Method, device, program product, and storage medium for estimating state of charge of a battery

Abstract

This application provides a method, apparatus, program product, and storage medium for estimating the state of charge (SOC) of a battery. The method includes acquiring the mapping relationship between open-circuit voltage and SOC, voltage measurements, current measurements, and a parameter set of a battery equivalent circuit model. Based on the mapping relationship and parameter set, a battery equivalent circuit model incorporating hysteresis characteristics is determined. Based on this model, an extended Kalman filter algorithm is used for SOC prediction to obtain the predicted SOC and voltage values ​​at the current moment. Based on the current voltage measurements, predicted voltage values, and current measurements, it is determined whether a correction condition is met. If met, based on the characteristic interval corresponding to the predicted SOC at the current moment, a target gain update strategy corresponding to that characteristic interval is used to update the Kalman gain to obtain a target Kalman gain. The predicted SOC at the current moment is updated based on the target Kalman gain to obtain an estimated SOC value at the current moment.

Description

Technical Field

[0001] This application relates to battery management technology, and more particularly to a method, apparatus, program product, and storage medium for estimating the state of charge of a battery. Background Technology

[0002] Against the backdrop of global energy structure transformation and the integration of high proportions of renewable energy into the grid, energy storage systems using lithium iron phosphate (LFP) batteries have become a key means to meet the second-level frequency regulation requirements of the power grid. This frequency regulation task typically requires the battery to operate in a medium state of charge (SOC) range. However, this range corresponds to the plateau region in the LFP battery open-circuit voltage-SOC relationship curve, where the voltage is extremely insensitive to changes in SOC, leading to a significant decrease in the observability of state estimation. Simultaneously, the significant voltage hysteresis effect during battery charging and discharging causes the open-circuit voltage corresponding to the same SOC to differ depending on the current direction. Faced with these characteristics, traditional state estimation methods typically employ static average open-circuit voltage lookup tables or simple charge / discharge curve switching strategies, coupled with fixed filter gains. These methods struggle to accurately characterize battery dynamics under frequent charge / discharge switching conditions, making them susceptible to noise interference in the plateau region due to weak observational information, leading to overcorrection or instability. In the high-slope region, they may converge slowly due to gain mismatch, ultimately resulting in low accuracy and robustness in both battery SOC estimation. Summary of the Invention

[0003] This application provides a method, device, program product, and storage medium for estimating the state of charge (SOC) of a battery, achieving fast convergence speed, high accuracy, and long-term stable SOC estimation results.

[0004] The technical solution of this application embodiment is implemented as follows: This application provides a method for estimating the state of charge of a battery, the method comprising: Obtain the mapping relationship between the open-circuit voltage and the state of charge of the battery, the voltage measurement value and current measurement value of the battery, and the parameter set of the battery equivalent circuit model; Based on the mapping relationship and the parameter set, a battery equivalent circuit model containing hysteresis characteristics is determined. Based on the battery equivalent circuit model, the extended Kalman filter algorithm is used to predict the state of charge at the current moment and the voltage at the current moment. Based on the voltage measurement value at the current moment, the voltage prediction value at the current moment and the current measurement value at the current moment, it is determined whether the correction condition is met. If the correction condition is met, based on the target characteristic interval corresponding to the predicted state of charge at the current time, the Kalman gain of the extended Kalman filter algorithm is updated using the target gain update strategy corresponding to the target characteristic interval to obtain the target Kalman gain; wherein, different characteristic intervals correspond to different gain update strategies; The predicted state of charge at the current moment is updated based on the target Kalman gain to obtain the estimated state of charge at the current moment.

[0005] In the above scheme, determining the battery equivalent circuit model containing hysteresis characteristics based on the mapping relationship and the parameter set includes: A state vector is constructed based on the charged state variables, the first polarization voltage variable of the first RC element, and the second polarization voltage variable of the second RC element. Based on the mapping relationship and the direction of the current measurement, the open-circuit voltage containing hysteresis characteristics is determined; Based on the state vector, the open-circuit voltage with hysteresis characteristics, the current variable, and the ohmic internal resistance, an equivalent circuit model of the battery is constructed; wherein the parameter set includes at least the first RC stage, the second RC stage, and the ohmic internal resistance.

[0006] In the above scheme, the step of using the extended Kalman filter algorithm to predict the state of charge and voltage at the current moment based on the battery equivalent circuit model includes: Based on the physical relationship between the state components in the state vector, the battery equivalent circuit model is discretized to obtain the state transition equation of the extended Kalman filter algorithm. Based on the state prediction value and the current measurement value at the previous moment, the state prediction value at the current moment is calculated through the state transition equation; wherein, the state prediction value at the current moment includes the state of charge prediction value at the current moment. Based on the battery equivalent circuit model, the current state prediction value, and the current measurement value, the current voltage prediction value is determined.

[0007] In the above scheme, determining whether the correction condition is met based on the current voltage measurement value, the current voltage prediction value, and the current current measurement value includes: The target voltage is determined based on the measured voltage value at the current moment and the predicted voltage value at the current moment; If the target voltage is greater than or equal to the voltage threshold, and the current measurement value at the current moment is greater than or equal to the current threshold, then the correction condition is determined to be met.

[0008] In the above scheme, the step of updating the Kalman gain of the extended Kalman filter algorithm based on the target characteristic interval corresponding to the predicted state of charge at the current time, and using the target gain update strategy corresponding to the target characteristic interval to obtain the target Kalman gain, includes: Based on the mapping relationship, the charging state is divided into a high-slope region and a plateau region; If the predicted state of charge at the current moment is in the high slope region, the Kalman gain is updated using the first gain update strategy to obtain the target Kalman gain; If the predicted state of charge at the current moment is in the plateau region, the Kalman gain is updated using the second gain update strategy to obtain the target Kalman gain; wherein, the maximum Kalman gain allowed by the first gain update strategy is greater than the maximum Kalman gain allowed by the second gain update strategy.

[0009] In the above scheme, the voltage threshold is dynamically adjusted based on the battery's historical operating state information; wherein: The historical operating status information includes at least the total frequency of the predicted state of charge of the battery entering the plateau region within the target time period; the voltage threshold increases with the increase of the total frequency.

[0010] In the above scheme, updating the predicted state of charge (SOC) value at the current moment based on the target Kalman gain to obtain the estimated SOC value at the current moment includes: The predicted state of charge at the current moment is updated based on the target voltage and the target Kalman gain to obtain the estimated state of charge at the current moment.

