Energy storage system state of charge estimation method, device and program product

Abstract

The invention is suitable for the technical field of battery management, and provides an energy storage system charge state estimation method and device and a program product, and the method comprises the steps: employing a first estimation algorithm to estimate the charge state of a battery; wherein the first estimation algorithm is a filtering algorithm based on a lightweight battery model, and the lightweight battery model is obtained through identification based on test data collected in a non-platform working interval of the battery; and according to the real-time working condition parameters of the battery, adaptively adjusting noise covariance parameters in the first estimation algorithm. The complexity of model identification and the data storage requirement are greatly reduced, and meanwhile, the influence of process noise and measurement noise is effectively inhibited.

Description

Technical Field

[0001] This application belongs to the field of battery management technology, and in particular relates to a method, device and program product for estimating the state of charge of an energy storage system. Background Technology

[0002] State of Charge (SOC) is a core parameter of energy storage systems, and its estimation directly affects the operational safety, lifespan, and overall system performance of the energy storage battery. Lithium iron phosphate (LFP) batteries have become the mainstream choice for energy storage systems due to their advantages of low cost, long cycle life, and high safety; however, SOC estimation based on these batteries still faces key technical challenges.

[0003] In existing technologies, battery model-based filtering algorithms (such as various Kalman filtering algorithms) are commonly used for SOC estimation, but their effectiveness is highly dependent on the adaptability of the battery model. Traditional battery models require extensive test data for parameter identification to cover various battery operating characteristics, resulting in a large number of model parameters. Energy storage systems often need to adapt to cells from different suppliers and models; storing multiple sets of model parameters would occupy a significant amount of controller flash memory, creating a significant conflict with the limited hardware storage resources of the energy storage controller. Furthermore, the actual operating conditions of energy storage batteries are dynamically changing, placing higher demands on the parameter adaptability of filtering algorithms, making it difficult to maintain stable SOC estimation accuracy, further increasing the difficulty of stable application in complex real-world scenarios. Summary of the Invention

[0004] This application provides a method, device, and program product for estimating the state of charge of an energy storage system. It can solve the technical problems in the prior art where storing multiple sets of model parameters in the controller is contradictory to the limited storage resources, and the estimation accuracy is unstable due to insufficient adaptation of filter algorithm parameters under dynamic changes in operating conditions.

[0005] In a first aspect, embodiments of this application provide a method for estimating the state of charge (SOC) of an energy storage system, including:

[0006] The state of charge of the battery is estimated using a first estimation algorithm; wherein the first estimation algorithm is a filtering algorithm based on a lightweight battery model, the lightweight battery model being identified based on test data collected from the non-platform operating range of the battery;

[0007] Furthermore, the noise covariance parameter in the first estimation algorithm is adaptively adjusted based on the real-time operating parameters of the battery.

[0008] In one possible implementation of the first aspect, the method further includes:

[0009] Obtain the current state parameters of the battery, and determine the current working range of the battery based on the current state parameters;

[0010] When the battery is in a non-platform operating range, the step of estimating the state of charge of the battery using the first estimation algorithm is executed;

[0011] When the battery is in the platform operating range, the state of charge of the battery is estimated by a second estimation algorithm based on the integration of the battery current.

[0012] In one possible implementation of the first aspect, determining the current operating range of the battery based on the current state parameter includes:

[0013] Compare the current state of charge estimate with the preset upper limit threshold and lower limit threshold of the state of charge.

[0014] If the current state of charge estimate is less than the lower limit threshold of the state of charge, or greater than the upper limit threshold of the state of charge, then the battery is determined to be currently in the non-platform operating range.

[0015] If the current state of charge estimate is less than the upper limit threshold of the state of charge and greater than the lower limit threshold of the state of charge, then the battery is determined to be currently in the platform's operating range.

[0016] or,

[0017] Obtain the local slope of the state-of-charge-open-circuit voltage curve corresponding to the current state-of-charge estimate;

[0018] The local slope is compared with a preset slope threshold;

[0019] If the local slope is greater than the preset slope threshold, then the battery is determined to be currently in the non-platform working range;

[0020] If the local slope is less than the preset slope threshold, then the battery is determined to be currently in the platform's working range.

[0021] In one possible implementation of the first aspect, the construction process of the lightweight battery model includes:

[0022] The battery is tested within the non-platform operating range to collect test data; wherein, the test data includes measured voltage, measured current, and temperature;

[0023] Based on the test data and the preset second-order RC equivalent circuit model, the output model predicts the voltage.

[0024] An objective function is constructed with the goal of minimizing the error between the measured voltage and the model-predicted voltage.

[0025] The objective function is iteratively optimized using the nonlinear least squares method to identify the parameters of the second-order RC equivalent circuit model within the non-platform operating range, thereby constructing the lightweight battery model.

[0026] In one possible implementation of the first aspect, the filtering algorithm is an extended Kalman filter algorithm; the parameters of the second-order RC equivalent circuit model in the non-platform operating range include: ohmic internal resistance R0, first polarization resistor R1, first polarization capacitor C1, second polarization resistor R2, and second polarization capacitor C2, and the first polarization resistor R1 and the first polarization capacitor C1 form a first RC network, and the second polarization resistor R2 and the second polarization capacitor C2 form a second RC network;

[0027] The construction process of the extended Kalman filter algorithm includes:

[0028] The state of charge, electrochemical polarization voltage, and concentration polarization voltage to be estimated are used as state vectors; wherein, the electrochemical polarization voltage is obtained based on the first polarization resistor R1 and the first polarization capacitor C1; and the concentration polarization voltage is obtained based on the second polarization resistor R2 and the second polarization capacitor C2.

[0029] Based on the dynamic characteristics of the lightweight battery model, a state equation f(·) is constructed to describe the update relationship of the state vector with the measured current and sampling time.

[0030] Based on the aforementioned lightweight battery model, observation equations are constructed to establish the relationship between the state vector and the measured terminal voltage z. k The mapping relationship.

[0031] In one possible implementation of the first aspect, estimating the state of charge of the battery using the first estimation algorithm includes:

[0032] Using the state equation f(·), the state vector updated from the previous time step is obtained. Given the input current at time k, calculate the predicted state vector at time k.

[0033] Based on the observation equation h(·), the predicted state vector Calculate the predicted observations

[0034] Calculate the Kalman gain K k ;

[0035] Based on the predicted state vector The Kalman gain Kk The measured terminal voltage z k and the predicted observations Obtain the updated state vector at time k. Among them, in the updated state vector In this study, the component of the state of charge is used as the estimated value of the battery's state of charge.

[0036] In one possible implementation of the first aspect, the real-time operating parameters include at least one of the following: measured terminal voltage z. k Measured current, cell temperature, ambient temperature, and charge / discharge rate;

[0037] The noise covariance parameter includes at least one of the following: process noise covariance and measurement noise covariance.

[0038] In one possible implementation of the first aspect, the method further includes:

[0039] Based on the state of charge estimate obtained by the first estimation algorithm within the non-platform operating range, the initial value of the second estimation algorithm is adjusted.

