Energy storage system SOC dynamic correction method and system based on three-dimensional fusion

Abstract

The invention discloses an energy storage system SOC dynamic correction method and system based on three-dimensional fusion. The method comprises the steps that a battery dynamic model, a temperature-internal resistance-SOC correlation model and a power grid frequency interaction model of an energy storage system are established respectively; establishing a first objective function, a second objective function and a third objective function corresponding to the battery dynamic model, the temperature-internal resistance-SOC correlation model and the power grid frequency interaction model respectively; unifying the first target function, the second target function and the third target function into a total optimization target based on a weighted fusion function; and performing minimization calculation processing on the total optimization target according to the real-time working condition characteristics of the energy storage system and the weight parameter of the total optimization target based on an adaptive switching strategy, and performing dynamic correction processing on the SOC of the energy storage system according to the minimization total optimization target. In the embodiment of the invention, full-link accurate mapping is realized, and the SOC estimation error is reduced to be within + / -1.5% from the original + / -5%.

Description

Technical Field

[0001] This invention relates to the field of energy storage system technology, and in particular to a method and system for dynamic correction of the State of Charge (SOC) of an energy storage system based on three-dimensional fusion. Background Technology

[0002] With the increasing penetration rate of new energy power generation and the growing complexity of distributed power grid structures, energy storage systems are playing an increasingly prominent role in grid frequency regulation, voltage support, and power quality management. The State of Charge (SOC) dynamic correction mechanism, as a core control technology for the coordinated operation of energy storage systems and the power grid, directly affects the lifespan, response accuracy, and grid stability of energy storage devices.

[0003] While current mainstream SOC estimation methods possess basic estimation capabilities, they are susceptible to measurement noise, temperature drift, and model lag in dynamic scenarios such as grid frequency fluctuations and sudden changes in renewable energy output. This leads to the accumulation of SOC estimation errors, potentially causing overcharging / over-discharging risks in energy storage. Existing dynamic SOC correction technologies are mostly based on static threshold rules or single-time-scale model predictive control. Although proposed SOC reconfiguration strategies based on AGC commands can reduce ineffective charging and discharging of energy storage through dead-zone boundary control, their model parameters, which rely on historical data fitting, are difficult to adapt to rapid changes in operating conditions under scenarios with high renewable energy penetration.

[0004] Furthermore, traditional methods often focus on single-node SOC balancing, lacking the ability to globally optimize the SOC distribution across multiple nodes in a distribution area. This leads to fragmented energy storage resource scheduling and an inability to achieve coordinated stability of grid frequency and voltage. However, existing SOC balancing control algorithms still have the following shortcomings: they cannot adapt to long-term changes in characteristics caused by battery aging and temperature variations, requiring frequent manual calibration and maintenance; in distribution systems with multiple energy storage nodes, traditional methods struggle to achieve coordinated SOC management, resulting in low energy storage resource utilization efficiency; and they rely solely on basic parameters such as voltage and current, failing to fully utilize the rich data resources provided by modern monitoring devices. Traditional methods cannot effectively integrate this high-precision data to improve SOC estimation accuracy. Therefore, there is an urgent need for a SOC dynamic correction algorithm based on 3D fusion to achieve efficient, safe and intelligent operation of the system. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art. This invention provides a method and system for dynamic correction of SOC of energy storage system based on three-dimensional fusion, which realizes accurate mapping of the entire link and reduces the SOC estimation error from the original ±5% to within ±1.5%.

[0006] To address the aforementioned technical problems, embodiments of the present invention provide a method for dynamic SOC correction of an energy storage system based on three-dimensional fusion, the method comprising: A battery dynamic model, a temperature-internal resistance-SOC correlation model, and a grid frequency interaction model for the energy storage system are established respectively. Establish the first objective function, the second objective function, and the third objective function corresponding to the battery dynamic model, the temperature-internal resistance-SOC correlation model, and the power grid frequency interaction model, respectively; The first objective function, the second objective function, and the third objective function are unified into a total optimization objective based on the weighted fusion function; Based on the adaptive switching strategy, the overall optimization objective is minimized according to the real-time operating characteristics of the energy storage system and the weight parameters of the overall optimization objective, and dynamic correction processing is performed on the SOC of the energy storage system according to the minimized overall optimization objective.

[0007] Optionally, the equations of the battery dynamic model are as follows: ; ; in, This is an open-circuit voltage function, representing the relationship between open-circuit voltage and state of charge; The internal resistance is a function of SOC and temperature T; This is the polarization voltage, used to simulate transient response; This is process noise; Let be the current at time t; Let be the voltage at time t; The sampled SOC value at time t+1; Let SOC be the sampled value at time t; The sampling time interval; This refers to the battery's rated capacity.

[0008] Optionally, the formula for the temperature-internal resistance-SOC correlation model is as follows: ; ; ; in, This is the internal resistance value at the current temperature; The reference internal resistance value is 25℃. The temperature detected during operation; This is the operating current after temperature compensation; To detect current; The current temperature coefficient; This is the OCV-SOC curve at the current temperature; OCV is the temperature coefficient; The OCV-SOC curve is shown at a reference temperature of 25℃.

[0009] Optionally, the formula for the power grid frequency interaction model is as follows: ; in, This is the correction amount for SOC; This refers to the power grid frequency deviation. , The proportional and integral coefficients are dynamically adjusted using an adaptive algorithm. Let t be the discharge power of the energy storage system at time t; This refers to the rated capacity of the energy storage system. This is used as a time interval.

[0010] Optionally, the formula for establishing the first objective function corresponding to the battery dynamic model is as follows: ; in, The SOC value estimated by the battery dynamic model; This is the SOC reference value; This refers to the actual measured terminal voltage; The predicted terminal voltage value calculated from the battery dynamic model.