[0011] This application provides a battery state-of-charge estimation device, the device comprising: The acquisition unit is used to acquire the mapping relationship between the open-circuit voltage and the state of charge of the battery, the voltage measurement value and current measurement value of the battery, and the parameter set of the battery equivalent circuit model; A construction unit is used to determine a battery equivalent circuit model containing hysteresis characteristics based on the mapping relationship and the parameter set. The processing unit is used to perform state prediction based on the battery equivalent circuit model using the extended Kalman filter algorithm, to obtain the predicted value of the state of charge and the predicted value of the voltage at the current moment, and to determine whether the correction conditions are met based on the measured value of the voltage at the current moment, the predicted value of the voltage at the current moment, and the measured value of the current moment. The processing unit is further configured to, if the correction condition is met, update the Kalman gain of the extended Kalman filter algorithm based on the target characteristic interval corresponding to the predicted state of charge at the current time, using the target gain update strategy corresponding to the target characteristic interval, to obtain the target Kalman gain; wherein, different characteristic intervals correspond to different gain update strategies. The determining unit is used to update the predicted value of the state of charge at the current moment based on the target Kalman gain, so as to obtain the estimated value of the state of charge at the current moment.

[0012] This application provides a battery state-of-charge estimation device, the device comprising: Memory is used to store executable instructions or computer programs. The processor, when executing computer-executable instructions or computer programs stored in the memory, implements the method provided in the embodiments of this application.

[0013] This application provides a computer program product, including a computer program or computer executable instructions, which, when executed by a processor, implement the method provided in this application.

[0014] This application provides a computer-readable storage medium storing a computer program or computer-executable instructions for implementing the method provided in this application when executed by a processor.

[0015] The embodiments of this application have the following beneficial effects: By acquiring the battery open-circuit voltage-state-of-charge mapping relationship, real-time measurement data, and equivalent circuit parameter set, and constructing a battery equivalent circuit model including hysteresis characteristics, the voltage prediction accuracy of the model under frequent battery charge-discharge switching conditions is fundamentally improved, laying an accurate model foundation for high-precision state estimation. Furthermore, based on this model, the extended Kalman filter algorithm is used for state prediction, and a correction condition judgment mechanism based on voltage error and current magnitude is introduced, effectively avoiding miscorrection when the observation information is unreliable (such as when the battery is stationary or at low current), and significantly improving the robustness of the algorithm in actual complex operating environments. Furthermore, when the correction conditions are met, the Kalman gain is adaptively adjusted according to the characteristic interval to which the predicted state of charge belongs (using the corresponding target gain update strategy). Finally, the predicted value is updated using this target Kalman gain, realizing intelligent and robust correction of the state of charge. Thus, through the synergistic effect of accurate hysteresis modeling, intelligent gating correction, and interval adaptive gain, the technical problems of inaccurate estimation and easy divergence of LFP batteries in the plateau region, as well as large model deviations caused by hysteresis effects under frequent switching conditions, are effectively solved. Finally, a state of charge estimation result with fast convergence speed, high accuracy, and long-term stability is obtained in the entire operating range. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the EKF in the battery state-of-charge estimation method provided in the embodiments of this application; Figure 2 This is a first flowchart illustrating the battery state-of-charge estimation method provided in this application embodiment; Figure 3 This is a schematic diagram of the second process of the battery state-of-charge estimation method provided in the embodiments of this application; Figure 4 This is a schematic diagram of the characteristic range in the battery state-of-charge estimation method provided in the embodiments of this application; Figure 5 This is a schematic diagram illustrating the determination of the estimated state of charge value in the battery state of charge estimation method provided in this application embodiment; Figure 6 This is a schematic diagram of the first experimental result in the battery state-of-charge estimation method provided in the embodiments of this application; Figure 7 This is a schematic diagram of the second experimental result in the battery state-of-charge estimation method provided in the embodiments of this application; Figure 8 This is a schematic diagram of the third experimental result in the battery state-of-charge estimation method provided in the embodiments of this application; Figure 9 This is a schematic diagram of the fourth experimental result in the battery state-of-charge estimation method provided in the embodiments of this application; Figure 10 This is a schematic diagram of the fifth experimental result in the battery state-of-charge estimation method provided in the embodiments of this application; Figure 11 This is a schematic diagram of the battery state-of-charge estimation device provided in an embodiment of this application; Figure 12 This is a schematic diagram of the battery state-of-charge estimation device provided in an embodiment of this application.

[0017] It should be noted that the terms "first" and "second" mentioned above are only used to distinguish between different options and do not represent the degree of superiority or inferiority of the options or their priority in the implementation process. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] It should be noted that against the backdrop of accelerated global energy structure transformation, the scale of grid-connected renewable energy sources such as wind power and photovoltaics continues to expand. The randomness and volatility of their output place higher demands on grid frequency stability, requiring power systems to have frequency regulation capabilities with second-level response to suppress frequency deviations. Energy storage, as a core support for frequency regulation, needs to balance power fluctuations in real time between primary frequency regulation (rapid adjustment) and secondary frequency regulation (longer-term control). For frequency regulation applications, current engineering practices widely adopt lithium iron phosphate (LFP) batteries as energy storage media due to their safety and cycle life advantages. At the same time, large-scale energy storage typically operates in a medium state of charge (SOC) range (approximately 40%–60%, usually around 50%) for extended periods during frequency regulation tasks. This operating characteristic constitutes a core requirement for the accuracy and stability of SOC estimation. However, the high-frequency charge / discharge switching and the randomness of power commands in frequency modulation scenarios pose multiple challenges to battery state estimation: on the one hand, it is necessary to provide a reliable SOC in the transient state to ensure power allocation and boundary management (such as State of Power (SOP) linkage); on the other hand, it is also necessary to balance lifespan and availability under hundreds of cycles per day to avoid overcharge / over-discharge risks and economic losses caused by SOC deviation.

[0020] From the perspective of electrochemical measurement characteristics, LFP exhibits a significant open circuit voltage (OCV) plateau phenomenon in the medium SOC range, meaning that OCV is insensitive to changes in SOC, leading to decreased system observability. However, in the low / high SOC edge region, the OCV-SOC slope is steep. This "plateau region – high slope region" characteristic makes it difficult to balance stability and accuracy across different ranges using a uniform gain strategy. Furthermore, LFP also exhibits a significant OCV hysteresis effect along the charge / discharge path, meaning that the open circuit voltage at the same SOC point differs during charging and discharging, forming a hysteresis loop. Ignoring hysteresis or replacing it with a single curve (or average OCV) in modeling and estimation directly introduces voltage prediction bias into the model, which further propagates to SOC estimation errors. This problem is particularly prominent in operating conditions such as frequency modulation where the charge / discharge direction frequently switches. The commonly used equivalent circuit model (ECM) in engineering is the mainstream for Battery Management Systems (BMS) due to its simple structure, easily identifiable parameters, and moderate computational load (see reference). Figure 1(As shown). However, under frequency modulation conditions, it needs to take into account the dynamics of multiple time scales (ohmic internal resistance and polarization process), and often uses a second-order RC network to characterize the fast and slow polarization processes respectively - which matches the time characteristics of primary and secondary frequency modulation. However, there is a problem: if the model parameters are not adapted with the SOC and operating conditions, or if the hysteresis effect and the difference in observability between intervals are not properly handled, the filtering algorithm is very prone to overcorrection in the plateau region, or numerical instability under external disturbances.