[0040] Secondly, embodiments of this application provide a state of charge estimation device for an energy storage system, comprising:

[0041] The first estimation module is used to estimate the state of charge of the battery using a first estimation algorithm; wherein the first estimation algorithm is a filtering algorithm based on a lightweight battery model, and the lightweight battery model is identified based on test data collected from the non-platform operating range of the battery;

[0042] An adaptive adjustment module is used to adaptively adjust the noise covariance parameter in the first estimation algorithm based on the real-time operating parameters of the battery.

[0043] Thirdly, embodiments of this application provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the energy storage system state of charge estimation method described in any of the first aspects above.

[0044] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the energy storage system state-of-charge estimation method described in any one of the first aspects.

[0045] Fifthly, embodiments of this application provide a computer program product that, when run on a computer device, causes the computer device to execute the energy storage system state of charge estimation method described in any of the first aspects above.

[0046] In this embodiment, a first estimation algorithm is used to estimate the state of charge (SOC) of a battery. This first estimation algorithm is a filtering algorithm based on a lightweight battery model, which is identified from test data collected within the battery's non-platform operating range. Furthermore, the noise covariance parameter in the first estimation algorithm is adaptively adjusted according to the battery's real-time operating parameters. By constructing the lightweight battery model through parameter identification only on test data from the non-platform region, the complexity of model identification and data storage requirements are significantly reduced, enabling easy deployment in energy storage controllers with limited storage resources. Simultaneously, the filtering algorithm possesses closed-loop correction characteristics that fuse model predictions with sensor measured data, effectively suppressing the influence of process noise and measurement noise, and achieving reliable estimation of the SOC within the non-platform operating range.

[0047] It is understood that the beneficial effects of the second to fifth aspects mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments of this application, 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 this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 This is a schematic flowchart of a method for estimating the state of charge of an energy storage system according to an embodiment of this application;

[0050] Figure 2 This is a schematic flowchart of a method for estimating the state of charge of an energy storage system according to another embodiment of this application;

[0051] Figure 3 This is a flowchart of an embodiment of the extended Kalman filter for implementing state of charge estimation.

[0052] Figure 4 This is a schematic flowchart of a method for estimating the state of charge of an energy storage system according to another embodiment of this application;

[0053] Figure 5 This is a curve comparing the accuracy of the interval-based State of Charge (SOC) estimation method provided in this application embodiment with that of the single ampere-hour integral method;

[0054] Figure 6 This is a schematic diagram of the structure of the energy storage system state of charge estimation device provided in the embodiments of this application;

[0055] Figure 7 This is a schematic diagram of the structure of the computer device provided in the embodiments of this application. Detailed Implementation

[0056] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0057] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0058] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0059] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0060] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0061] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0062] Figure 1 A schematic flowchart of a method for estimating the state of charge of an energy storage system according to an embodiment of this application is shown.

[0063] S101, the state of charge of the battery is estimated using a first estimation algorithm; wherein, the first estimation algorithm is a filtering algorithm based on a lightweight battery model, and the lightweight battery model is identified based on test data collected from the non-platform operating range of the battery.

[0064] The first estimation algorithm refers to a filtering estimation algorithm based on a lightweight battery model. Its purpose is to indirectly estimate the state of charge (SOC) of a battery. Its core feature is to integrate the predictive ability of the battery model with the correction ability of the sensor's measured data to achieve high-precision state estimation.

[0065] Lightweight battery model: A dynamic characteristic model of battery built to adapt to the limited hardware resources of energy storage controller. Its feature is that it completes parameter identification only through test data in the non-platform operating range of the battery, without relying on full-condition data. While ensuring estimation accuracy, it greatly simplifies the model complexity and reduces the amount of parameter storage.

[0066] Non-platform operating range: The specific operating range of the battery's state of charge (SOC). Within this range, the battery's open-circuit voltage (OCV) is significantly sensitive to changes in SOC, providing effective observation information for filtering algorithms and serving as a data source range for identifying parameters in lightweight battery models.

[0067] Test data identification: In offline mode, battery operation data within the non-platform working range is collected through standardized testing, and then mathematical optimization methods (such as nonlinear least squares method) are used to iteratively solve the problem to determine the key parameters of the lightweight battery model, so that the error between the model output and the measured data meets the preset requirements.

[0068] Optionally, the filtering algorithm is an extended Kalman filter algorithm.

[0069] In this embodiment, the battery's state of charge is first loaded with pre-collected standardized test data and lightweight battery model parameters within the battery's non-platform operating range. Then, the measured data during the battery's operation is collected by sensors. Subsequently, a filtering estimation algorithm based on the lightweight battery model is used to indirectly estimate the battery's state of charge by combining the model's predictive ability with the correction ability of the sensor's measured data, and the state of charge estimation result is output.

[0070] S102, based on the real-time operating parameters of the battery, adaptively adjust the noise covariance parameter in the first estimation algorithm.

[0071] Among them, real-time operating parameters refer to key parameters that change dynamically during battery operation and can reflect the current operating status, including but not limited to the measured terminal voltage z. k Measured current, cell temperature, ambient temperature, and charge / discharge rate are the core basis for judging operating condition fluctuations and triggering parameter adjustments.

[0072] Adaptive adjustment: refers to the process of dynamically optimizing algorithm parameters based on preset rules or logic and real-time collected operating condition parameters.

[0073] Noise covariance parameter: refers to the key parameter used in the filtering algorithm to quantify uncertainty, including process noise covariance and measurement noise covariance. The former reflects the dynamic characteristic uncertainty of the lightweight battery model itself, while the latter reflects the error uncertainty of the sensor measured data. Its value directly affects the weight allocation of the filtering algorithm to the model prediction and the measured data.

[0074] In this embodiment, the real-time operating parameters of the battery are first collected synchronously by sensors. Then, the trend of operating condition changes is determined according to preset adaptation rules (such as whether the charge / discharge rate increases or whether the cell temperature deviates from the reference range). Subsequently, the process noise covariance and / or measurement noise covariance in the first estimation algorithm are adjusted accordingly. For example, when the operating condition fluctuation is small, the parameters are kept relatively stable to maintain the continuity of estimation; when the operating condition fluctuation is large, the parameter weights are optimized to enhance the algorithm's dependence on measured data. Finally, the adjusted noise covariance parameters are fed back to the first estimation algorithm in real time for subsequent iterative calculation of state of charge estimation, thereby achieving accurate adaptation of the algorithm to dynamic operating conditions.