[0011] Optionally, the formula for the second objective function corresponding to the temperature-internal resistance-SOC correlation model is as follows: ; in, This refers to the weighting coefficient for the temperature difference term; This refers to the internal temperature of the battery cluster. This refers to the weighting coefficient of the over-temperature penalty term; This is the safe temperature threshold.

[0012] Optionally, the formula for establishing the third objective function corresponding to the power grid frequency interaction model is as follows: ; in, The required SOC correction amount; This is the actual SOC correction amount; For the response time term, a weighting coefficient is used. This refers to the response delay time.

[0013] Optionally, the formula for unifying the first objective function, the second objective function, and the third objective function into a total optimization objective based on the weighted fusion function is as follows: ; in, , , These are weighting coefficients, dynamically adjusted using fuzzy logic.

[0014] Optionally, the formula for the adaptive switching strategy is as follows: ; Where ECM represents the high-order equivalent circuit model; Simplified represents the simplified electrochemical model; SOH-Compensated represents the health state compensation model; and C represents the current value corresponding to the rated capacity. Absolute value of the rate of temperature change; Temperature steady state, i.e. Hold for 5 minutes; These are the weighting coefficients.

[0015] In addition, embodiments of the present invention also provide a dynamic SOC correction system for energy storage systems based on three-dimensional fusion, the system comprising: The first module is used to establish the battery dynamic model, the temperature-internal resistance-SOC correlation model, and the grid frequency interaction model of the energy storage system, respectively. The second module is used to establish the first objective function, the second objective function, and the third objective function corresponding to the battery dynamic model, the temperature-internal resistance-SOC correlation model, and the power grid frequency interaction model, respectively. Weighted fusion module: used to unify the first objective function, the second objective function, and the third objective function into a total optimization objective based on a weighted fusion function; Dynamic correction module: It is used to perform dynamic correction processing on the SOC of the energy storage system based on the adaptive switching strategy, according to the real-time operating characteristics of the energy storage system and the weight parameters of the overall optimization objective, and to perform dynamic correction processing on the SOC of the energy storage system based on the minimized overall optimization objective.

[0016] In the specific implementation of this invention, by establishing a multi-dimensional state model, the temperature evolution of battery cells in the energy storage system is incorporated into a unified prediction model, and based on this, adaptive and rolling dynamic optimization control is achieved, effectively overcoming the shortcomings of traditional control algorithms such as response lag, high energy consumption, and uneven temperature distribution; thereby achieving accurate mapping across the entire link, and reducing the SOC estimation error from ±5% of the traditional method to within ±1.5%. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the SOC dynamic correction method for energy storage systems based on three-dimensional fusion in an embodiment of the present invention. Figure 2 This is a schematic diagram of the structural composition of the SOC dynamic correction system for an energy storage system based on three-dimensional fusion in an embodiment of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1, please refer to Figure 1 , Figure 1 This is a flowchart illustrating the SOC dynamic correction method for energy storage systems based on three-dimensional fusion in an embodiment of the present invention.

[0021] like Figure 1 As shown, a dynamic SOC correction method for an energy storage system based on three-dimensional fusion is proposed, the method comprising: S101: Establish the battery dynamic model, temperature-internal resistance-SOC correlation model, and grid frequency interaction model of the energy storage system respectively. In a specific implementation of this invention, the equations of the battery dynamic model are as follows: ; ; in, This is an open-circuit voltage function, representing the relationship between open-circuit voltage and state of charge; The internal resistance is a function of SOC and temperature T; This is the polarization voltage, used to simulate transient response; This is process noise; Let be the current at time t; Let be the voltage at time t; The sampled SOC value at time t+1; Let SOC be the sampled value at time t; The sampling time interval; This refers to the battery's rated capacity.

[0022] Furthermore, the formula for the temperature-internal resistance-SOC correlation model is as follows: ; ; ; in, This is the internal resistance value at the current temperature; The reference internal resistance value is 25℃. The temperature detected during operation; This is the operating current after temperature compensation; To detect current; The current temperature coefficient; This is the OCV-SOC curve at the current temperature; OCV is the temperature coefficient; The OCV-SOC curve is shown at a reference temperature of 25℃.

[0023] Furthermore, the formula for the power grid frequency interaction model is as follows: ; in, This is the correction amount for SOC; This refers to the power grid frequency deviation. , The proportional and integral coefficients are dynamically adjusted using an adaptive algorithm. Let t be the discharge power of the energy storage system at time t; This refers to the rated capacity of the energy storage system. This is used as a time interval.

[0024] Specifically, estimating the SOC of a battery first requires accurately describing its internal chemical reaction processes. Therefore, based on the improved Thevenin equivalent circuit model, a battery dynamic equation is established to describe the relationship between the terminal voltage and the internal state. The equations of the battery dynamic model are as follows: ; ; in, This is an open-circuit voltage function, representing the relationship between open-circuit voltage and state of charge; The internal resistance is a function of SOC and temperature T; This is the polarization voltage, used to simulate transient response; This is process noise; Let be the current at time t; Let be the voltage at time t; The sampled SOC value at time t+1; Let SOC be the sampled value at time t; The sampling time interval; This refers to the battery's rated capacity.

[0025] Temperature is a key factor affecting the accuracy of SOC estimation. Traditional methods often treat temperature as an external correction term, ignoring the real-time impact of the internal temperature gradient on internal resistance. In this embodiment, a thermodynamic transfer dimension is introduced to establish a correlation model between temperature, internal resistance, and SOC. The formula for the temperature-internal resistance-SOC correlation model is as follows: ; ; ; in, This is the internal resistance value at the current temperature; The reference internal resistance value is 25℃. The temperature detected during operation; This is the operating current after temperature compensation; To detect current; The current temperature coefficient; This is the OCV-SOC curve at the current temperature; OCV is the temperature coefficient; The OCV-SOC curve is shown at a reference temperature of 25℃.