[0021] Based on this, embodiments of this application provide a method for estimating the state of charge (SOC) of a battery, applied to a battery SOC estimation device, with reference to... Figure 2 As shown, the specific steps may include: Step 101: Obtain the mapping relationship between the battery's open-circuit voltage and state of charge, the battery's voltage and current measurements, and the parameter set of the battery's equivalent circuit model.

[0022] In this embodiment, the mapping relationship between open-circuit voltage and state of charge refers to the set of open-circuit voltage data corresponding to different states of charge of the battery, which is experimentally calibrated. It is usually represented by an OCV-SOC lookup table or a fitted curve. The voltage measurement value and current measurement value refer to the voltage between the positive and negative terminals of the battery and the current flowing through the battery at multiple moments obtained by sensors in real time. The sign of the current is usually defined as positive for charging and negative for discharging. The parameter set of the battery equivalent circuit model can be determined offline. It is a set of parameters required in the equivalent circuit model that describes the dynamic electrical behavior of the battery. Specifically, it can include the ohmic internal resistance (R0), the resistance (R1, R2) and capacitance (C1, C2) of the two RC links (characterizing polarization processes with different time constants), and these parameters usually change with SOC. They can be identified offline (such as by the Particle Swarm Optimization (PSO) algorithm) to minimize the root mean square error (RMSE) between the model voltage and the measured voltage, forming a parameter table (i.e., ECM parameter table) interpolated with SOC and stored for online retrieval. In one feasible implementation, the battery can specifically be a lithium iron phosphate battery.

[0023] Step 102: Based on the mapping relationship and parameter set, determine the battery equivalent circuit model that includes hysteresis characteristics.

[0024] In this embodiment, the battery equivalent circuit model incorporating hysteresis characteristics differs from traditional equivalent circuit models. It expands the open-circuit voltage (OCV) from a static value solely related to the state of charge (SOC) to a dynamic variable simultaneously dependent on both SOC and the direction of current (i.e., the charging / discharging direction). (i.e., hysteresis OCV state) to simulate the hysteresis phenomenon where the battery charge and discharge voltage curves do not coincide. A state vector can be constructed first, and then, based on each state component, mapping relationship, and parameter set in the state vector, an equivalent battery circuit model incorporating hysteresis characteristics can be determined. This fundamentally improves the voltage prediction accuracy of the model under frequent battery charge and discharge switching conditions. By explicitly modeling the hysteresis effect, the systematic bias caused by the use of average OCV curves in traditional methods is overcome, laying an accurate model foundation for subsequent high-precision state estimation.

[0025] Step 103: Based on the battery equivalent circuit model, the extended Kalman filter algorithm is used to predict the state of charge and the voltage at the current moment. Based on the voltage measurement, voltage prediction, and current measurement at the current moment, it is determined whether the correction conditions are met.

[0026] In this embodiment, the extended Kalman filter (EKF) algorithm is used to discretize the battery equivalent circuit model to obtain the state transition equation. Based on the state estimate (including SOC and polarization voltage) at the previous moment and the current measurement value at the current moment, the predicted value of the state of charge (i.e., the prior value) and the predicted value of the voltage at the current moment output by the model are predicted using the state transition equation. Then, the voltage measurement value and the voltage prediction value at the current moment can be processed first, and then the processing result and the current measurement value at the current moment are judged to determine whether the current and voltage meet the correction conditions. If the correction conditions are met, the subsequent filtering correction steps are allowed to be executed. If the correction conditions are not met, only the predicted value of the state of charge is used, and no correction is performed.

[0027] It should be noted that the state prediction realizes the recursive tracking of the battery's internal state (SOC), and introduces a correction condition as a gating mechanism. When the voltage error is small or the current is too small (when the observation information is unreliable), the correction update of the Kalman filter is paused, thereby effectively avoiding miscorrection and filter divergence caused by measurement noise, instantaneous model error or static conditions, and significantly improving the robustness and reliability of the algorithm in real complex operating environments.

[0028] Step 104: If the correction condition is met, based on the target characteristic interval corresponding to the predicted state of charge at the current time, the Kalman gain of the extended Kalman filter algorithm is updated using the target gain update strategy corresponding to the target characteristic interval to obtain the target Kalman gain.

[0029] Different characteristic ranges correspond to different gain update strategies.

[0030] In this embodiment, the target characteristic interval refers to the SOC region divided according to the local slope characteristics of the mapping relationship (OCV-SOC), mainly divided into a high-slope region (OCV changes significantly with SOC) and a plateau region (OCV changes gently with SOC). The gain update strategy refers to the rules for constraining or adjusting the Kalman gain used for correction state calculation in EKF, and different gain update strategies are mainly reflected in the different maximum allowed Kalman gain values. When the correction condition is met, it is first determined whether the current SOC prediction value is in the high-slope region or the plateau region. Then, the gain update strategy matching the region is used to calculate the Kalman gain. In this way, the adaptive gain strategy of partition governance balances the convergence speed and stability of the estimation process. That is, in the high-slope region, accuracy and fast tracking are pursued, while in the plateau region, stability is prioritized and divergence is prevented. This greatly improves the adaptability, accuracy and long-term reliability of the SOC estimation algorithm across the entire SOC range.

[0031] Step 105: Update the predicted state of charge at the current moment based on the target Kalman gain to obtain the estimated state of charge at the current moment.

[0032] In this embodiment, the estimated state of charge (i.e., the posterior value) at the current moment is obtained by correcting the predicted state of charge at the current moment using the target Kalman gain; the target Kalman gain optimized by the aforementioned interval adaptive strategy is used to correct and update the prior state prediction (i.e., the predicted state of charge at the current moment) in the extended Kalman filter (EKF) framework, thereby obtaining the final output estimated state of charge at the current moment.

[0033] The battery state-of-charge estimation method provided in this application improves the voltage prediction accuracy under conditions of frequent current direction switching by acquiring the battery's mapping relationship and real-time data and constructing an equivalent circuit model containing hysteresis characteristics. This lays the foundation for high-precision estimation. Furthermore, after using extended Kalman filtering for state prediction, a correction condition based on voltage error and current magnitude is introduced, effectively filtering out conditions with unreliable observation information and significantly enhancing the robustness of the algorithm. Moreover, when the correction condition is met, the Kalman gain is adaptively adjusted according to the characteristic interval where the predicted state of charge value is located, using the corresponding gain update strategy. Through the synergistic effect of accurate hysteresis modeling, intelligent gating correction, and interval adaptive gain, the shortcomings of traditional methods, such as overcorrection in the plateau region and slow convergence in the high-slope region, are effectively overcome. Ultimately, this method achieves excellent results with high estimation accuracy and strong algorithm robustness across the entire state of charge range.

[0034] Based on the foregoing embodiments, this application provides yet another method for estimating the state of charge (SOC) of a battery, applied to a battery SOC estimation device, with reference to... Figure 2 As shown, the specific steps may include: Step 201: Obtain the mapping relationship between the battery's open-circuit voltage and state of charge, the battery's voltage and current measurements, and the parameter set of the battery's equivalent circuit model.