[0075] In this embodiment, a lightweight battery model is constructed by identifying parameters only from test data in the non-platform region, and the battery model parameters are identified offline. Considering the significant differences in the dynamic characteristics of lithium iron phosphate batteries at different charge / discharge rates, their equivalent circuit model parameters (such as ohmic internal resistance) change with the current rate. Traditional solutions require meticulous identification for each charge / discharge rate, which is cumbersome and has limited adaptability. However, the Kalman filter algorithm (first estimation algorithm) used in this embodiment dynamically updates the process noise covariance and measurement noise covariance, enabling it to adaptively respond to dynamic changes in actual operating conditions. It maintains high-precision state-of-charge estimation without relying on the cumbersome parameter identification process for different charge / discharge rates. Meanwhile, this embodiment strictly limits the range of parameter identification to the non-platform region of the lithium iron phosphate cell (i.e., the state of charge (SOC) range where voltage changes are significant). Within this range, the battery electrical parameters are more sensitive and more identifiable, which not only effectively captures the core dynamic characteristics of the battery, but also significantly reduces the dimensionality and computational burden of parameter identification, further enhancing the lightweight characteristics of the battery model. While ensuring estimation accuracy, it improves the real-time performance and engineering feasibility of the system.

[0076] Figure 2 A schematic flowchart of a method for estimating the state of charge of an energy storage system according to another embodiment of this application is shown.

[0077] S201, Obtain the current state parameters of the battery, and determine the current working range of the battery based on the current state parameters.

[0078] Among them, the current state parameters refer to the parameters that can characterize the current operating state of the battery and are used to determine the operating range, including but not limited to the estimated value of the current state of charge (SOC) and the local slope of the state of charge-open circuit voltage (SOC-OCV) curve corresponding to the current state of charge, which are the basis for distinguishing between the platform operating range and the non-platform operating range.

[0079] Operating range: refers to the operating range of a battery based on the correlation characteristics of its state of charge (SOC) and open-circuit voltage (OCV). It is divided into two categories: platform operating range and non-platform operating range. Different ranges correspond to different state of charge estimation algorithms, which is the basis for achieving accurate estimation in different ranges.

[0080] In this embodiment, the current state parameters, such as the estimated state of charge (SOC) and the local slope of the SOC-OCV curve, are first obtained through sensor or algorithm iteration results. Then, these parameters are analyzed and judged according to preset rules to determine whether the battery is currently in a platform working range or a non-platform working range, providing a clear basis for subsequent targeted selection of the first estimation algorithm or the second estimation algorithm.

[0081] S202, when the battery is in a non-platform operating range, the step of estimating the state of charge of the battery using the first estimation algorithm is executed.

[0082] Among them, the non-platform operating range is the specific operating range of the battery's state of charge (SOC). Within this range, the battery's open-circuit voltage (OCV) is highly sensitive to changes in SOC, which can provide effective observation information for filtering algorithms and is suitable for filtering estimation algorithms based on lightweight battery models.

[0083] In this embodiment, the judgment result that the battery is in the non-platform working range output by S201 is first received. Then, the first estimation algorithm based on the lightweight battery model is triggered and executed. The real-time operating parameters of the battery are collected synchronously and the noise covariance parameter in the algorithm is adaptively adjusted. By utilizing the correlation characteristics between the battery open circuit voltage and the state of charge in this range, combined with the prediction and correction capabilities of the algorithm, a stable and reliable battery state of charge estimation result in the non-platform working range is finally output.

[0084] S203, when the battery is in the platform working range, the state of charge of the battery is estimated by a second estimation algorithm based on the integration of the battery current.

[0085] Among them, the platform working range is the specific operating range of the battery's state of charge (SOC). Within this range, the battery's open-circuit voltage (OCV) has low sensitivity to changes in SOC (or the local slope is less than a preset threshold), making it difficult to provide effective observation information for the filtering algorithm through voltage signals. This is suitable for current integral estimation algorithms.

[0086] The second estimation algorithm refers to the state of charge estimation algorithm based on the battery current integral. Its core feature is that it derives the state of charge by quantifying the change in charge, without relying on the observation sensitivity of the voltage signal, and is adapted to the estimation requirements of the platform's working range.

[0087] Integrating battery current: This refers to the core logic of calculating the change in battery charge or discharge by integrating the charging and discharging current over time during battery operation, thereby deriving the state of charge.

[0088] In this embodiment, the judgment result of the battery being in the platform working range output by S201 is first received. Then, the second estimation algorithm is started. The current charging and discharging current data of the battery is collected in real time by the sensor. The current data is integrated over time to calculate the cumulative change in charging and discharging capacity. Then, combined with the preset initial state of charge value, the estimated value of the current state of charge of the battery is derived. Finally, a stable and reliable state of charge estimation result within the platform working range is output.

[0089] In an optional embodiment, S201 determines the current operating range of the battery based on the current state parameters as follows:

[0090] Method 1: Compare the current estimated state of charge with the preset upper limit threshold and lower limit threshold of the state of charge; if the current estimated state of charge is less than the lower limit threshold of the state of charge, or greater than the upper limit threshold of the state of charge, then determine that the battery is currently in the non-platform working range; if the current estimated state of charge is less than the upper limit threshold of the state of charge and greater than the lower limit threshold of the state of charge, then determine that the battery is currently in the platform working range.

[0091] Among them, the upper limit threshold of state of charge refers to the preset upper limit value standard of state of charge (SOC) used to distinguish the working range. It is pre-calibrated based on the battery's own state of charge-open circuit voltage (SOC-OCV) characteristics and is the quantitative basis for judging whether the battery exceeds the upper limit of the platform's working range.

[0092] State of Charge (SOC) lower limit threshold: refers to the preset lower limit value standard of the state of charge used to distinguish the working range. It is used in conjunction with the upper limit threshold of the state of charge. It is also pre-calibrated based on the battery SOC-OCV characteristics and is the quantitative basis for judging whether the battery is below the lower limit of the platform's working range.

[0093] In this embodiment of the application, when executing method one, the current state of charge (SOC) estimate of the battery is first obtained, and then the estimate is compared and analyzed with the preset upper limit threshold and lower limit threshold of the state of charge. If the current SOC estimate is less than the lower limit threshold or greater than the upper limit threshold, the battery is determined to be in a non-platform operating range. If the current SOC estimate is between the lower limit threshold and the upper limit threshold, the battery is determined to be in a platform operating range. Finally, the judgment result is output, providing a clear basis for the subsequent selection of the first estimation algorithm or the second estimation algorithm.

[0094] Method 2: Obtain the local slope of the state-of-charge-open-circuit voltage curve corresponding to the current state-of-charge estimate; compare the local slope with a preset slope threshold; if the local slope is greater than the preset slope threshold, determine that the battery is currently in the non-platform operating range; if the local slope is less than the preset slope threshold, determine that the battery is currently in the platform operating range.

[0095] The local slope of the state-of-charge (SOC)-open-circuit voltage (OCV) curve refers to the slope of the tangent line corresponding to the current estimated SOC value on the preset SOC-OCV curve. It directly reflects the sensitivity of the OCV to changes in SOC, serving as a quantitative indicator for distinguishing working intervals. The magnitude of the local slope directly reflects the sensitivity of the OCV to changes in SOC; a larger slope indicates stronger sensitivity and provides more effective observation information for the filtering algorithm (adapting to the first estimation algorithm); a smaller slope indicates weaker sensitivity and makes it difficult to infer the SOC from the voltage signal (adapting to the second estimation algorithm).