[0026] When an energy storage system participates in grid frequency regulation, its State of Charge (SOC) needs to be dynamically adjusted according to the grid frequency deviation. This embodiment constructs a grid interaction dimension to establish a dynamic correlation between frequency deviation and SOC correction. The formula for the grid frequency interaction model is as follows: ; in, This is the correction amount for SOC; This refers to the power grid frequency deviation. , The proportional and integral coefficients are dynamically adjusted using an adaptive algorithm. Let t be the discharge power of the energy storage system at time t; This refers to the rated capacity of the energy storage system. This is used as a time interval.

[0027] S102: Establish the first objective function, the second objective function, and the third objective function corresponding to the battery dynamic model, the temperature-internal resistance-SOC correlation model, and the power grid frequency interaction model, respectively; In a specific implementation of this invention, the formula for establishing the first objective function corresponding to the battery dynamic model is as follows: ; in, The SOC value estimated by the battery dynamic model; This is the SOC reference value; This refers to the actual measured terminal voltage; The predicted terminal voltage value calculated from the battery dynamic model.

[0028] Furthermore, the formula for the second objective function corresponding to the temperature-internal resistance-SOC correlation model is as follows: ; in, This refers to the weighting coefficient for the temperature difference term; This refers to the internal temperature of the battery cluster. This refers to the weighting coefficient of the over-temperature penalty term; This is the safe temperature threshold.

[0029] Furthermore, the formula for establishing the third objective function corresponding to the power grid frequency interaction model is as follows: ; in, The required SOC correction amount; This is the actual SOC correction amount; For the response time term, a weighting coefficient is used. This refers to the response delay time.

[0030] Specifically, in actual operation, the three temperatures in S101 are interdependent. The grid dispatch demand changes the charging and discharging current, the current change causes the battery to heat up, and the temperature change affects the electrochemical parameters (internal resistance, OCV), which in turn affects the accuracy of SOC estimation. Therefore, it is necessary to establish a unified optimization framework to coordinate the three dimensions.

[0031] To achieve the optimal synergy between electrochemical accuracy, thermodynamic safety, and grid response speed, this embodiment constructs a unified state-space model that integrates electrochemical parameters (internal resistance, polarization voltage), thermodynamic parameters (temperature gradient, heat flux density), and grid signals (frequency deviation, power demand) into a single optimization problem.

[0032] First, define the objective function in three dimensions: the electrochemical dimension objective function. The formula for establishing the first objective function corresponding to the battery dynamic model is as follows: ; in, The SOC value estimated by the battery dynamic model; This is the SOC reference value; This refers to the actual measured terminal voltage; The predicted terminal voltage value calculated from the battery dynamic model.

[0033] The formula for the second objective function corresponding to the temperature-internal resistance-SOC correlation model is as follows: ; in, This refers to the weighting coefficient for the temperature difference term; This refers to the internal temperature of the battery cluster. This refers to the weighting coefficient of the over-temperature penalty term; This is the safe temperature threshold.

[0034] The formula for establishing the third objective function corresponding to the power grid frequency interaction model is as follows: ; in, The required SOC correction amount; This is the actual SOC correction amount; For the response time term, a weighting coefficient is used. This refers to the response delay time.

[0035] S103 unifies the first objective function, the second objective function, and the third objective function into a total optimization objective based on a weighted fusion function; In a specific implementation of this invention, the formula for unifying the first objective function, the second objective function, and the third objective function into a total optimization objective based on the weighted fusion function is as follows: ; in, , , These are weighting coefficients, dynamically adjusted using fuzzy logic.

[0036] Specifically, the three objectives are unified into a single overall optimization objective through a weighted fusion function: ; in, , , These are weighting coefficients, dynamically adjusted using fuzzy logic.

[0037] In this formula, J is the overall optimization objective function. The core of the algorithm is to minimize this J value. Through this dynamic weighting mechanism, the three dimensions achieve dynamic balance and collaborative optimization under a unified architecture.

[0038] S104: Based on the adaptive switching strategy, the total optimization objective is minimized according to the real-time operating characteristics of the energy storage system and the weight parameters of the total optimization objective, and the SOC of the energy storage system is dynamically corrected according to the minimized total optimization objective.

[0039] In a specific implementation of this invention, the formula for the adaptive switching strategy is as follows: ; Where ECM represents the high-order equivalent circuit model; Simplified represents the simplified electrochemical model; SOH-Compensated represents the health state compensation model; and C represents the current value corresponding to the rated capacity. Absolute value of the rate of temperature change; Temperature steady state, i.e. Hold for 5 minutes; These are the weighting coefficients.

[0040] Specifically, battery state changes involve different time scales: transient response at the second level, SOC evolution at the hour level, and aging degradation at the month / year level. In this embodiment, a multi-time-scale joint estimation architecture is designed to cover accurate management across all time scales. This architecture forms a tight input-output closed loop with the previous dimensions. The joint estimation framework mainly includes second-level, hour-level, and month / year-level scales.

[0041] Second-level scale: Estimating internal states such as lithium-ion concentration.

[0042] On a timescale of seconds, the battery exhibits rapid polarization characteristics and ion diffusion behavior. Based on the electrochemical mechanism of the battery dynamic model and combined with the real-time temperature distribution provided by the temperature-internal resistance-SOC correlation model, a reduced-order partial differential equation (PDE) model is used to estimate the lithium-ion concentration. ; denoted as ion concentration; D as diffusion coefficient; F as Faraday constant; I as current; this concentration estimate provides more accurate internal state information on an hourly scale, directly serving the... Refinement of the mid-terminal voltage prediction term.

[0043] Hourly Scale: Estimating SOC The hourly scale is the main time window for SOC estimation. An extended Kalman filter (EKF) fusion method is used to improve the ampere-hour integral method, integrating multi-dimensional information such as lithium-ion concentration estimated at the second scale, temperature measurements obtained from the temperature-internal resistance-SOC correlation model, and grid interaction signals from the grid frequency interaction model into a unified framework. Based on the fusion of the improved ampere-hour integral and EKF, characteristic quantities... As an extension of the observation vector, it is used for EKF updates as follows: ; in, It represents the battery state of charge at the (k+1)th sampling time. Let SOC be the value at the k-th sampling time. The battery current at the kth sampling time; Sampling time interval; Battery rated capacity; Kalman gain coefficient; Observation vector; Terminal voltage; Current; temperature; AC impedance amplitude; AC impedance phase angle; τ relaxation voltage time constant; Current SOC estimate: Lithium ion concentration; The observation equations represent electrochemistry, thermodynamics, and new characteristic quantities.