[0035] Step 202: Construct a state vector based on the charged state variables, the first polarization voltage variable of the first RC element, and the second polarization voltage variable of the second RC element.

[0036] In this embodiment, the state of charge variable is a key state quantity reflecting the remaining capacity of the battery, usually expressed as a percentage, and is a core parameter for the battery management system to perform energy management and life assessment; the first polarization voltage variable of the first RC stage refers to the voltage across the first RC stage used to simulate the rapid polarization process of the battery (usually corresponding to a shorter time constant), denoted as U1; the second polarization voltage variable refers to the voltage across the second RC stage used to simulate the slow polarization process of the battery (usually corresponding to a longer time constant), denoted as U2; the state vector refers to a vector composed of internal state variables; the state of charge variable reflecting the macroscopic energy state of the battery and the two polarization voltage variables U1 and U2 reflecting the microscopic polarization dynamics inside the battery are combined to form a three-dimensional state vector ( Specifically, it can be expressed as .

[0037] Step 203: Determine the open-circuit voltage containing hysteresis characteristics based on the mapping relationship and the direction of the current measurement value.

[0038] In this application embodiment, the open-circuit voltage including hysteresis characteristics refers to an improved concept of conventional battery open-circuit voltage, denoted as... It is not only a single-valued function of the state of charge, but also depends on the dynamic variables of the state of charge and the direction of current (charging or discharging). This variable simulates the hysteresis phenomenon of voltage path non-overlap during the charging and discharging process of lithium iron phosphate batteries. It is calculated based on the state of charge and real-time current direction (sign(I)) in the mapping relationship, through a specific dynamic relationship (such as a first-order inertial element) between the pre-calibrated charging and discharging master curves. This simulates the voltage hysteresis effect.

[0039] Step 204: Based on the state vector, open-circuit voltage and current variables including hysteresis characteristics, construct the battery equivalent circuit model.

[0040] The parameter set includes at least a first RC circuit, a second RC circuit, and an ohmic internal resistance.

[0041] In this embodiment, the traditional method of using a static average OCV lookup table or a simple charge / discharge curve switching method is abandoned. Instead, the dynamic hysteresis model is directly embedded into the state observation equation of the EKF. This makes OCV estimation no longer an isolated lookup behavior, but a part of the filter state, which can reflect the influence of historical current paths in real time. Specifically, by calculating the state vector, the open-circuit voltage containing hysteresis characteristics, the current variable (I), and the ohmic internal resistance (R0) in the parameter set, the battery equivalent circuit model is obtained, which can be expressed as follows: .

[0042] Step 205: Based on the physical relationship between the state components in the state vector, the battery equivalent circuit model is discretized to obtain the state transition equation of the extended Kalman filter algorithm.

[0043] In this embodiment, discretizing the battery equivalent circuit model refers to the process of transforming the continuous-time differential equation (based on physical laws) describing the battery's dynamic behavior into a discrete-time difference equation suitable for recursive calculation by a digital controller (such as a microprocessor in a BMS) at a fixed sampling period. This transformation is necessary because the EKF algorithm runs on a microcontroller with discrete time steps; the state transition equation, within the EKF framework, is a mathematical equation used to describe the system state (here...). How the system evolves from the previous time step (k-1) to the current time step (k) is essentially a discretized dynamic model of the system. Based on the physical relationships between the state components in the state vector (i.e., the charged state variable (SOC), the first polarization voltage variable (U1), and the second polarization voltage variable (U2))—the change in SOC follows the ampere-hour integral law, while the dynamics of the polarization voltage are described by the first-order differential equation of the RC circuit—the equivalent circuit model of the battery, which includes hysteresis characteristics, is discretized. This process uses numerical methods such as the Euler method to transform the continuous differential equation into a discrete difference equation, thus obtaining the state transition equation required by the Extended Kalman Filter (EKF) algorithm, whose expression is: , where f is the discretized state transition function, and the state transition equation defines how the state evolves from the previous time step to the current time step.

[0044] Step 206: Based on the state prediction value and the current measurement value at the previous moment, calculate the state prediction value at the current moment through the state transition equation.

[0045] The current state prediction value includes the current state of charge prediction value.

[0046] In this embodiment, the error covariance matrix is ​​used to quantify the uncertainty or magnitude of the error in the state estimate. Its diagonal elements represent the variance of the estimates of each state variable (such as SOC, U1, U2), and the off-diagonal elements represent the error correlation between the state variables. In EKF, the error covariance matrix evolves continuously with the prediction and update steps; the corrected state prediction value from the previous time step ( ) and the current measurement value at the previous moment ( Substituting this into the state transition equation, we can calculate the predicted state value at the current time. ), and the current state prediction value specifically includes the current SOC prediction value ( ), the predicted value of U1 at the current moment ( ) and the current U2 prediction value ( ),Right now .

[0047] It should be noted that it is also necessary to predict how the uncertainty of this state estimation propagates, i.e., to predict the error covariance matrix, which is calculated using the following formula: ,in, Is the state transition function f in The Jacobian matrix at a given point describes the local linear relationship of the state change; It is the error covariance matrix at the current moment. It is the error covariance matrix of the previous time step; It is the process noise covariance matrix, representing the uncertainty of the model itself and the unmodeled dynamics; therefore, this calculation formula represents the current uncertainty ( This stems from the uncertainty of the previous moment. The result after propagation through system dynamics (A), plus the process noise of the model itself ( ).

[0048] Step 207: Based on the battery equivalent circuit model, the current state prediction value, and the current current measurement value, determine the current voltage prediction value.

[0049] In this embodiment of the application, the current state prediction value (including , and ) and the current measurement value at the current moment ( By substituting the values ​​into the battery's equivalent circuit model, the predicted voltage value at the current moment can be obtained. ).

[0050] Step 208: Determine the target voltage based on the current voltage measurement value and the current voltage prediction value.

[0051] In this embodiment, the target voltage measures the accuracy of the model prediction. It can be obtained by subtracting the current voltage measurement and the current voltage prediction, and then performing an absolute value calculation. Specifically, it can characterize the voltage error and can be denoted as... Specifically, the calculation process for the target voltage can be expressed as follows: .

[0052] Step 209: If the target voltage is greater than or equal to the voltage threshold, and the current measurement value at the current moment is greater than or equal to the current threshold, then the correction condition is satisfied.

[0053] The voltage threshold is dynamically adjusted based on the battery's historical operating status information. The historical operating status information includes at least the total frequency of the battery's predicted state of charge entering the plateau region within the target time period. The voltage threshold increases with the increase of the total frequency.