[0096] Preset slope threshold: refers to a quantitative standard pre-calibrated based on offline battery test data, used to judge the local slope sensitivity of the SOC-OCV curve, and is a key reference value for dividing the platform working range and the non-platform working range.

[0097] In this embodiment of the application, when executing method two, the current state of charge (SOC) estimate of the battery is first obtained, and then the preset SOC-OCV curve is queried to obtain the local slope of the curve corresponding to the SOC estimate. Subsequently, the local slope is compared and analyzed with a preset slope threshold: if the local slope is greater than the preset slope threshold, it is determined that the battery is currently in a non-platform working range; if the local slope is less than the preset slope threshold, it is determined that the battery is currently in a platform working range. Finally, the judgment result is output to provide a basis for the selection of subsequent estimation algorithms.

[0098] In an optional embodiment, the construction process of the lightweight battery model includes:

[0099] Step a1: Test the battery within the non-platform working range to collect test data; wherein, the test data includes measured voltage, measured current and temperature.

[0100] Among them, test data refers to the core data set collected synchronously during targeted testing of the battery outside the platform's working range, which is used for the identification of lightweight battery model parameters and is the basic input for building the model.

[0101] In this embodiment, the battery is driven to operate stably within the specified range according to a pre-defined standardized test procedure (such as constant current charge-discharge, pulse charge-discharge, etc.). Simultaneously, voltage, current, and temperature sensors are used to synchronously collect measured voltage, current, and temperature data during battery operation, ensuring the real-time and synchronous nature of data acquisition. Finally, the collected data undergoes preliminary preprocessing (such as removing outliers and aligning the data time axis) to obtain a set of test data that meets the requirements for subsequent model parameter identification, providing basic data support for the construction of a lightweight battery model.

[0102] Step a2: Based on the test data and the preset second-order RC equivalent circuit model, output the model predicted voltage.

[0103] Among them, the second-order RC equivalent circuit model refers to the preset circuit model used to simulate the dynamic electrical characteristics of the battery. It consists of an ohmic internal resistance and two independent RC networks (polarization resistor and polarization capacitor connected in series). It can accurately fit the voltage response characteristics of the battery in the non-platform operating range and is the carrier for deriving the model's predicted voltage.

[0104] Model predicted voltage: refers to the simulated battery voltage value calculated by the internal circuit characteristics of the model after inputting the test data into the second-order RC equivalent circuit model. It is used to compare with the measured voltage to optimize the model parameters and is the core intermediate output for model parameter identification.

[0105] In this embodiment, a preset second-order RC equivalent circuit model is first invoked to clarify the circuit structure and parameter correlation of the model; then, the test data (including measured current, temperature, etc.) collected and preprocessed in step a1 is used as the model input. Based on the dynamic characteristics of the circuit model, such as ohmic voltage drop and polarization response, the simulated voltage values ​​corresponding to each test moment are calculated through circuit theory; finally, according to the time axis alignment rule, the model predicted voltage sequence corresponding one-to-one with the measured voltage is output, providing basic data for subsequent construction of error objective function and optimization of model parameters.

[0106] Step a3: Construct an objective function with the goal of minimizing the error between the measured voltage and the model-predicted voltage.

[0107] The objective function refers to the mathematical expression for the quantitative error relationship constructed with the core objective of minimizing the deviation between the measured voltage and the model-predicted voltage. It serves as the basis for subsequent optimization algorithms to solve for the optimal parameters of the model.

[0108] In this embodiment, the measured voltage sequence collected in step a1 and the model predicted voltage sequence output in step a2 are first obtained to ensure that the two sets of data are perfectly aligned along the time axis. Then, based on the goal of minimizing error, an appropriate error quantization method (such as mean square error, absolute error, etc.) is selected to construct a mathematical function that reflects the degree of deviation between the measured voltage and the model predicted voltage. This function takes the parameters to be identified in the second-order RC equivalent circuit model as independent variables and the error value as dependent variables, and finally forms an objective function that can be used for iterative optimization, providing a clear optimization direction for solving the optimal parameters of the model through the nonlinear least squares method.

[0109] Step a4: The objective function is iteratively optimized using the nonlinear least squares method to identify the parameters of the second-order RC equivalent circuit model in the non-platform operating range, thereby constructing the lightweight battery model.

[0110] Among them, the nonlinear least squares method refers to a mathematical optimization algorithm used to solve the optimal solution of a nonlinear function. It iteratively adjusts the parameters to be optimized by minimizing the sum of squares of the objective function error, and finally finds the parameter combination that minimizes the error.

[0111] The parameters of the second-order RC equivalent circuit model in the non-platform operating range include: ohmic internal resistance R0, first polarization resistor R1, first polarization capacitor C1, second polarization resistor R2, and second polarization capacitor C2. The first polarization resistor R1 and the first polarization capacitor C1 form a first RC network, and the second polarization resistor R2 and the second polarization capacitor C2 form a second RC network.

[0112] Iterative optimization: refers to the cyclic optimization process based on the nonlinear least squares method, that is, repeatedly adjusting the parameters to be identified, calculating the error of the objective function, and correcting the direction of the parameters until the error meets the preset convergence condition.

[0113] In this embodiment, the nonlinear least squares method is first invoked, and the error objective function constructed in step a3 is used as the optimization object. At the same time, the initial values ​​of the parameters to be identified (R0, R1, C1, R2, C2) of the second-order RC equivalent circuit model are initialized. Then, according to the algorithm iteration rules, the measured voltage of step a1 and the model predicted voltage of step a2 are substituted into the objective function to calculate the error value corresponding to the current parameter combination. Then, the parameter value direction is adjusted according to the error feedback, the parameters to be identified are updated, and the model predicted voltage and error are recalculated. The above iteration process is repeated until the error reaches the preset convergence threshold (or the number of iterations meets the upper limit), the optimization is stopped, and the parameter combination at this time is locked. Finally, the optimized parameters are embedded into the second-order RC equivalent circuit model to form a lightweight battery model that is only suitable for non-platform operating ranges, and the model construction is completed.

[0114] In an optional embodiment, the construction process of the extended Kalman filter algorithm includes:

[0115] Step b1: The state of charge, electrochemical polarization voltage, and concentration polarization voltage to be estimated are used as state vectors; wherein, the electrochemical polarization voltage is obtained based on the first polarization resistor R1 and the first polarization capacitor C1; and the concentration polarization voltage is obtained based on the second polarization resistor R2 and the second polarization capacitor C2.

[0116] Among them, the state vector refers to the set of vectors used in the extended Kalman filter algorithm to uniformly describe the core operating state of the battery. It consists of three key state variables to be estimated: the state of charge (SOC), the electrochemical polarization voltage, and the concentration polarization voltage. It is the object of the algorithm to achieve state prediction and update.