[0044] Kalman gain The mathematical essence and ; Kalman gain The calculation formula is: ; This formula is derived by solving the following optimization problem: ; And this optimization objective is The first item: ; The difference between the terminal voltage predicted by the observation equation and the measured value corresponds to... The second point; EKF continuously corrects this error through the innovation process; therefore, it directly achieves Minimize.

[0045] Observation vector pairs and Support; the temperature T in the observation vector is The key state variable—accurate temperature estimation—provides the data foundation for the temperature-internal resistance-SOC correlation model, enabling thermodynamic objectives to be achieved. The realization of this becomes possible; the characteristic quantity of AC impedance is The key state quantity—impedance information—reflects the battery's response capability to grid frequency regulation, providing a basis for power control and grid interaction in the grid frequency interaction model.

[0046] Monthly / Grade Scale: Estimate SOH (Health Status) As batteries age, their rated capacity gradually decreases. This invention estimates the State of Charge (SOH) using a capacity decay model and feeds the updated rated capacity back into the hourly-scale SOC recursive formula, thus eliminating long-term drift errors at their source. The capacity decay model is as follows: ; This is the rated capacity at the current moment; The initial rated capacity is λ, the aging factor is η, and the cycle decay factor is η. The updated rated capacity is fed back to the hourly SOC recursive formula. ; This mechanism ensures Feasibility throughout the entire lifecycle—if the rated capacity is not updated, as the battery ages, It will gradually increase.

[0047] However, in actual operation, different operating conditions require different levels of model complexity: high-order models are needed to ensure accuracy during drastic fluctuations, while simplified models can be used to reduce computational burden during steady-state operation; therefore, this embodiment introduces an adaptive model switching strategy based on real-time operating conditions.

[0048] The formula for the adaptive handover strategy is as follows: ; Where ECM represents the high-order equivalent circuit model; Simplified represents the simplified electrochemical model; SOH-Compensated represents the health state compensation model; and C represents the current value corresponding to the rated capacity. Absolute value of the rate of temperature change; Temperature steady state, i.e. Hold for 5 minutes; These are the weighting coefficients.

[0049] Construct three models with different levels of complexity: (1) ECM (Higher-order equivalent circuit model) ECM (Higher Order Equivalent Circuit Model): Employs a second- or third-order RC network to accurately simulate the battery's polarization response at different frequencies; internal resistance R and capacitance C change in real time with SOC and temperature T, and are updated through online parameter identification; the hysteresis effect of open-circuit voltage OCV is considered to improve the estimation accuracy during charge-discharge switching. Suitable for complex operating conditions such as drastic current changes (|I|>2C), rapid temperature fluctuations (|dT / dt|>5°C / min), or large fluctuations in grid frequency (|Δf|>0.2Hz).

[0050] (2) Simplified electrochemical model The battery dynamic model is reasonably simplified by reducing the multi-order RC network to a first-order RC network and ignoring the high-frequency dynamic response. The internal resistance R is approximated as constant within the local SOC range to reduce the computational burden. The hysteresis effect of OCV is not considered. It is suitable for normal operating conditions with stable current (|I|<0.1C), stable temperature, and normal grid frequency.

[0051] (3) SOH-Compensated (Health Status Compensation Model) Based on the Simplified model, the SOH value estimated at the monthly / year scale above is introduced for compensation; this includes replacing the rated capacity in the SOC recursive formula with the real-time updated actual capacity, compensating for internal resistance based on the SOH value, and appropriately correcting the OCV-SOC curve according to the degree of aging; it is suitable for scenarios with stable operating conditions and batteries that have entered the middle and late stages of their lifespan (SOH<80%).

[0052] The selected model is passed as an instruction to the joint estimation framework, which directly determines the algorithm complexity and computation method of the joint estimation framework at the three time scales. Through this "on-demand computation" mechanism, the joint estimation framework can run at the most appropriate complexity under different model instructions, ensuring that the overall optimization objective J is always effectively controlled.

[0053] When selecting the ECM model; for second-level (lithium-ion concentration estimation): enable full PDE solution without any simplification to obtain the most accurate lithium-ion concentration. This provides high-precision internal state information for hourly-scale terminal voltage prediction, directly reducing J... elec The terminal voltage prediction error term in the text.

[0054] Hourly Scale (SOC Estimation): Full EKF enabled. A state-space model using a 2nd or 3rd order RC network is employed, with the state vector containing multiple polarization voltages; the observation vector contains all characteristic quantities [V,I,T|Z|θ]T; internal resistance R and capacitance C are identified online in real-time with respect to SOC and temperature T. Accurate Kalman gain calculation directly realizes J... electro Minimization, while supporting J through temperature T and AC impedance characteristic quantities. thermal and J grid Optimization.

[0055] Monthly / Grade Scale (SOH Estimation): Regularly update SOH; periodically calculate actual capacity C according to the capacity decay model. actual This information is fed back to the SOC recursive formula on an hourly scale to ensure long-term accuracy.

[0056] When selecting the Simplified model, for second-scale (lithium-ion concentration estimation): a reduced-order PDE is used to solve the diffusion equation, simplifying it (e.g., ignoring higher-order diffusion terms or using a steady-state approximation) to quickly obtain c. Li To approximate the value while maintaining basic accuracy, the computational burden is reduced; for hourly-scale (SOC estimation): a reduced-order EKF is used. A first-order RC network is used, with the state vector containing only one polarization voltage; the observation vector is simplified to [V,I,T]T, ignoring AC impedance characteristics; the internal resistance R is approximated as constant within the local SOC interval to reduce the computational burden. The goal remains to minimize the SOC estimation error. elec Maintaining at an acceptable level, temperature T observations provide limited thermal compensation; monthly / grade scale (SOH estimation): reduce the SOH update frequency, or only update when the cumulative number of cycles reaches a certain threshold, to save computational resources.

