focus of stability control. In charging mode, the SOH (State of Health) constraint on control quantities focuses on preventing voltage spikes caused by overcharging, while in discharging mode, it focuses on preventing DC power failure. This asymmetric constraint strategy is designed based on the asymmetry in the terminal voltage and internal resistance of aged batteries under charging and discharging states.

[0143] The specific implementation process of step S5 is as follows:

[0144] Define operating mode factors :

[0145] (16)

[0146] In the formula: A value of 1 represents the discharge mode, and a value of -1 represents the charging mode. To output active power for AC; to assess transient instability risk through quantitative acceleration area:

[0147] (17)

[0148] In the formula, To accelerate the area; The initial phase angle; This is the fault clearing angle; To simulate the mechanical power input of a generator; simultaneously, to reduce the transient current limit based on the degree of aging to protect the battery hardware:

[0149] (18)

[0150] In the formula, This is the transient current limiting value; γ is the rated current; γ is the sensitivity coefficient; This indicates the real-time health status of the energy storage battery.

[0151] (19)

[0152] In the formula, For the final adaptive virtual reactance; Flag is determined by Triggered in conjunction with current over-limit conditions; This is the upper limit of virtual reactance adaptive, which is the maximum equivalent reactance value that the energy storage system is allowed to adjust to under the current SOH. This is the adaptive lower limit of virtual reactance, which is the minimum virtual reactance value of the system under normal operating conditions.

[0153] like Figure 4The grid-type energy storage controller shown uses the final adaptive virtual reactance determined by the system in conjunction with the current SOH state. The output characteristics of the converter terminal voltage are dynamically reshaped by adjusting... exist[ , The value within the range changes the equivalent impedance voltage drop between the potential within the energy storage system and the grid connection point voltage, thereby achieving transient limiting protection of the output current at the physical level.

[0154] S6. Based on the transient risk assessment results, an exponential correction operator reflecting the physical health of the battery is introduced. Through a dual self-adjusting adaptive actuator, the virtual inertia and virtual damping are saturated and limited respectively within the safe adjustment range, and the final control parameters are output. Based on the final control parameters, the control command of the grid-type converter is generated within the safe adjustment range. The control command performs adaptive adjustment of the virtual inertia coefficient and virtual damping coefficient according to the frequency fluctuation characteristics to achieve adaptive support for the power grid.

[0155] The control parameter generation logic used in this invention is as follows: Figure 5 As shown. Figure 5 and Figure 6 Both correspond to S6 in describing the execution of the dual-adjustment adaptive controller. Figure 5 The controller's internals and algorithms will be showcased in detail. Figure 6 Show the overall architecture of the system.

[0156] The specific implementation process of step S6 is as follows:

[0157] S6.1. Based on the transient risk assessment results, an exponential correction operator reflecting the physical health of the battery is introduced. A dual self-adjusting adaptive actuator is used to saturate and limit the virtual inertia and virtual damping within the safe adjustment range, respectively, and outputs the final control parameters. Details are as follows:

[0158] Under the dual premise of satisfying physical boundary and stability constraints, the system executes an adaptive control law. For example... Figure 6 As shown, this invention designs a dual self-adjusting adaptive actuator constrained by a stable energy boundary, used to coordinate grid support capability and battery physical safety under different operating conditions.

[0159] First, the basic adaptive inertia needs to be calculated based on the power grid frequency disturbance. :

[0160] (20)

[0161] In the formula, It is the static inertia; The first adjustable gain; This is the second adjustable gain; This is for frequency deviation; The frequency change rate is used; an exponential correction operator reflecting the physical health of the battery is introduced. :

[0162] (twenty one)

[0163] The virtual inertia is then saturated and limited by the critical energy stability boundary, and the final adaptive inertia is output. for:

[0164] (twenty two)

[0165] In the formula, It is a saturation limiting function; This is the minimum limit for virtual inertia; This represents the upper limit of virtual inertia. (Virtual inertia upper limit) This is determined based on the physical power limit under the current battery health condition. Specifically, the system uses a feedforward neural network to sense the polarization voltage drop compensation caused by battery aging, and then evaluates the maximum transient power allowed to be output by the energy storage converter under the current SOH constraint; subsequently, combined with the transient stability critical energy value determined by the energy function method, the maximum inertia coefficient that satisfies the requirement of system instability and does not trigger battery overcurrent protection is calculated, and it is defined as... .

[0166] Synchronous online adjustment of damping coefficient:

[0167] (twenty three)

[0168] In the formula, This is the final adaptive damping coefficient; The basic damping coefficient; The gain is adjusted for damping. For frequency deviation, The initial resistance, The real-time ohmic internal resistance.