[0117] Electrochemical polarization voltage: refers to the polarization voltage component generated during battery operation due to the mismatch between the internal electrochemical reaction rate and the charge transfer rate. Its dynamic characteristics are determined by the charging and discharging behavior of the first RC network (first polarization resistor R1 and first polarization capacitor C1), and it is a key state quantity reflecting the battery's electrochemical dynamic response.

[0118] Concentration polarization voltage: refers to the polarization voltage component generated during battery operation due to the uneven distribution of ion concentration on the electrode surface and inside the electrolyte. Its dynamic characteristics are determined by the charging and discharging behavior of the second RC network (second polarization resistor R2 and second polarization capacitor C2), and it is a key state quantity reflecting the dynamic response of battery concentration.

[0119] In this embodiment, the core structure of the second-order RC equivalent circuit model is first defined: the ohmic internal resistance R0, the first RC network (R1 and C1), the second RC network (R2 and C2), and the polarization characteristics of each network. Then, based on the model structure, the core quantities to be estimated in the extended Kalman filter algorithm are determined: the state of charge (SOC), which can directly characterize the remaining battery capacity; the electrochemical polarization voltage associated with the first RC network; and the concentration polarization voltage associated with the second RC network. Finally, these three key quantities to be estimated are integrated into a unified state vector, which is used to construct state equations to describe the state update law and construct observation equations to establish the mapping relationship between the state and the measured voltage.

[0120] Step b2: Based on the dynamic characteristics of the lightweight battery model, construct the state equation f(·) to describe the update relationship of the state vector with the measured current and sampling time.

[0121] Among them, the state equation f(·) refers to the function used in the extended Kalman filter algorithm to describe the change of the state vector with the input (measured current) and time (sampling time). It is the mathematical carrier of the algorithm's state prediction step. Its input is the current state vector, the measured current and the sampling time, and its output is the predicted state vector at the next moment.

[0122] The dynamic characteristics of lightweight battery models refer to the response of the battery's electrical state (such as polarization voltage and state of charge) as a function of charging and discharging current and time, as represented by the lightweight battery model. This is the physical basis for constructing the state equation.

[0123] Sampling time: refers to the time interval between the sensor collecting measured data of the battery (such as current and voltage), that is, the time difference between two adjacent data acquisitions, which is the time step of the state vector update.

[0124] In this embodiment, the dynamic response law of the lightweight battery model (second-order RC equivalent circuit model) is first invoked: for example, the electrochemical polarization voltage (corresponding to the first RC network) changes with the measured current and sampling time, following the transient response formula of the RC circuit; the concentration polarization voltage (corresponding to the second RC network) is similar; and the state of charge follows the law of charge change of the current integral. Subsequently, these laws are transformed into the update relationship of the state vector (state of charge, electrochemical polarization voltage, concentration polarization voltage), with the measured current as the input and the sampling time as the time step, and the next time step value of the state vector is expressed as a function of the current time step value, the measured current and the sampling time. Finally, the state equation f(·) describing the update relationship is constructed, providing mathematical model support for the state prediction step of the extended Kalman filter.

[0125] Step b3: Construct observation equations based on the lightweight battery model to establish the relationship between the state vector and the measured terminal voltage z. k The mapping relationship.

[0126] Among them, the observation equation is the equation used in the extended Kalman filter algorithm to establish the mathematical relationship between the state vector and the measured physical quantity (measured terminal voltage in this step). It is the basis for the state update step of the algorithm. The predicted value of the measured physical quantity can be derived from the state vector, and then compared with the actual measured value to achieve state correction.

[0127] Measured terminal voltage z k : refers to the voltage data at both ends of the battery actually collected by the voltage sensor at time k. It is one of the inputs to the observation equation and is used to compare with the voltage prediction value derived from the state vector to quantify the degree of error in the state estimation.

[0128] In this embodiment, the voltage composition logic of the lightweight battery model is first invoked: the battery terminal voltage is composed of the open-circuit voltage (directly related to the state of charge in the state vector), the ohmic voltage drop (the product of the measured current and the ohmic internal resistance), the electrochemical polarization voltage (one of the state vectors), and the concentration polarization voltage (one of the state vectors). Subsequently, this voltage composition relationship is transformed into a mathematical expression, using the state vectors (state of charge, electrochemical polarization voltage, concentration polarization voltage) as input variables and the measured terminal voltage z as the input variable. k For the output variable, a mapping relationship between the two is constructed, namely the observation equation. This observation equation can derive the predicted value of the terminal voltage from the state vector, providing support for the subsequent correction of the state vector using the measured terminal voltage in the extended Kalman filter.

[0129] In an optional embodiment, S101 uses a first estimation algorithm to estimate the state of charge (SOC) of the battery. Estimating the SOC using the first estimation algorithm includes:

[0130] Step c1: Using the state equation f(·), the state vector updated from the previous time step is... Given the input current at time k, calculate the predicted state vector at time k.

[0131] Wherein, the state vector updated in the previous time step The state vector at time k-1 is corrected by the extended Kalman filter state update step. Its value has been corrected for error by combining the measured terminal voltage at that time, and it is the accurate basic input for the current state prediction.

[0132] Input current at time k: refers to the battery charging and discharging current data collected by the current sensor at time k. It is the external input quantity that drives the battery state change and determines the update trend of the state vector.

[0133] The predicted state vector at time k The estimated value of the state vector at time k, derived through the state equation, is the output of the extended Kalman filter state prediction stage and has not yet been combined with the measured terminal voltage at the current time for error correction.

[0134] In this embodiment, the state vector corrected by the state update process at time k-1 is first called. Simultaneously, the battery charging and discharging input current at the current time k is collected; then, these two data points are substituted into the constructed state equation f(·), and combined with the time step of the sampling time, the dynamic update law of the battery state described by the state equation is used to calculate the estimated value of the state vector at time k, i.e., the predicted state vector. This step completes the state prediction stage of the extended Kalman filter, providing a basic estimate for subsequent state correction based on measured terminal voltage.

[0135] Step c2, based on the observation equation h(·), the predicted state vector... Calculate the predicted observations

[0136] Among them, the observation equation h(·) is a mathematical function used to establish the mapping relationship between the state vector and the observed physical quantity (the battery terminal voltage in this step). It is the core tool for deriving the observation estimate from the predicted state vector. Its input is the state vector and its output is the corresponding observation estimate.

[0137] Predicted state vector The estimated value of the current state vector obtained in step c1 through the state equation is the input variable of the observation equation and has not yet been corrected by combining it with the measured data.

[0138] Predicted observations The estimated battery terminal voltage is derived from the predicted state vector through the observation equation. It is a key reference quantity in the state update stage of the extended Kalman filter and is used to compare with the measured terminal voltage to quantify the state estimation error.

[0139] In this embodiment of the application, the predicted state vector at the current time output in step c1 is first called. Then, substituting this into the constructed observation equation h(·), and using the mapping relationship between the state vector described by the observation equation and the terminal voltage, the predicted value of the battery terminal voltage at the current moment is calculated, i.e., the predicted observation value. This step completes the prediction of the observed values ​​in the extended Kalman filter, providing a benchmark for subsequent correction of the state vector by the deviation between the measured terminal voltage and the predicted observed values.