[0054] In this embodiment of the application, the target voltage is greater than or equal to the voltage threshold ( This indicates that the current model prediction error exceeds the allowable threshold; the current measurement value at the current moment being greater than or equal to the current threshold indicates that the current battery is in a sufficiently obvious dynamic operating condition (charging or discharging), and its measurement information has sufficient reliability; the target voltage can be compared with the voltage threshold, and the current measurement value at the current moment can be compared with the current threshold, and the current threshold can be set to the minimum current ( Therefore, the activation correction condition is only satisfied when the target voltage is greater than or equal to the voltage threshold and the current measurement value at the current moment is large enough. It should be noted that the voltage threshold is not a fixed value, but can be dynamically adjusted. It is determined based on the total frequency of the predicted state of charge of the battery entering the plateau region (i.e., the low SOC count). This means that the algorithm has a certain memory capacity, and when the battery frequently enters the low SOC region with poor observability, the algorithm will become more cautious and increase the voltage threshold. This adaptive conservative strategy can effectively improve the long-term stability and reliability of the algorithm throughout its entire life cycle.

[0055] It should be noted that if historical records show that the battery frequently enters the plateau region (high total frequency), it indicates that the battery has been operating under conditions of low observability and easy estimation divergence for a long time. In order to improve long-term stability, the algorithm will adopt a more conservative strategy, that is, automatically increase the voltage threshold. This means that corrections are only allowed when the model predictions show a larger deviation, thus effectively suppressing overcorrection and estimation fluctuations caused by minor noise or model mismatch in the plateau region.

[0056] Step 210: If the correction conditions are met, the state of charge is divided based on the mapping relationship to obtain the high-slope region and the plateau region.

[0057] In the embodiments of this application, reference is made to Figure 4 As shown, the high-slope region refers to the SOC interval on the OCV-SOC curve where the absolute value of the local slope is relatively large. In these regions, small changes in SOC will cause significant changes in OCV, meaning that OCV is very sensitive to changes in SOC and the state is highly observable. The plateau region refers to the SOC interval on the OCV-SOC curve where the absolute value of the local slope is very small (especially the middle SOC range of LFP cells). In these regions, large changes in SOC will only cause small changes in OCV, meaning that OCV is not sensitive to changes in SOC and the state is poorly observable.

[0058] In this embodiment of the application, the continuous SOC range can be divided into intervals based on the local slope characteristics of the mapping relationship (OCV-SOC). Typically, a preset slope threshold is used for judgment, that is, the SOC value is substituted into the OCV-SOC curve (or its derivative curve). If the absolute value of the slope at that point is greater than the target threshold, it is determined to be a high slope region; otherwise, it is determined to be a plateau region.

[0059] It should be noted that after step 210, either step 211 or step 212 can be executed. Step 211: If the predicted state of charge at the current moment is in the high slope region, the Kalman gain is updated using the first gain update strategy to obtain the target Kalman gain.

[0060] In this embodiment, the first gain update strategy is a Kalman gain calculation and constraint rule applied to the high-slope region. Its core feature is that it allows a larger upper limit for the Kalman gain. When it is determined that the current SOC prediction value is in the high-slope region, the first gain update strategy is used to calculate and determine the target Kalman gain for final state correction. Specifically, in the EKF standard gain calculation formula ( Based on this, the first gain update strategy imposes a relatively loose constraint on the calculated gain, allowing it to reach a relatively high numerical upper limit to obtain the target Kalman gain. This means that in the high slope region, the algorithm believes that the small changes in the observed voltage do reflect the real changes in SOC, and therefore tends to trust the observation information and use a larger gain to quickly correct the SOC prediction value, thereby pursuing high accuracy and fast convergence tracking.

[0061] Step 212: If the predicted state of charge at the current moment is in the plateau region, the Kalman gain is updated using the second gain update strategy to obtain the target Kalman gain.

[0062] The maximum Kalman gain allowed by the first gain update strategy is greater than the maximum Kalman gain allowed by the second gain update strategy.

[0063] In this embodiment, the second gain update strategy is a Kalman gain calculation and constraint rule applied to the plateau region. Its core feature is to strictly limit the upper limit of the Kalman gain, keeping it at a small value. When it is determined that the current SOC prediction value is in the plateau region, the second gain update strategy can be adopted. After calculating the standard Kalman gain, the second gain update strategy will implement strict amplitude limiting, restricting the gain value (especially K_SOC) to a lower maximum allowable value. This is because in the plateau region, OCV is not sensitive to SOC, and small noise or model error in the observed voltage can easily be misjudged as a large change in SOC. If a large gain is still used at this time, it will cause the estimated value to overreact to noise, causing violent fluctuations or even divergence. Therefore, the second gain update strategy takes a conservative approach, suppressing the amplification of observation noise and model uncertainty by limiting the gain, and prioritizing the numerical stability and long-term reliability of the estimation process, even if this may come at the cost of sacrificing short-term convergence speed.

[0064] It should be noted that this application does not adopt the traditional single EKF correction strategy. Instead, it establishes a judgment mechanism for the high slope region and the plateau region based on the plateau region characteristics of the LFP battery OCV curve, and matches different gain control strategies (high slope region trust model, plateau region suppression update). This effectively solves the industry problem of inaccurate SOC estimation and easy divergence caused by the partial OCV insensitivity of LFP batteries.

[0065] It should be noted that step 213 can be executed after steps 211 and 212; Step 213: Update the predicted state of charge at the current moment based on the target voltage and the target Kalman gain to obtain the estimated state of charge at the current moment.

[0066] In this embodiment of the application, the estimated value of the state of charge at the current moment can be obtained by calculating the target voltage and the predicted value of the state of charge at the current moment, which can be denoted as: Specifically, it can be done through formulas. The calculation yielded that, This is an adaptive weighting coefficient that determines the extent to which we should believe that this observation bias is caused by the actual change in SOC; it should be noted that when the correction is in an active state (i.e., the condition in step 208 is met), Used; if not activated, then It is considered 0 at this time. .

[0067] It should be noted that after updating the state estimate, the error covariance matrix (P) of the state estimate also needs to be updated simultaneously to reflect the reduction in uncertainty of the state estimate after this correction. Typically, the Josephus form, which has better numerical stability, is used for updating to ensure that the error covariance matrix always remains symmetric and positive definite. The updated error covariance matrix will be used in the prediction step of the next time step. Simultaneously, the updated complete state vector (P) is also updated. The error covariance matrix (P) and the error covariance matrix (P) are stored as initial values ​​for the next round of EKF iteration (prediction-update loop).