[0169] S6.2. Based on the final control parameters, control commands for the grid-connected converter are generated within the safe adjustment range. These control commands adaptively adjust the virtual inertia coefficient and virtual damping coefficient according to the frequency fluctuation characteristics to achieve adaptive support for the power grid. Specifically:

[0170] The control commands of a grid-type converter ultimately converge into the phase angle closed-loop output equation:

[0171] (twenty four)

[0172] In the formula, Phase angle; The reference angular frequency; Given power; For the Laplace operator; To output active power in AC mode; This is the final adaptive damping coefficient; For the final adaptive inertia; The angular frequency is [value missing]. This command drives the converter actuator, achieving optimal active support to the power grid while ensuring that the aging battery is not overloaded.

[0173] Before generating control commands, it is necessary to identify the operating mode of the grid-type energy storage system and adjust the direction of control parameter adjustment according to the operating mode.

[0174] The dual self-adjusting adaptive actuator includes two independent and cooperative adjustment paths: a first self-adjusting path, used to adaptively adjust the grid control parameters according to the grid frequency disturbance information; and a second self-adjusting path, used to dynamically constrain the adjustment range of the grid control parameters according to the real-time health status (SOH) and critical stable energy boundary of the energy storage battery. The dual self-adjusting adaptive actuator establishes a hierarchical control principle prioritizing safety over response through the cooperative logic of the first and second self-adjusting paths. The grid control parameters include at least one of virtual inertia parameters, virtual damping parameters, and virtual impedance parameters. Specifically: the first self-adjusting path adaptively adjusts the virtual inertia and damping coefficient based on the frequency disturbance characteristics to improve the transient response capability of the grid system. In the first self-adjusting path, the system collects the real-time grid frequency f and calculates the basic adaptive inertia used for transient response according to equation (20). The second adjustment path introduces a health correction operator based on SOH (State of Health) and, combined with the critical energy boundary, dynamically limits the adaptive control input. For example... Figure 6 As shown, The input is connected to the saturation limiting module, and the exponential correction operator is driven by SOH. With virtual inertia limit This also constitutes the dynamic boundary constraints of the module. Based on the real-time health state (SOH) of the energy storage battery and the critical stable energy boundary determined by equation (15), the system dynamically generates the upper limit of the virtual inertia allowed under the current physical state. .

[0175] The dual self-adjusting adaptive actuator establishes a hierarchical control principle that prioritizes safety over response through the collaborative logic of the first and second adjustment paths. Specifically, the second adjustment path, as a higher-level safety constraint layer, has absolute binding force on the primary control commands generated by the first adjustment path.

[0176] When a power grid frequency disturbance triggers the first adjustment path, a large basic adaptive inertia is calculated. At that time, the second adjustment path is based on the DC-side power support limit of the real-time SOH mapping (i.e., the maximum power provided by the battery in the current health state). The physical safety limit (i.e., virtual inertia upper limit) under the current battery health state is simultaneously generated, along with the critical energy stability boundary. The actuator executes the final decision logic through the saturation limiting function `sat`: if... Not crossed The system will output according to the response requirements; if If an attempt is made to breach the physical safety red line, the second adjustment path will execute a "one-vote veto" hard constraint, forcibly clamping the output inertia to... Within the range.

[0177] This hierarchical design ensures that the grid control parameters always operate within the intersection of the battery's physical tolerance and the system's synchronous stability throughout the entire life cycle, fundamentally eliminating the risk of DC bus voltage collapse caused by blindly pursuing AC side support performance.

[0178] After completing the aforementioned adaptive adjustment, the system integrates the final control quantity into the phase angle closed-loop control equation, outputs the phase angle command of the grid-connected converter, and sends the generated reference voltage vector to the pulse width modulation module. By comparing the carrier signal, trigger pulse signals are generated for the converter's power devices, driving the grid-connected energy storage converter to operate. The converter performs switching actions according to the trigger pulses, extracting energy from the energy storage battery and converting it into AC power for injection into the grid. The system also monitors the current and voltage at the converter's output port in real time, feeding them back to the power calculation module and the SOH evaluation module, forming a dynamic closed-loop control that takes physical constraints into account. Simultaneously, current limiting protection is performed based on the final adaptive virtual reactance adjustment command described in step S5. By reshaping the equivalent output impedance characteristics of the converter, it ensures that during grid disturbances, the output power of the grid-connected energy storage system remains within the safe boundary allowed by the current SOH, driving the energy storage system to achieve safe and effective support for the grid under both physical and stability constraints.