[0140] Step c3, calculate the Kalman gain K k .

[0141] Wherein, Kalman gain K k : is the weight parameter used in the extended Kalman filter algorithm to balance the state prediction error and the observation error. Its value determines the degree of influence of the deviation between the predicted observation and the measured observation on the state vector correction. It is a key coefficient that connects the state prediction and state update links and achieves accurate state correction.

[0142] In this embodiment, the covariance matrix corresponding to the predicted state vector at the current time (derived from the Jacobian matrix of the state equation and the state covariance matrix of the previous time), the Jacobian matrix of the observation equation at the predicted state vector, and the preset observation noise covariance matrix are first obtained. Then, these matrices are substituted into the standard formula for calculating Kalman gain, and the Kalman gain K at the current time is obtained through matrix multiplication, inversion, and other operations. k This provides a weighting basis for subsequent correction and prediction of the state vector using measured terminal voltage.

[0143] Step c4, based on the predicted state vector The Kalman gain K k The measured terminal voltage z k and the predicted observations Obtain the updated state vector at time k. Among them, in the updated state vector In this study, the component of the state of charge is used as the estimated value of the battery's state of charge.

[0144] Among them, the measured terminal voltage z k : refers to the voltage data at both ends of the battery actually collected by the voltage sensor at the current time k. It is the actual observation data used to correct the predicted state vector and is one of the inputs to the state update process.

[0145] Updated state vector The current state vector after Kalman gain weighting correction is a precise state quantity that combines the estimated value of state prediction with the actual data of measured observation. It is the final output of the extended Kalman filter state update stage.

[0146] State of charge estimate: refers to the state of charge component extracted from the updated state vector. It is the final output of the first estimation algorithm and directly represents the current remaining charge state of the battery.

[0147] In this embodiment of the application, when performing this step, the predicted state vector of step c1 is first retrieved. Kalman gain K in step c3 k The measured terminal voltage z at the current moment k and the predicted observations in step c2 Then the deviation between the measured terminal voltage and the predicted observed value was calculated (i.e. The deviation is multiplied by the Kalman gain to obtain the correction amount for the state vector; then the predicted state vector is added to the correction amount to obtain the updated state vector at the current time step. Finally, from the updated state vector In the process, the component corresponding to the state of charge is extracted, which is the estimated value of the current state of charge of the battery, thus completing the state of charge estimation process of the first estimation algorithm.

[0148] For ease of understanding, combined with Figure 3 The flowchart shown is for implementing state of charge estimation using the Extended Kalman Filter (EKF). The complete process for implementing state of charge estimation using the EKF is described below.

[0149] Step d1, initialization, includes:

[0150] State vector initialization: The initial state is yes The initial state reference for extended Kalman filter startup. SOC0 is the initial value of the state of charge (SOC), obtained by back-calculating the OCV-SOC curve after the battery has been at rest. V 10 This is the initial value of the electrochemical polarization voltage. V 20 This is the initial value of the concentration polarization voltage. Since the polarization dissipates after the battery is left to stand, the initial value is set to 0.

[0151] Error covariance matrix initialization: P0 is the initial value of the covariance matrix, which describes the degree of uncertainty in the initial state estimation and is determined by the confidence level of the initial state (such as the accuracy of OCV-SOC back-calculation). It is the covariance matrix at time k.

[0152] Initialization of process / observation noise covariance matrix: Q k =Q0, R k =R0. Q k R is the process noise covariance, where Q0 is the initial value of the process noise covariance, describing the statistical characteristics of the error in state equation modeling. k R0 is the observation noise covariance, which is the initial value of the observation noise covariance and describes the statistical characteristics of the sensor (voltage / current) measurement error.

[0153] Step d2: Using the state equation f(·), update the state from the previous time step. Given the input current at time k, calculate the predicted state vector at time k. in, u k The input current at the current moment k, i.e. the battery charging / discharging current, is acquired in real time by the current sensor and is the external excitation that drives the changes in battery state.

[0154] In this embodiment, the corrected updated state vector from the previous time step k-1 is first retrieved. The battery charging / discharging input current at the current time k is simultaneously acquired via a current sensor. These two data points are then substituted into the state equation f(·) corresponding to the lightweight second-order RC equivalent circuit. The calculation is performed using the current integral update law of the state of charge, the transient response formulas of the first RC network (electrochemical polarization) and the second RC network (concentration polarization), and a preset sampling time step. Finally, the uncorrected predicted state vector at the current time k is obtained. correspond Figure 3 The mid-state observer drives the battery model to generate estimated values ​​for the link.

[0155] Step d3: Calculate the Jacobian matrix F of the state equation. k And through the Jacobian matrix F k Update the prediction covariance matrix using the covariance matrix at the current time k. in, It is the Jacobian matrix F k The transpose of .

[0156] Among them, the Jacobian matrix F of the state equation k This is a matrix that locally linearizes the nonlinear battery state equation (based on a second-order RC equivalent circuit model) at the current predicted state vector, used to approximate the local rate of change of the state equation.

[0157] The covariance matrix at current time k: refers to the corrected covariance matrix at the previous time k-1. It is a statistical matrix that describes the degree of uncertainty of the estimated state vector at the previous moment.

[0158] Predicting covariance matrix The uncertainty matrix corresponding to the predicted state vector at the current moment reflects the confidence level of the predicted state vector and is one of the inputs for subsequent calculation of the Kalman gain.

[0159] In this embodiment of the application, the predicted state vector at the current time is first used as the basis. By taking the partial derivatives of each state component (state of charge, electrochemical polarization voltage, concentration polarization voltage) in the state equation f(·) (corresponding to the state update relation of the second-order RC cell model), the Jacobian matrix F of the state equation is constructed. k This completes the local linearization of the nonlinear state equation; then, the updated covariance matrix from the previous time step is retrieved. Through matrix operations ( Q k (For process noise covariance), update to obtain the prediction covariance matrix at the current time. This provides a basis for uncertainty quantification in the subsequent calculation of Kalman gain.

[0160] Step d4, based on the observation equation h(·), the predicted state vector is... Calculate the predicted observations in,

[0161] Among them, the observation equation h(·) is a mathematical equation that establishes the mapping relationship between the state vector and the battery terminal voltage based on the voltage composition logic of the second-order RC equivalent circuit model. It describes the superposition relationship between open-circuit voltage, ohmic voltage drop, electrochemical polarization voltage and concentration polarization voltage. It is a tool in the extended Kalman filter (EKF) for deriving the observation prediction value from the state vector. It is a nonlinear observation equation.

[0162] In this embodiment of the application, the predicted state vector output in step d2 is retrieved first. Then, the observation equation h(·) based on the second-order RC equivalent circuit model is invoked; subsequently, the charged state components in the predicted state vector are mapped to the corresponding open-circuit voltages, and combined with the electrochemical polarization voltage, concentration polarization voltage, and the product of the current input current and the ohmic internal resistance (ohmic voltage drop) within the vector, they are substituted into the voltage superposition relationship of the observation equation for calculation, and finally the predicted observation value at the current moment is obtained.