[0068] It should be noted that this application provides a complete closed-loop system that includes offline calibration and online real-time estimation. For example... Figure 5 As shown, its core online process begins at the data input layer, where battery voltage and current are acquired in real time, along with the parameter set of the corresponding mapping relationship (OCV-SOC) and battery equivalent circuit model retrieved from the pre-stored database. Next, the algorithm enters the model construction and state prediction layer. Based on the open-circuit voltage model with hysteresis characteristics and the second-order RC equivalent circuit, a state-space equation is constructed, and the extended Kalman filter (EKF) is used for recursive prediction of the state and covariance, yielding the predicted state of charge (SOC) and voltage values. Subsequently, the algorithm enters the intelligent decision layer. First, a "correction activation gating" mechanism consisting of a dynamic voltage threshold and a fixed current threshold is used to determine whether the correction conditions are met. If met, the algorithm further switches to the corresponding adaptive gain update strategy (such as the first or second gain update strategy) based on the OCV-SOC curve interval (high slope region or plateau region) where the current SOC prediction value is located, calculating the constrained target Kalman gain. Finally, in the state update and output layer, the predicted state of charge is corrected using the target Kalman gain to obtain a high-precision estimate of the state of charge, which is then output. Simultaneously, the error covariance matrix is ​​updated to prepare for the next time step estimation. Furthermore, the architecture includes an offline parameter calibration module, which identifies model parameters using a particle swarm optimization (PSO) algorithm and determines the hysteresis factor through grid search. This provides an accurate model foundation for online estimation. The entire architecture enables collaborative work from data to model, and from prediction to adaptive correction, aiming to achieve robust and accurate SOC estimation under complex operating conditions.

[0069] It should be noted that this application aims to address two major challenges in estimating the state of charge (SOC) of lithium iron phosphate (LFP) batteries: firstly, the weak observability of the SOC-OCV curve plateau region leads to estimation divergence; secondly, the model bias introduced by the hysteresis effect of the charge / discharge path. To address these challenges, this scheme constructs an open-circuit voltage (OCV) model incorporating the hysteresis state and a second-order RC equivalent circuit. It employs an extended Kalman filter (EKF) with an interval adaptive gain mechanism, combined with a dual-threshold gating correction strategy for voltage and current. The hysteresis factor is determined through grid search, and the circuit parameters are identified using a particle swarm optimization (PSO) algorithm, supporting real-time interpolation updates with SOC. This method has low computational complexity, is easy to deploy, and is suitable for real-time applications in battery management systems (BMS) and frequency-modulated energy storage systems. The proposed estimation algorithm has been designed and experimentally verified based on the GB / T 38661-2020 standard. Specific verification results are as follows: Verification Result 1 (refer to...) Figure 6 As shown): Initial SOC = 90%, with a downward bias of 15%; experiments show that for the downward bias experiment with an initial SOC of 0.9, the average error of the model correction over the entire cycle is 0.26%, and the final convergence error is approximately 0.07%. Validation result 2 (refer to...) Figure 7 As shown): Initial SOC = 75%, with a downward bias of 15%; experiments show that for the downward bias experiment with an initial SOC of 0.75, the average error of the model correction over the entire cycle is 0.92%, and the final convergence error is approximately 0.07%. Validation result 3 (refer to...) Figure 8 As shown): Initial SOC = 75%, with an upward bias of 15%; experiments show that for the upward bias experiment with an initial SOC of 0.75, the average error of the model correction over the entire cycle is 0.47%, and the final convergence error is approximately 0.75%. Validation result 4 (refer to...) Figure 9 As shown in Figure 5): Initial SOC = 35%, with a downward bias of 15%; experiments show that for the downward bias experiment with an initial SOC of 0.35, the average error of the model correction over the entire cycle is 0.26%, and the final convergence error is approximately 0.07%. Validation result 5 (refer to...) Figure 10 As shown in the figure): the initial SOC=35%, the bias is set to: upper bias of 15%; the experiment shows that: for the lower bias experiment with an initial SOC of 0.75, the average error of the model correction over the whole cycle is 2.46%, and the final convergence error is about 2%.

[0070] It should be noted that the descriptions of the same steps and contents as in other embodiments in this embodiment can be found in the descriptions in other embodiments, and will not be repeated here.

[0071] The battery state-of-charge estimation method provided in this application improves the voltage prediction accuracy under conditions of frequent current direction switching by acquiring the battery's mapping relationship and real-time data and constructing an equivalent circuit model containing hysteresis characteristics. This lays the foundation for high-precision estimation. Furthermore, after using extended Kalman filtering for state prediction, a correction condition based on voltage error and current magnitude is introduced, effectively filtering out conditions with unreliable observation information and significantly enhancing the robustness of the algorithm. Moreover, when the correction condition is met, the Kalman gain is adaptively adjusted according to the characteristic interval where the predicted state of charge value is located, using the corresponding gain update strategy. Through the synergistic effect of accurate hysteresis modeling, intelligent gating correction, and interval adaptive gain, the shortcomings of traditional methods, such as overcorrection in the plateau region and slow convergence in the high-slope region, are effectively overcome. Ultimately, this method achieves excellent results with high estimation accuracy and strong algorithm robustness across the entire state of charge range.

[0072] Based on the foregoing embodiments, this application provides a battery state-of-charge estimation device, which can be applied to... Figure 2 and Figure 3 In the corresponding embodiment, the battery state-of-charge estimation method is provided, referring to Figure 11 As shown, the battery state-of-charge estimation device 3 may include: an acquisition unit 31, a construction unit 32, a processing unit 33, and a determination unit 34, wherein: The acquisition unit 31 is used to acquire the mapping relationship between the open-circuit voltage and the state of charge of the battery, the measured voltage and current values ​​of the battery, and the parameter set of the battery equivalent circuit model. Building unit 32 is used to determine the battery equivalent circuit model containing hysteresis characteristics based on the mapping relationship and parameter set; The processing unit 33 is used to perform state prediction based on the battery equivalent circuit model and the extended Kalman filter algorithm to obtain the predicted value of the state of charge and the predicted value of the voltage at the current time, and to determine whether the correction conditions are met based on the measured value of the voltage at the current time, the predicted value of the voltage at the current time and the measured value of the current time. The processing unit 33 is further configured to, if the correction condition is met, update the Kalman gain of the extended Kalman filter algorithm based on the target characteristic interval corresponding to the predicted state of charge at the current time, using the target gain update strategy corresponding to the target characteristic interval, to obtain the target Kalman gain; wherein, different characteristic intervals correspond to different gain update strategies. The determination unit 34 is used to update the predicted value of the state of charge at the current time based on the target Kalman gain, so as to obtain the estimated value of the state of charge at the current time.

[0073] In other embodiments of this application, the construction unit 32 is also configured to perform the following steps: A state vector is constructed based on the charged state variables, the first polarization voltage variable of the first RC element, and the second polarization voltage variable of the second RC element. Based on the mapping relationship and the direction of the current measurement, the open-circuit voltage containing hysteresis characteristics is determined; Based on the state vector, open-circuit voltage and current variables with hysteresis characteristics, and ohmic internal resistance, an equivalent circuit model of the battery is constructed; wherein the parameter set includes at least a first RC stage, a second RC stage, and ohmic internal resistance.

[0074] In other embodiments of this application, the processing unit 33 is further configured to perform the following steps: Based on the physical relationship between the state components in the state vector, the battery equivalent circuit model is discretized to obtain the state transition equation of the extended Kalman filter algorithm. Based on the state prediction value and the current measurement value at the previous moment, the state prediction value at the current moment is calculated through the state transition equation; wherein, the state prediction value at the current moment includes the state of charge prediction value at the current moment. Based on the battery equivalent circuit model, the current state prediction value, and the current current measurement value, the current voltage prediction value is determined.