[0179] This invention addresses the issues of limited DC-side energy support and unstable grid control caused by battery aging. It establishes a mapping relationship between the real-time state of health (SOH) of the energy storage battery and its ohmic resistance and polarization characteristics, and introduces a neural network to compensate for voltage prediction errors under aging conditions, thereby achieving accurate perception of the battery's physical constraint boundaries. Simultaneously, it utilizes the dynamic balance principle of DC bus voltage to evaluate the maximum power support capability that the energy storage system can provide to the AC side under the current SOH state. Furthermore, it constructs a nonlinear state-space model of grid-based energy storage considering DC dynamics, and uses the energy function method to determine the critical energy boundary for transient synchronization and stability of the system and the safe adjustment range of control parameters. For operating modes and transient disturbance characteristics, a dual self-adjusting adaptive actuator constrained by the energy boundary is designed. Under the dual premises of physical and stability constraints, it adaptively adjusts the virtual inertia and damping coefficient, and introduces a health correction mechanism based on SOH to dynamically limit the control parameters. This method achieves adaptive support capability optimization of grid-based energy storage throughout its entire life cycle while ensuring DC-side operational safety and system transient synchronization and stability.

[0180] This invention optimizes the transient stability of energy storage systems throughout their entire lifecycle by establishing a mapping relationship between the microscopic degradation mechanism of batteries and the DC-side support capability of new energy power grids.

Claims

1. An adaptive control method for energy storage batteries considering dynamic constraints on State of Health (SOH), characterized in that, include: Obtain real-time health status (SOH) parameters of the energy storage battery and generate equivalent physical parameters related to the battery aging status. The real-time ohmic internal resistance and polarization resistance under the current state of harmonics (SOH) of the battery are calculated based on equivalent physical parameters. The calculated values ​​are input into a pre-constructed RC equivalent model of the battery using a mechanism and data fusion approach. The model outputs the battery output voltage. A neural network error compensation module is introduced to correct the nonlinear error in the battery output voltage under aging conditions, thus obtaining the battery terminal voltage. Based on the battery's terminal voltage, real-time ohmic internal resistance, and polarization resistance under the current state of harmonics (SOH), combined with the dynamic balance equation of the DC bus voltage, the maximum output power that the grid-connected energy storage system can provide to the AC side under the current healthy state is calculated. A nonlinear state equation for the power angle-frequency of a grid-connected energy storage system is constructed. Using the energy function method, the critical energy and critical energy stability boundary for maintaining transient synchronization stability of the grid-connected energy storage system are determined based on the upper limit of the maximum output power. The critical energy stability boundary is then used as the safe adjustment range of the control parameters of the grid-connected energy storage system. The system identifies the charging and discharging operation modes of the energy storage system in real time, thereby identifying potential transient instability risks and generating transient risk assessment results. By adjusting the final adaptive virtual reactance, transient current limiting is achieved, thus protecting the hardware safety of aging batteries. Based on the transient risk assessment results, an exponential correction operator reflecting the battery's physical health is introduced. A dual self-adjusting adaptive actuator performs saturation and limiting processing on the virtual inertia and virtual damping respectively within the safe adjustment range, outputting the final control parameters. Based on the final control parameters, control commands for the grid-type converter are generated within the safe adjustment range. These control commands perform adaptive adjustment of the virtual inertia coefficient and virtual damping coefficient according to the frequency fluctuation characteristics, so as to achieve adaptive support for the power grid.

2. The adaptive control method for energy storage batteries considering dynamic constraints of State of Harmony (SOH) according to claim 1, characterized in that, The calculation of the battery's real-time ohmic internal resistance and polarization resistance under the current state of harmonics (SOH) based on equivalent physical parameters includes: Considering the decrease in conductivity due to battery degradation, a linear mapping relationship between the degree of aging and the physical ohmic internal resistance is established: ； In the formula, For real-time ohmic internal resistance; The initial ohmic internal resistance; Internal resistance at the end of the lifespan; This refers to the real-time health status of the energy storage battery. To simulate the nonlinear characteristic of a sharp increase in polarization resistance during the later stages of aging, the following formula is used: ； In the formula, For the present The polarization internal resistance is below; The initial polarization resistance, This is the aging sensitivity coefficient.

3. The adaptive control method for energy storage batteries considering dynamic constraints of State of Harm (SOH) according to claim 1, characterized in that, The neural network error compensation module corrects the nonlinear error in the battery output voltage under aging conditions to obtain the battery terminal voltage, including: A feedforward neural network is introduced to predict polarization residual error, and a data-model fusion-driven terminal voltage prediction equation is constructed. The expression for polarization residual error is a complex function of SOH, charging and discharging current, temperature, and resting time. ； In the formula, Calculate the difference between the actual measured voltage and the voltage from the physical model; This is a nonlinear mapping function constructed by a feedforward neural network; This refers to the charging and discharging current. For temperature; This refers to the settling time; The expression for the data-model fusion-driven terminal voltage prediction equation is as follows: ； In the formula, Terminal voltage; This is the open-circuit voltage; This refers to the charging and discharging current. This is the polarization voltage.