[0163] Step d5: Calculate the Jacobian matrix H of the observation equation. k And through the Jacobian matrix H kThe residual covariance S is obtained by combining the observation noise covariance at the current time k. k .

[0164] Among them, the Jacobian matrix H of the observation equation k This is a matrix that locally linearizes the nonlinear observation equation at the current predicted state vector. It is used to approximate the local change relationship between the state vector and the observed value (terminal voltage) in the observation equation, and is a tool for extending the Kalman filter to adapt to the nonlinear characteristics of the observation loop.

[0165] Residual covariance S k The uncertainty statistical matrix describes the deviation between predicted and measured observations and is one of the parameters used to weigh the reliability of observational data.

[0166] In this embodiment of the application, the predicted state vector at the current time is first used as the basis. By taking partial derivatives of each state component (charge state, electrochemical polarization voltage, concentration polarization voltage) in the observation equation (corresponding to the voltage superposition relationship of the second-order RC model), the Jacobian matrix H of the observation equation is constructed. k This completes the local linearization of the observation equation; subsequently, the prediction covariance matrix obtained in step d3 is retrieved. and the observation noise covariance R at the current moment k Through matrix operations The residual covariance S was calculated. k This provides a foundation for the subsequent calculation of Kalman gain.

[0167] Step d6, based on the prediction covariance matrix at the current time k The Jacobian matrix H at time k k The residual covariance S at the current time k k The Kalman gain K is calculated. k .

[0168] Wherein, Kalman gain K k It combines the uncertainty of state prediction (prediction covariance matrix) and the linearization relationship of the observation equation (Jacobi matrix H). k The weighting coefficients, calculated from the uncertainty of observation bias (residual covariance), serve to balance the reliability of state prediction with the reliability of actual observation, providing quantified weights for subsequent correction of the predicted state vector using observation bias.

[0169] In this embodiment of the application, the prediction covariance matrix at the current time output in step d3 is first retrieved. The Jacobian matrix H of the observation equation obtained in step d5 k With residual covariance S kThese matrices are then substituted into the standard formula for calculating Kalman gain. Perform matrix multiplication and inverse matrix operations sequentially to obtain the Kalman gain K at the current time step. k This gain will serve as an adjustment weight between the predicted state vector and the observation bias, ensuring that the subsequently corrected state both closely reflects the dynamic changes of the battery and matches the actual state of the measured data.

[0170] Step d7, the updated state vector at the current time. From the updated state vector In the process, the component corresponding to the state of charge is extracted, which is the estimated value of the current state of charge of the battery, thus completing the state of charge estimation process of the first estimation algorithm.

[0171] Wherein, the updated state vector It is the current state vector after Kalman gain weighting correction, which integrates the dynamic trend of state prediction and the actual data of actual observation. It is the final output of the extended Kalman filter state update stage, containing accurate state of charge, electrochemical polarization voltage, and concentration polarization voltage components.

[0172] In this embodiment of the application, the predicted state vector of step d2 is retrieved first. Kalman gain K in step d6 k The measured terminal voltage z at the current moment k Compared with the predicted observations in step d4 The deviation between the measured terminal voltage and the predicted observed value was then calculated. This deviation is compared with the Kalman gain K. k Multiplying them yields the state correction, which is then multiplied by the predicted state vector. Add, using the formula Obtain the updated state vector Finally from Extract the predefined state of charge component, which is the estimated state of charge of the current battery, and complete the closed-loop process of the first estimation algorithm (extended Kalman filter) for state of charge estimation.

[0173] Optionally, the covariance matrix can be updated.

[0174] This step is the covariance iterative update stage of the first estimation algorithm (Kalman filter algorithm), which realizes adaptive adjustment of error and noise parameters to adapt to dynamic operating conditions in non-platform operating ranges. Specifically, it includes:

[0175] 1) Error covariance matrix update

[0176] After updating the state vector corresponding to the State of Charge (SOC), the error covariance matrix is ​​updated synchronously, with the expression: Let I be the error covariance matrix after correction at time k. I is the identity matrix.

[0177] 2) Dynamic updates of process noise covariance and observation noise covariance

[0178] In this embodiment, the process noise covariance Q of the first estimation algorithm k Covariance of observation noise R k Instead of taking a fixed value, it is dynamically adjusted based on the current operating conditions and sensor data. The update expression is:

[0179]

[0180] In the formula, Q k+1 yes k+1 Process noise covariance at any given time;

[0181] R k+1 It is the observation noise covariance at time k+1;

[0182] u k It is the charging and discharging current at the current moment k;

[0183] V k It is the battery terminal voltage at the current moment k;

[0184] T k It is the cell temperature at the current moment k;

[0185] Y k It is the observation residual term at the current time k, that is, the deviation between the measured terminal voltage and the model predicted terminal voltage;

[0186] ΔX m This is the m-th type of operating condition variation item, including: the change in charging and discharging current (i.e., the charging and discharging rate), the change in cell temperature (i.e., temperature fluctuation), etc.

[0187] α m This is the process weighting coefficient corresponding to the change term of the m-th type of operating condition, where the change term of the m-th type of operating condition (such as current change) is a dynamic parameter of the operating condition, α. m Its purpose is to quantify the degree of impact of this type of operating condition change on the process noise covariance update results;

[0188] μ m It is an auxiliary weighting coefficient that matches the variation term of the m-th type of working condition, and needs to be matched with α. m Synergistic effect: By associating the actual fluctuation amplitude of the m-th type of operating condition change item, the adjustment accuracy of the process noise covariance by the change of operating conditions is further refined, and the dynamic operating condition characteristics of the non-platform working range are adapted.

[0189] M is the total number of categories for the variable item of the m-th working condition;

[0190] γ is the residual weight corresponding to the observation residual term. The observation residual term is the deviation between the measured terminal voltage and the model predicted terminal voltage mentioned above. The role of γ is to adjust the influence of this residual term on the process noise covariance update and enhance the algorithm's adaptive response capability to model prediction errors.

[0191] v is the residual covariance matrix corresponding to the observation residual term, which is used to quantify the degree of uncertainty of the deviation between the measured terminal voltage and the model predicted terminal voltage;

[0192] β m It is the weighting coefficient corresponding to the change in operating conditions in the observation noise covariance update;

[0193] η m It is the covariance matrix corresponding to the change in operating conditions in the observation noise covariance update;

[0194] g1(·) is the update equation for the process noise covariance, which is related to the current operating conditions (charge and discharge rate, temperature fluctuation) and sensor data (terminal voltage, current, cell temperature).

[0195] g2(·) is the update equation for the observation noise covariance, which is related to the current operating conditions (charge / discharge rate, temperature fluctuation) and sensor data (terminal voltage, current, cell temperature).