[0075] In other embodiments of this application, the processing unit 33 is further configured to perform the following steps: Determine the target voltage based on the current voltage measurement and the current voltage prediction. If the target voltage is greater than or equal to the voltage threshold, and the current measurement value at the current moment is greater than or equal to the current threshold, the correction condition is determined to be met.

[0076] In other embodiments of this application, the processing unit 33 is further configured to perform the following steps: Based on the mapping relationship, the charging state is divided into high-slope region and plateau region; If the predicted state of charge at the current moment is in the high slope region, the Kalman gain is updated using the first gain update strategy to obtain the target Kalman gain. If the predicted state of charge at the current moment is in the plateau region, the Kalman gain is updated using the second gain update strategy to obtain the target Kalman gain; wherein, the maximum Kalman gain allowed by the first gain update strategy is greater than the maximum Kalman gain allowed by the second gain update strategy.

[0077] In other embodiments of this application, the processing unit 33 is further configured to perform the following steps: The voltage threshold is dynamically adjusted based on the battery's historical operating status information; Historical operating status information includes at least the total frequency of the predicted state of charge of the battery entering the plateau region within the target time period; the voltage threshold increases with the increase of the total frequency.

[0078] In other embodiments of this application, the determining unit 34 is further configured to perform the following steps: The predicted state of charge (SOC) at the current moment is updated based on the target voltage and the target Kalman gain to obtain the estimated SOC at the current moment.

[0079] It should be noted that the specific implementation process of the steps performed by each module in the embodiments of this application can be referred to Figure 2 and Figure 3 The implementation process of the battery state-of-charge estimation method provided in the corresponding embodiment will not be described in detail here.

[0080] The battery state-of-charge estimation device provided in this application improves the voltage prediction accuracy of the model under conditions of frequent current direction switching by acquiring the battery mapping relationship and real-time data and constructing an equivalent circuit model containing hysteresis characteristics. This lays the model foundation for high-precision estimation. Furthermore, after using extended Kalman filtering for state prediction, a correction condition based on voltage error and current magnitude is introduced, effectively filtering out conditions with unreliable observation information and significantly enhancing the robustness of the algorithm. Moreover, when the correction condition is met, the Kalman gain is adaptively adjusted according to the characteristic interval where the predicted state of charge value is located, using the corresponding gain update strategy. Thus, through the synergistic effect of accurate hysteresis modeling, intelligent gating correction, and interval adaptive gain, the defects of traditional methods such as overcorrection in the plateau region and slow convergence in the high-slope region are effectively overcome. Ultimately, it achieves excellent results with high estimation accuracy and strong algorithm robustness across the entire state of charge range.

[0081] Based on the foregoing embodiments, embodiments of this application provide a battery state of charge estimation device, which can be applied to... Figure 2 and Figure 3 In the corresponding embodiment, the battery state-of-charge estimation method is provided, referring to Figure 12 As shown, the battery state-of-charge estimation device 4 may include: a processor 41, a memory 42, and a communication bus 43, wherein: Communication bus 43 is used to realize the communication connection between processor 41 and memory 42; The processor 41 is used to execute the battery state-of-charge estimation program in the memory 42 to perform the following steps: Obtain the mapping relationship between the battery's open-circuit voltage and state of charge, the battery's voltage and current measurements, and the parameter set of the battery's equivalent circuit model; Based on the mapping relationship and parameter set, the equivalent circuit model of the battery containing hysteresis characteristics is determined. Based on the battery equivalent circuit model, the extended Kalman filter algorithm is used for state prediction to obtain the predicted value of the state of charge and the predicted value of the voltage at the current time. Based on the measured value of the voltage, the predicted value of the voltage at the current time, and the measured value of the current time, it is determined whether the correction condition is met. If the correction condition is met, based on the target characteristic interval corresponding to the predicted state of charge at the current time, the Kalman gain of the extended Kalman filter algorithm is updated using the target gain update strategy corresponding to the target characteristic interval to obtain the target Kalman gain; where different characteristic intervals correspond to different gain update strategies. The predicted state of charge at the current moment is updated based on the target Kalman gain to obtain the estimated state of charge at the current moment.

[0082] In other embodiments of this application, processor 41 is used to execute a battery state-of-charge estimation program in memory 42 based on mapping relationships and parameter sets to determine a battery equivalent circuit model containing hysteresis characteristics, in order to perform the following steps: A state vector is constructed based on the charged state variables, the first polarization voltage variable of the first RC element, and the second polarization voltage variable of the second RC element. Based on the mapping relationship and the direction of the current measurement, the open-circuit voltage containing hysteresis characteristics is determined; Based on the state vector, open-circuit voltage and current variables with hysteresis characteristics, and ohmic internal resistance, an equivalent circuit model of the battery is constructed; wherein the parameter set includes at least a first RC stage, a second RC stage, and ohmic internal resistance.

[0083] In other embodiments of this application, the processor 41 is used to execute the battery state-of-charge estimation program in the memory 42 based on the battery equivalent circuit model, and to perform state prediction using the extended Kalman filter algorithm to obtain the predicted value of the state of charge and the predicted value of the voltage at the current moment, so as to implement the following steps: Based on the physical relationship between the state components in the state vector, the battery equivalent circuit model is discretized to obtain the state transition equation of the extended Kalman filter algorithm. Based on the state prediction value and the current measurement value at the previous moment, the state prediction value at the current moment is calculated through the state transition equation; wherein, the state prediction value at the current moment includes the state of charge prediction value at the current moment. Based on the battery equivalent circuit model, the current state prediction value, and the current current measurement value, the current voltage prediction value is determined.

[0084] In other embodiments of this application, the processor 41 is configured to execute a battery state-of-charge estimation program in the memory 42, based on the current voltage measurement, the current voltage prediction, and the current current measurement, to determine whether a correction condition is met, in order to implement the following steps: Determine the target voltage based on the current voltage measurement and the current voltage prediction. If the target voltage is greater than or equal to the voltage threshold, and the current measurement value at the current moment is greater than or equal to the current threshold, the correction condition is determined to be met.

[0085] In other embodiments of this application, the processor 41 is used to execute the target characteristic interval corresponding to the predicted value of the battery state of charge at the current time in the battery state of charge estimation program stored in the memory 42, and to update the Kalman gain of the extended Kalman filter algorithm using the target gain update strategy corresponding to the target characteristic interval to obtain the target Kalman gain, so as to implement the following steps: Based on the mapping relationship, the charging state is divided into high-slope region and plateau region; If the predicted state of charge at the current moment is in the high slope region, the Kalman gain is updated using the first gain update strategy to obtain the target Kalman gain. If the predicted state of charge at the current moment is in the plateau region, the Kalman gain is updated using the second gain update strategy to obtain the target Kalman gain; wherein, the maximum Kalman gain allowed by the first gain update strategy is greater than the maximum Kalman gain allowed by the second gain update strategy.