4. The adaptive control method for energy storage batteries considering dynamic constraints of State of Harmony (SOH) according to claim 1, characterized in that, The expression for the dynamic balance equation of the DC bus voltage is as follows: ； in, For DC bus capacitors; This is the DC bus voltage; This represents the battery's output power under its current healthy condition. It provides DC output active power; It outputs active power for AC.

5. The adaptive control method for energy storage batteries considering dynamic constraints of State of Harm (SOH) according to claim 1, characterized in that, The maximum output power that the grid-connected energy storage system can provide to the AC side under its current healthy state is calculated using the following formula: ； in, The maximum power provided by the battery under the current health condition is the upper limit of the maximum output power that the grid-connected energy storage system can provide to the AC side under the current health condition. This is the DC bus voltage; This is the minimum permissible operating voltage threshold for the DC bus. To account for the battery's equivalent total internal resistance after SOH, its value is equal to the sum of the battery's real-time ohmic internal resistance and the polarization internal resistance under the current SOH.

6. The adaptive control method for energy storage batteries considering dynamic constraints of State of Health (SOH) according to claim 1, characterized in that, The power angle-frequency nonlinear state equation of the grid-connected energy storage system is constructed as follows: A DC voltage compensation circuit is introduced to correct the grid-type angular frequency in real time. ； In the formula, Angular frequency; The reference angular frequency; This is the droop coefficient; To output active power in AC mode; The active power setpoint; DC voltage compensation gain; This is the DC bus voltage; This is the reference value for the DC bus voltage; Define the power angle-frequency nonlinear state equation of a grid-connected energy storage system: ； In the formula, δ is the power angle; Angular frequency; The rated angular frequency of the power grid; It is the equivalent inertia coefficient; To simulate the mechanical power input to a generator; This is the equivalent damping coefficient; It outputs active power for AC.

7. The adaptive control method for energy storage batteries considering dynamic SOH constraints according to claim 6, characterized in that, The critical energy and critical energy stability boundary for the grid-connected energy storage system to maintain transient synchronization and stability are calculated using the following formulas: ； in, This is the energy function value; For follow The critical energy that is modified due to degradation; To simulate the mechanical power input to the generator; δ is the power angle; This refers to the angular frequency deviation. The initial stable power angle of the grid-connected energy storage system before the occurrence of disturbance; This represents the active power amplitude of the AC side power angle characteristic curve.

8. The adaptive control method for energy storage batteries considering dynamic constraints of State of Health (SOH) according to claim 1, characterized in that, The real-time identification of the charging and discharging operation mode of the energy storage system, thereby identifying potential transient instability risks and generating transient risk assessment results, including: Define operating mode factors : ； In the formula: Setting Mode to 1 represents discharge mode; setting Mode to -1 represents charging mode. To output active power for AC; to assess transient instability risk through quantitative acceleration area: ； In the formula, To accelerate the area; The initial phase angle; This is the fault clearing angle; To simulate the mechanical power input of a generator; simultaneously, to reduce the transient current limit based on the degree of aging to protect the battery hardware: ； In the formula, This is the transient current limiting value; γ is the rated current; γ is the sensitivity coefficient; This refers to the real-time health status of the energy storage battery. ； In the formula, For the final adaptive virtual reactance; Flag is determined by Triggered in conjunction with current over-limit conditions; The virtual reactance is adaptively capped; This is the adaptive lower limit for virtual reactance.

9. The adaptive control method for energy storage batteries considering dynamic SOH constraints according to claim 1, characterized in that, Based on the transient risk assessment results, an exponential correction operator reflecting the physical health of the battery is introduced. A dual self-adjusting adaptive actuator performs saturation and limiting processing on the virtual inertia and virtual damping within the safe adjustment range, respectively, and outputs the final control parameters, including: Calculate the basic adaptive inertia based on power grid frequency disturbances. : ； In the formula, It is the static inertia; The first adjustable gain; This is the second adjustable gain; This is for frequency deviation; The frequency change rate is used; an exponential correction operator reflecting the physical health of the battery is introduced. : ； The virtual inertia is saturated and limited by combining the critical energy stability boundary, and the final adaptive inertia is output. for: ； In the formula, It is a saturation limiting function; This is the minimum limit for virtual inertia; The virtual inertia upper limit is set; online adjustment of the damping coefficient is performed synchronously. ； In the formula, This is the final adaptive damping coefficient; The basic damping coefficient; The gain is adjusted for damping. This is for frequency deviation; The initial resistance; The real-time ohmic internal resistance.

10. The adaptive control method for energy storage batteries considering dynamic constraints of State of Health (SOH) according to claim 1, characterized in that, The control commands for the grid-type converter are summarized in the phase angle closed-loop output equation: ； In the formula, Phase angle; The reference angular frequency; Given power; For the Laplace operator; To output active power in AC mode; This is the final adaptive damping coefficient; For the final adaptive inertia; ω is the angular frequency.

Images