[0057] When selecting the SOH-Compensated model, at the second-level scale (lithium ion concentration estimation): if the operating conditions are stable, lithium ion concentration estimation can be paused, and empirical values ​​can be used directly or this step can be ignored, relying primarily on SOH compensation to maintain SOC accuracy; at the hour-level scale (SOC estimation): it mainly relies on the ampere-hour integral of SOH compensation; the core formula is... C actual Real-time updates from monthly / grade-level scales; EKF reduced-order operation, observation vectors can be further simplified (e.g., retaining only Vt and I); emphasis is placed on preventing accumulated errors due to aging through capacity correction, maintaining J elec Long-term stability. Monthly / grade-level (SOH estimation): Focus on implementing SOH estimation, regularly update actual capacity, and promptly feed the results back to the hourly scale to form a closed-loop correction.

[0058] In the specific implementation of this invention, by establishing a multi-dimensional state model, the temperature evolution of battery cells in the energy storage system is incorporated into a unified prediction model, and based on this, adaptive and rolling dynamic optimization control is achieved, effectively overcoming the shortcomings of traditional control algorithms such as response lag, high energy consumption, and uneven temperature distribution; thereby achieving accurate mapping across the entire link, and reducing the SOC estimation error from ±5% of the traditional method to within ±1.5%.

[0059] Example 2, please refer to Figure 2 , Figure 2 This is a schematic diagram of the structural composition of the SOC dynamic correction system for an energy storage system based on three-dimensional fusion in an embodiment of the present invention.

[0060] like Figure 2 As shown, a dynamic SOC correction system for an energy storage system based on three-dimensional fusion is disclosed. The system includes: The first module 201 is used to establish the battery dynamic model, the temperature-internal resistance-SOC correlation model, and the grid frequency interaction model of the energy storage system, respectively. In a specific implementation of this invention, the equations of the battery dynamic model are as follows: ; ; in, This is an open-circuit voltage function, representing the relationship between open-circuit voltage and state of charge; The internal resistance is a function of SOC and temperature T; This is the polarization voltage, used to simulate transient response; This is process noise; Let be the current at time t; Let be the voltage at time t; The sampled SOC value at time t+1; Let SOC be the sampled value at time t; The sampling time interval; This refers to the battery's rated capacity.

[0061] Furthermore, the formula for the temperature-internal resistance-SOC correlation model is as follows: ; ; ; in, This is the internal resistance value at the current temperature; The reference internal resistance value is 25℃. The temperature detected during operation; This is the operating current after temperature compensation; To detect current; The current temperature coefficient; This is the OCV-SOC curve at the current temperature; OCV is the temperature coefficient; The OCV-SOC curve is shown at a reference temperature of 25℃.

[0062] Furthermore, the formula for the power grid frequency interaction model is as follows: ; in, This is the correction amount for SOC; This refers to the power grid frequency deviation. , The proportional and integral coefficients are dynamically adjusted using an adaptive algorithm. Let t be the discharge power of the energy storage system at time t; This refers to the rated capacity of the energy storage system. This is used as a time interval.

[0063] Specifically, estimating the SOC of a battery first requires accurately describing its internal chemical reaction processes. Therefore, based on the improved Thevenin equivalent circuit model, a battery dynamic equation is established to describe the relationship between the terminal voltage and the internal state. The equations of the battery dynamic model are as follows: ; ; in, This is an open-circuit voltage function, representing the relationship between open-circuit voltage and state of charge; The internal resistance is a function of SOC and temperature T; This is the polarization voltage, used to simulate transient response; This is process noise; Let be the current at time t; Let be the voltage at time t; The sampled SOC value at time t+1; Let SOC be the sampled value at time t; The sampling time interval; This refers to the battery's rated capacity.

[0064] Temperature is a key factor affecting the accuracy of SOC estimation. Traditional methods often treat temperature as an external correction term, ignoring the real-time impact of the internal temperature gradient on internal resistance. In this embodiment, a thermodynamic transfer dimension is introduced to establish a correlation model between temperature, internal resistance, and SOC. The formula for the temperature-internal resistance-SOC correlation model is as follows: ; ; ; in, This is the internal resistance value at the current temperature; The reference internal resistance value is 25℃. The temperature detected during operation; This is the operating current after temperature compensation; To detect current; The current temperature coefficient; This is the OCV-SOC curve at the current temperature; OCV is the temperature coefficient; The OCV-SOC curve is shown at a reference temperature of 25℃.

[0065] When an energy storage system participates in grid frequency regulation, its State of Charge (SOC) needs to be dynamically adjusted according to the grid frequency deviation. This embodiment constructs a grid interaction dimension to establish a dynamic correlation between frequency deviation and SOC correction. The formula for the grid frequency interaction model is as follows: ; in, This is the correction amount for SOC; This refers to the power grid frequency deviation. , The proportional and integral coefficients are dynamically adjusted using an adaptive algorithm. Let t be the discharge power of the energy storage system at time t; This refers to the rated capacity of the energy storage system. This is used as a time interval.

[0066] The second establishment module 202 is used to establish the first objective function, the second objective function, and the third objective function corresponding to the battery dynamic model, the temperature-internal resistance-SOC correlation model, and the power grid frequency interaction model, respectively. In a specific implementation of this invention, the formula for establishing the first objective function corresponding to the battery dynamic model is as follows: ; in, The SOC value estimated by the battery dynamic model; This is the SOC reference value; This refers to the actual measured terminal voltage; The predicted terminal voltage value calculated from the battery dynamic model.