[0086] In other embodiments of this application, the processor 41 is used to execute a battery state-of-charge estimation program in the memory 42 to perform the following steps: The voltage threshold of the battery state-of-charge estimation program is dynamically adjusted based on the battery's historical operating state information; Historical operating status information includes at least the total frequency of the predicted state of charge of the battery entering the plateau region within the target time period; the voltage threshold increases with the increase of the total frequency.

[0087] In other embodiments of this application, the processor 41 is used to execute the battery state-of-charge estimation program in the memory 42 to update the predicted value of the current state of charge based on the target Kalman gain, so as to obtain the estimated value of the current state of charge, in order to implement the following steps: The predicted state of charge (SOC) at the current moment is updated based on the target voltage and the target Kalman gain to obtain the estimated SOC at the current moment.

[0088] It should be noted that a detailed description of the steps performed by the processor can be found in [reference needed]. Figure 2 and Figure 3The implementation process of the battery state-of-charge estimation method provided in the corresponding embodiment will not be described in detail here.

[0089] The battery state-of-charge estimation device provided in this application improves the voltage prediction accuracy under conditions of frequent current direction switching by acquiring the battery's mapping relationship and real-time data, and constructing an equivalent circuit model containing hysteresis characteristics. This lays the foundation for high-precision estimation. Furthermore, after using extended Kalman filtering for state prediction, a correction condition based on voltage error and current magnitude is introduced, effectively filtering out conditions with unreliable observation information and significantly enhancing the robustness of the algorithm. Moreover, when the correction condition is met, the Kalman gain is adaptively adjusted according to the characteristic interval where the predicted state of charge value is located, using the corresponding gain update strategy. Through the synergistic effect of accurate hysteresis modeling, intelligent gating correction, and interval adaptive gain, the defects of traditional methods such as overcorrection in the plateau region and slow convergence in the high-slope region are effectively overcome. Ultimately, it achieves excellent results with high estimation accuracy and strong algorithm robustness across the entire state of charge range.

[0090] Based on the foregoing embodiments, this application provides a computer program product, including a computer program, which implements [the following] when executed by a processor. Figure 2 and Figure 3 The steps in the battery state-of-charge estimation method provided in the corresponding embodiment.

[0091] Based on the foregoing embodiments, this application provides a computer-readable storage medium storing one or more programs that can be executed by one or more processors to achieve... Figure 2 and Figure 3 The steps in the battery state-of-charge estimation method provided in the corresponding embodiment.

[0092] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, and improvements made within the spirit and scope of this application are included within the scope of protection of this application.

Claims

1. A method for estimating the state of charge of a battery, characterized in that, The method includes: Obtain the mapping relationship between the open-circuit voltage and the state of charge of the battery, the voltage measurement value and current measurement value of the battery, and the parameter set of the battery equivalent circuit model; Based on the mapping relationship and the parameter set, a battery equivalent circuit model containing hysteresis characteristics is determined. Based on the battery equivalent circuit model, the extended Kalman filter algorithm is used to predict the state of charge at the current moment and the voltage at the current moment. Based on the voltage measurement value at the current moment, the voltage prediction value at the current moment and the current measurement value at the current moment, it is determined whether the correction condition is met. If the correction condition is met, based on the target characteristic interval corresponding to the predicted state of charge at the current time, the Kalman gain of the extended Kalman filter algorithm is updated using the target gain update strategy corresponding to the target characteristic interval to obtain the target Kalman gain; wherein, different characteristic intervals correspond to different gain update strategies; The predicted state of charge at the current moment is updated based on the target Kalman gain to obtain the estimated state of charge at the current moment.

2. The method according to claim 1, characterized in that, The process of determining the battery equivalent circuit model containing hysteresis characteristics based on the mapping relationship and the parameter set includes: A state vector is constructed based on the charged state variables, the first polarization voltage variable of the first RC element, and the second polarization voltage variable of the second RC element. Based on the mapping relationship and the direction of the current measurement, the open-circuit voltage containing hysteresis characteristics is determined; Based on the state vector, the open-circuit voltage with hysteresis characteristics, the current variable, and the ohmic internal resistance, an equivalent circuit model of the battery is constructed; wherein the parameter set includes at least the first RC stage, the second RC stage, and the ohmic internal resistance.

3. The method according to claim 2, characterized in that, Based on the battery equivalent circuit model, the extended Kalman filter algorithm is used for state prediction to obtain the predicted state of charge and the predicted voltage at the current moment, including: Based on the physical relationship between the state components in the state vector, the battery equivalent circuit model is discretized to obtain the state transition equation of the extended Kalman filter algorithm. Based on the state prediction value and the current measurement value at the previous moment, the state prediction value at the current moment is calculated through the state transition equation; wherein, the state prediction value at the current moment includes the state of charge prediction value at the current moment. Based on the battery equivalent circuit model, the current state prediction value, and the current measurement value, the current voltage prediction value is determined.

4. The method according to claim 1, characterized in that, The determination of whether the correction condition is met based on the current voltage measurement value, the current voltage prediction value, and the current current measurement value includes: The target voltage is determined based on the measured voltage value at the current moment and the predicted voltage value at the current moment; If the target voltage is greater than or equal to the voltage threshold, and the current measurement value at the current moment is greater than or equal to the current threshold, then the correction condition is determined to be met.

5. The method according to claim 4, characterized in that, The target Kalman gain is obtained by updating the Kalman gain of the extended Kalman filter algorithm based on the target characteristic interval corresponding to the predicted state of charge at the current time, using the target gain update strategy corresponding to the target characteristic interval, including: Based on the mapping relationship, the charging state is divided into a high-slope region and a plateau region; If the predicted state of charge at the current moment is in the high slope region, the Kalman gain is updated using the first gain update strategy to obtain the target Kalman gain; If the predicted state of charge at the current moment is in the plateau region, the Kalman gain is updated using the second gain update strategy to obtain the target Kalman gain; wherein, the maximum Kalman gain allowed by the first gain update strategy is greater than the maximum Kalman gain allowed by the second gain update strategy.

6. The method according to claim 4, characterized in that, The voltage threshold is dynamically adjusted based on the battery's historical operating status information; wherein: The historical operating status information includes at least the total frequency of the predicted state of charge of the battery entering the plateau region within the target time period; the voltage threshold increases with the increase of the total frequency.

7. The method according to claim 4, characterized in that, The step of updating the predicted state of charge (SOC) at the current moment based on the target Kalman gain to obtain the estimated SOC at the current moment includes: The predicted state of charge at the current moment is updated based on the target voltage and the target Kalman gain to obtain the estimated state of charge at the current moment.

8. A battery state-of-charge estimation device, characterized in that, The device includes: Memory is used to store executable instructions or computer programs. A processor, when executing computer-executable instructions or computer programs stored in the memory, implements the method according to any one of claims 1 to 7.

9. A computer program product, comprising a computer program, characterized in that, The computer program, when executed by a processor, implements the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores one or more programs that can be executed by one or more processors to implement the method of any one of claims 1 to 7.

Images