[0067] Furthermore, the formula for the second objective function corresponding to the temperature-internal resistance-SOC correlation model is as follows: ; in, This refers to the weighting coefficient for the temperature difference term; This refers to the internal temperature of the battery cluster. This refers to the weighting coefficient of the over-temperature penalty term; This is the safe temperature threshold.

[0068] Furthermore, the formula for establishing the third objective function corresponding to the power grid frequency interaction model is as follows: ; in, The required SOC correction amount; This is the actual SOC correction amount; For the response time term, a weighting coefficient is used. This refers to the response delay time.

[0069] Specifically, in actual operation, the three temperatures in S101 are interdependent. The grid dispatch demand changes the charging and discharging current, the current change causes the battery to heat up, and the temperature change affects the electrochemical parameters (internal resistance, OCV), which in turn affects the accuracy of SOC estimation. Therefore, it is necessary to establish a unified optimization framework to coordinate the three dimensions.

[0070] To achieve the optimal synergy between electrochemical accuracy, thermodynamic safety, and grid response speed, this embodiment constructs a unified state-space model that integrates electrochemical parameters (internal resistance, polarization voltage), thermodynamic parameters (temperature gradient, heat flux density), and grid signals (frequency deviation, power demand) into a single optimization problem.

[0071] First, define the objective function in three dimensions: the electrochemical dimension objective function. The formula for establishing the first objective function corresponding to the battery dynamic model is as follows: ; in, The SOC value estimated by the battery dynamic model; This is the SOC reference value; This refers to the actual measured terminal voltage; The predicted terminal voltage value calculated from the battery dynamic model.

[0072] The formula for the second objective function corresponding to the temperature-internal resistance-SOC correlation model is as follows: ; in, This refers to the weighting coefficient for the temperature difference term; This refers to the internal temperature of the battery cluster. This refers to the weighting coefficient of the over-temperature penalty term; This is the safe temperature threshold.

[0073] The formula for establishing the third objective function corresponding to the power grid frequency interaction model is as follows: ; in, The required SOC correction amount; This is the actual SOC correction amount; For the response time term, a weighting coefficient is used. This refers to the response delay time.

[0074] Weighted fusion module 203: used to unify the first objective function, the second objective function and the third objective function into a total optimization objective based on a weighted fusion function; In a specific implementation of this invention, the formula for unifying the first objective function, the second objective function, and the third objective function into a total optimization objective based on the weighted fusion function is as follows: ; in, , , These are weighting coefficients, dynamically adjusted using fuzzy logic.

[0075] Specifically, the three objectives are unified into a single overall optimization objective through a weighted fusion function: ; in, , , These are weighting coefficients, dynamically adjusted using fuzzy logic.

[0076] In this formula, J is the overall optimization objective function. The core of the algorithm is to minimize this J value. Through this dynamic weighting mechanism, the three dimensions achieve dynamic balance and collaborative optimization under a unified architecture.

[0077] Dynamic correction module 204: is used to perform a minimum calculation of the total optimization target based on the adaptive switching strategy according to the real-time operating characteristics of the energy storage system and the weight parameters of the total optimization target, and to perform dynamic correction processing on the SOC of the energy storage system according to the minimized total optimization target.

[0078] In a specific implementation of this invention, the formula for the adaptive switching strategy is as follows: ; Where ECM represents the high-order equivalent circuit model; Simplified represents the simplified electrochemical model; SOH-Compensated represents the health state compensation model; and C represents the current value corresponding to the rated capacity. Absolute value of the rate of temperature change; Temperature steady state, i.e. Hold for 5 minutes; These are the weighting coefficients.

[0079] Specifically, battery state changes involve different time scales: transient response at the second level, SOC evolution at the hour level, and aging degradation at the month / year level. In this embodiment, a multi-time-scale joint estimation architecture is designed to cover accurate management across all time scales. This architecture forms a tight input-output closed loop with the previous dimensions. The joint estimation framework mainly includes second-level, hour-level, and month / year-level scales.

[0080] Second-level scale: Estimating internal states such as lithium-ion concentration.

[0081] On a timescale of seconds, the battery exhibits rapid polarization characteristics and ion diffusion behavior. Based on the electrochemical mechanism of the battery dynamic model and combined with the real-time temperature distribution provided by the temperature-internal resistance-SOC correlation model, a reduced-order partial differential equation (PDE) model is used to estimate the lithium-ion concentration. ; denoted as ion concentration; D as diffusion coefficient; F as Faraday constant; I as current; this concentration estimate provides more accurate internal state information on an hourly scale, directly serving the... Refinement of the mid-terminal voltage prediction term.

[0082] Hourly Scale: Estimating SOC The hourly scale is the main time window for SOC estimation. An extended Kalman filter (EKF) fusion method is used to improve the ampere-hour integral method, integrating multi-dimensional information such as lithium-ion concentration estimated at the second scale, temperature measurements obtained from the temperature-internal resistance-SOC correlation model, and grid interaction signals from the grid frequency interaction model into a unified framework. Based on the fusion of the improved ampere-hour integral and EKF, characteristic quantities... As an extension of the observation vector, it is used for EKF updates as follows: ; in, It represents the battery state of charge at the (k+1)th sampling time. Let SOC be the value at the k-th sampling time. The battery current at the kth sampling time; Sampling time interval; Battery rated capacity; Kalman gain coefficient; Observation vector; Terminal voltage; Current; temperature; AC impedance amplitude; AC impedance phase angle; τ relaxation voltage time constant; Current SOC estimate: Lithium ion concentration; The observation equations represent electrochemistry, thermodynamics, and new characteristic quantities.

[0083] Kalman gain The mathematical essence and ; Kalman gain The calculation formula is: ; This formula is derived by solving the following optimization problem: ; And this optimization objective is The first item: ; The difference between the terminal voltage predicted by the observation equation and the measured value corresponds to... The second point; EKF continuously corrects this error through the innovation process; therefore, it directly achieves Minimize.

[0084] Observation vector pairs and Support; the temperature T in the observation vector is The key state variable—accurate temperature estimation—provides the data foundation for the temperature-internal resistance-SOC correlation model, enabling thermodynamic objectives to be achieved. The realization of this becomes possible; the characteristic quantity of AC impedance is The key state quantity—impedance information—reflects the battery's response capability to grid frequency regulation, providing a basis for power control and grid interaction in the grid frequency interaction model.

[0085] Monthly / Grade Scale: Estimate SOH (Health Status) As batteries age, their rated capacity gradually decreases. This invention estimates the State of Charge (SOH) using a capacity decay model and feeds the updated rated capacity back into the hourly-scale SOC recursive formula, thus eliminating long-term drift errors at their source. The capacity decay model is as follows: ; This is the rated capacity at the current moment; The initial rated capacity is λ, the aging factor is η, and the cycle decay factor is η. The updated rated capacity is fed back to the hourly SOC recursive formula. ; This mechanism ensures Feasibility throughout the entire lifecycle—if the rated capacity is not updated, as the battery ages, It will gradually increase.

[0086] However, in actual operation, different operating conditions require different levels of model complexity: high-order models are needed to ensure accuracy during drastic fluctuations, while simplified models can be used to reduce computational burden during steady-state operation; therefore, this embodiment introduces an adaptive model switching strategy based on real-time operating conditions.

[0087] The formula for the adaptive handover strategy is as follows: ; Where ECM represents the high-order equivalent circuit model; Simplified represents the simplified electrochemical model; SOH-Compensated represents the health state compensation model; and C represents the current value corresponding to the rated capacity. Absolute value of the rate of temperature change; Temperature steady state, i.e. Hold for 5 minutes; These are the weighting coefficients.

[0088] Construct three models with different levels of complexity: (1) ECM (Higher-order equivalent circuit model) ECM (Higher Order Equivalent Circuit Model): Employs a second- or third-order RC network to accurately simulate the battery's polarization response at different frequencies; internal resistance R and capacitance C change in real time with SOC and temperature T, and are updated through online parameter identification; the hysteresis effect of open-circuit voltage OCV is considered to improve the estimation accuracy during charge-discharge switching. Suitable for complex operating conditions such as drastic current changes (|I|>2C), rapid temperature fluctuations (|dT / dt|>5°C / min), or large fluctuations in grid frequency (|Δf|>0.2Hz).

[0089] (2) Simplified electrochemical model The battery dynamic model is reasonably simplified by reducing the multi-order RC network to a first-order RC network and ignoring the high-frequency dynamic response. The internal resistance R is approximated as constant within the local SOC range to reduce the computational burden. The hysteresis effect of OCV is not considered. It is suitable for normal operating conditions with stable current (|I|<0.1C), stable temperature, and normal grid frequency.

[0090] (3) SOH-Compensated (Health Status Compensation Model) Based on the Simplified model, the SOH value estimated at the monthly / year scale above is introduced for compensation; this includes replacing the rated capacity in the SOC recursive formula with the real-time updated actual capacity, compensating for internal resistance based on the SOH value, and appropriately correcting the OCV-SOC curve according to the degree of aging; it is suitable for scenarios with stable operating conditions and batteries that have entered the middle and late stages of their lifespan (SOH<80%).

[0091] The selected model is passed as an instruction to the joint estimation framework, which directly determines the algorithm complexity and computation method of the joint estimation framework at the three time scales. Through this "on-demand computation" mechanism, the joint estimation framework can run at the most appropriate complexity under different model instructions, ensuring that the overall optimization objective J is always effectively controlled.

[0092] When selecting the ECM model; for second-level (lithium-ion concentration estimation): enable full PDE solution without any simplification to obtain the most accurate lithium-ion concentration. This provides high-precision internal state information for hourly-scale terminal voltage prediction, directly reducing J... elec The terminal voltage prediction error term in the text.

[0093] Hourly Scale (SOC Estimation): Full EKF enabled. A state-space model using a 2nd or 3rd order RC network is employed, with the state vector containing multiple polarization voltages; the observation vector contains all characteristic quantities [V,I,T|Z|θ]T; internal resistance R and capacitance C are identified online in real-time with respect to SOC and temperature T. Accurate Kalman gain calculation directly realizes J... electro Minimization, while supporting J through temperature T and AC impedance characteristic quantities. thermal and J grid Optimization.

[0094] Monthly / Grade Scale (SOH Estimation): Regularly update SOH; periodically calculate actual capacity C according to the capacity decay model. actual This information is fed back to the SOC recursive formula on an hourly scale to ensure long-term accuracy.

[0095] When selecting the Simplified model, for second-scale (lithium-ion concentration estimation): a reduced-order PDE is used to solve the diffusion equation, simplifying it (e.g., ignoring higher-order diffusion terms or using a steady-state approximation) to quickly obtain c. Li To approximate the value while maintaining basic accuracy, the computational burden is reduced; for hourly-scale (SOC estimation): a reduced-order EKF is used. A first-order RC network is used, with the state vector containing only one polarization voltage; the observation vector is simplified to [V,I,T]T, ignoring AC impedance characteristics; the internal resistance R is approximated as constant within the local SOC interval to reduce the computational burden. The goal remains to minimize the SOC estimation error. elec Maintaining at an acceptable level, temperature T observations provide limited thermal compensation; monthly / grade scale (SOH estimation): reduce the SOH update frequency, or only update when the cumulative number of cycles reaches a certain threshold, to save computational resources.

[0096] When selecting the SOH-Compensated model, at the second-level scale (lithium ion concentration estimation): if the operating conditions are stable, lithium ion concentration estimation can be paused, and empirical values ​​can be used directly or this step can be ignored, relying primarily on SOH compensation to maintain SOC accuracy; at the hour-level scale (SOC estimation): it mainly relies on the ampere-hour integral of SOH compensation; the core formula is... C actual Real-time updates from monthly / grade-level scales; EKF reduced-order operation, observation vectors can be further simplified (e.g., retaining only Vt and I); emphasis is placed on preventing accumulated errors due to aging through capacity correction, maintaining J elecLong-term stability. Monthly / grade-level (SOH estimation): Focus on implementing SOH estimation, regularly update actual capacity, and promptly feed the results back to the hourly scale to form a closed-loop correction.

[0097] In the specific implementation of this invention, by establishing a multi-dimensional state model, the temperature evolution of battery cells in the energy storage system is incorporated into a unified prediction model, and based on this, adaptive and rolling dynamic optimization control is achieved, effectively overcoming the shortcomings of traditional control algorithms such as response lag, high energy consumption, and uneven temperature distribution; thereby achieving accurate mapping across the entire link, and reducing the SOC estimation error from ±5% of the traditional method to within ±1.5%.

[0098] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, which may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0099] Furthermore, the above provides a detailed description of the SOC dynamic correction method and system for energy storage systems based on three-dimensional fusion provided by the embodiments of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for dynamic SOC correction of an energy storage system based on three-dimensional fusion, characterized in that, The method includes: A battery dynamic model, a temperature-internal resistance-SOC correlation model, and a grid frequency interaction model for the energy storage system are established respectively. Establish the first objective function, the second objective function, and the third objective function corresponding to the battery dynamic model, the temperature-internal resistance-SOC correlation model, and the power grid frequency interaction model, respectively; The first objective function, the second objective function, and the third objective function are unified into a total optimization objective based on the weighted fusion function; Based on the adaptive switching strategy, the overall optimization objective is minimized according to the real-time operating characteristics of the energy storage system and the weight parameters of the overall optimization objective, and dynamic correction processing is performed on the SOC of the energy storage system according to the minimized overall optimization objective.

2. The method for dynamic SOC correction of an energy storage system according to claim 1, characterized in that, The equations of the battery dynamic model are as follows: ; ; in, This is an open-circuit voltage function, representing the relationship between open-circuit voltage and state of charge; The internal resistance is a function of SOC and temperature T; This is the polarization voltage, used to simulate transient response; This is process noise; Let be the current at time t; Let be the voltage at time t; The sampled SOC value at time t+1; Let SOC be the sampled value at time t; The sampling time interval; This refers to the battery's rated capacity.

3. The method for dynamic SOC correction of an energy storage system according to claim 1, characterized in that, The formula for the temperature-internal resistance-SOC correlation model is as follows: ； ； ; in, This is the internal resistance value at the current temperature; The reference internal resistance value is 25℃. The temperature detected during operation; This is the operating current after temperature compensation; To detect current; The current temperature coefficient; This is the OCV-SOC curve at the current temperature; OCV is the temperature coefficient; The OCV-SOC curve is shown at a reference temperature of 25℃.

4. The method for dynamic SOC correction of an energy storage system according to claim 1, characterized in that, The formula for the power grid frequency interaction model is as follows: ； in, This is the correction amount for SOC; This refers to the power grid frequency deviation. , The proportional and integral coefficients are dynamically adjusted using an adaptive algorithm. Let t be the discharge power of the energy storage system at time t; This refers to the rated capacity of the energy storage system. This is used as a time interval.

5. The method for dynamic SOC correction of an energy storage system according to claim 1, characterized in that, The formula for establishing the first objective function corresponding to the battery dynamic model is as follows: ； in, The SOC value estimated by the battery dynamic model; This is the SOC reference value; This refers to the actual measured terminal voltage; The predicted terminal voltage value calculated from the battery dynamic model.

6. The method for dynamic SOC correction of an energy storage system according to claim 1, characterized in that, The formula for the second objective function corresponding to the temperature-internal resistance-SOC correlation model is as follows: ; in, This refers to the weighting coefficient for the temperature difference term; This refers to the internal temperature of the battery cluster. This refers to the weighting coefficient of the over-temperature penalty term; This is the safe temperature threshold.

7. The method for dynamic SOC correction of an energy storage system according to claim 1, characterized in that, The formula for establishing the third objective function corresponding to the power grid frequency interaction model is as follows: ； in, The required SOC correction amount; This is the actual SOC correction amount; For the response time term, a weighting coefficient is used. This refers to the response delay time.

8. The method for dynamic SOC correction of an energy storage system according to claim 1, characterized in that, The formula for unifying the first objective function, the second objective function, and the third objective function into a total optimization objective based on the weighted fusion function is as follows: ； in, , , These are weighting coefficients, dynamically adjusted using fuzzy logic.

9. The method for dynamic SOC correction of an energy storage system according to claim 1, characterized in that, The formula for the adaptive handover strategy is as follows: ； Where ECM represents the high-order equivalent circuit model; Simplified represents the simplified electrochemical model; SOH-Compensated represents the health state compensation model; and C represents the current value corresponding to the rated capacity. Absolute value of the rate of temperature change; Temperature steady state, i.e. Hold for 5 minutes; These are the weighting coefficients.

10. A dynamic SOC correction system for an energy storage system based on three-dimensional fusion, characterized in that, The system includes: The first module is used to establish the battery dynamic model, the temperature-internal resistance-SOC correlation model, and the grid frequency interaction model of the energy storage system, respectively. The second module is used to establish the first objective function, the second objective function, and the third objective function corresponding to the battery dynamic model, the temperature-internal resistance-SOC correlation model, and the power grid frequency interaction model, respectively. Weighted fusion module: used to unify the first objective function, the second objective function, and the third objective function into a total optimization objective based on a weighted fusion function; Dynamic correction module: It is used to perform dynamic correction processing on the SOC of the energy storage system based on the adaptive switching strategy, according to the real-time operating characteristics of the energy storage system and the weight parameters of the overall optimization objective, and to perform dynamic correction processing on the SOC of the energy storage system based on the minimized overall optimization objective.

Images