[0196] In this embodiment, by dynamically updating the process noise covariance and observation noise covariance of the first estimation algorithm (Kalman filter algorithm), the algorithm can respond in real time to dynamic operating conditions such as charge / discharge rate fluctuations and cell temperature fluctuations within the non-platform operating range, without relying on cumbersome parameter identification for different operating conditions, thus improving operating condition tracking. The synergistic effect of the observation residual term and multiple operating condition change terms can accurately quantify the uncertainty of model prediction errors and operating condition changes, combined with α m β m Fine-tuning of the equal weighting coefficients effectively suppresses noise interference, ensuring high-precision estimation of the state of charge (SOC) within the non-platform operating range. In addition, the adaptive adjustment mechanism for process noise and observation noise can dynamically offset the effects of uncertainties such as sensor measurement errors and model parameter drift, improving the robustness and anti-interference capability of the first estimation algorithm, enabling it to stably output SOC estimation results under complex energy storage conditions.

[0197] In an optional embodiment, the method further includes: adjusting the initial value of the second estimation algorithm based on the state of charge estimate obtained by the first estimation algorithm within the non-platform operating range.

[0198] The initial value of the second estimation algorithm refers to the reference state of charge (SOC) value when the second estimation algorithm (ampere-hour integration method) starts or enters a new round of integration calculation. It is the starting point for calculating the subsequent state of charge (SOC) using the current integration formula. The accuracy of the initial value directly determines the upper limit of the estimation error within the platform's working range.

[0199] The specific process for adjusting the initial value of the second estimation algorithm based on the state of charge estimate obtained by the first estimation algorithm within the non-platform operating range includes:

[0200] Step e1: Monitor the battery's operating range status and parameters in real time. If the triggering conditions are met (such as switching from non-platform area to platform area, error exceeding limits, or cycle reaching the target), immediately start the initial value adjustment process. If the battery remains in the platform area, temporarily store the triggering state and perform the adjustment after obtaining the latest benchmark calibration value the next time it enters the non-platform area.

[0201] Step e2: The latest State of Charge (SOC) estimation result of the first estimation algorithm in the non-platform operating range is used as the benchmark calibration value. This value has been suppressed by the model prediction-measured voltage correction closed loop to suppress noise and model error. Moreover, the open circuit voltage (OCV) in the non-platform region is sensitive to changes in the State of Charge (SOC), ensuring the high reliability of the calibration value.

[0202] Among them, the benchmark calibration value is the estimated state of charge (SOC) output by the first estimation algorithm (a filtering algorithm based on a lightweight model) in the non-platform operating range.

[0203] Step e3: The benchmark calibration value is directly assigned as the current initial value of the second estimation algorithm, replacing the original initial value to start a new round of integration calculation. The auxiliary parameters of the ampere-hour integration method (such as battery rated capacity and coulomb efficiency coefficient) are corrected simultaneously to ensure the consistency between the initial value and the core parameters of the algorithm and to avoid error accumulation.

[0204] Step e4: Compare the deviation of the State of Charge (SOC) estimated value of the second estimation algorithm before and after adjustment with the benchmark calibration value of the first estimation algorithm. If the deviation is ≤ a preset threshold, such as 0.5%, the adjustment is deemed effective and the current initial value is maintained. If the deviation is > 0.5%, the latest result of the first estimation algorithm is retrieved again and adjusted until the accuracy requirements are met.

[0205] Step e5: After the adjustment is completed, the second estimation algorithm continues to perform current integration calculation in the plateau region based on the updated initial value; the switching status of the working area is monitored in real time. If the battery returns to the non-plateau region, the initial value fixing logic of the second estimation algorithm is paused. When switching back to the plateau region, the above process is repeated to update the initial value, forming a closed loop of non-plateau calibration-platform accurate estimation.

[0206] Figure 4This is a schematic flowchart of a method for estimating the state of charge (SOC) of an energy storage system according to another embodiment of this application. As shown in the figure, the lightweight battery model parameters specific to the non-platform operating range are first obtained through offline identification and stored in the flash memory of the energy storage system. Then, the operating condition of the energy storage battery is judged in real time. If it is in the non-platform operating range, the first estimation algorithm (extended Kalman filter) is started, and the process noise covariance and observation noise covariance are dynamically updated. The initial value of the second estimation algorithm (ampere-hour integral) is updated through the accurate result output by the algorithm. If it is in the platform operating range, the second estimation algorithm with the updated initial value is directly called to estimate the SOC. Finally, based on the algorithm output results of different operating ranges, the final estimated SOC value is obtained and output, thus fully realizing the interval-based SOC estimation process of parameter storage, operating condition judgment, dual algorithm collaboration, and result output.

[0207] In the embodiments of this application, see Figure 5 The figure shows a comparison of the accuracy of the interval-based State of Charge (SOC) estimation method with the single ampere-hour integral method. By comparing the curves of the ampere-hour integral + extended Kalman filter and the single ampere-hour integral method, the figure clearly presents the SOC estimation performance in different operating intervals: in the plateau region (the region with a relatively high SOC), the SOC curves of the two methods have a high degree of fit; however, after entering the non-plateau region, the SOC curve of the single ampere-hour integral method deviates significantly. The ampere-hour integral + extended Kalman filter scheme in this embodiment, relying on the accurate correction of the extended Kalman filter (the first estimation algorithm), maintains the stability of the SOC estimation, intuitively demonstrating the accuracy advantage of this embodiment in the non-plateau region.

[0208] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0209] Corresponding to the energy storage system state of charge estimation method described in the above embodiments, Figure 6 A structural block diagram of the energy storage system state of charge estimation device provided in the embodiments of this application is shown. For ease of explanation, only the parts related to the embodiments of this application are shown.

[0210] Reference Figure 6 The energy storage system's state of charge estimation device includes:

[0211] The first estimation module is used to estimate the state of charge of the battery using a first estimation algorithm; wherein the first estimation algorithm is a filtering algorithm based on a lightweight battery model, and the lightweight battery model is identified based on test data collected from the non-platform operating range of the battery;

[0212] An adaptive adjustment module is used to adaptively adjust the noise covariance parameter in the first estimation algorithm based on the real-time operating parameters of the battery.

[0213] In one possible implementation, the energy storage system state of charge estimation device further includes an acquisition module and a second estimation module;

[0214] The acquisition module is used to acquire the current state parameters of the battery and determine the current working range of the battery based on the current state parameters.

[0215] The first estimation module is used to perform the step of estimating the state of charge of the battery using the first estimation algorithm when the battery is in a non-platform operating range.

[0216] The second estimation module is used to estimate the state of charge of the battery by employing a second estimation algorithm based on the integration of the battery current when the battery is in the platform operating range.

[0217] In one possible implementation, the acquisition module is used for:

[0218] Compare the current state of charge estimate with the preset upper limit threshold and lower limit threshold of the state of charge.

[0219] If the current state of charge estimate is less than the lower limit threshold of the state of charge, or greater than the upper limit threshold of the state of charge, then the battery is determined to be currently in the non-platform operating range.

[0220] If the current state of charge estimate is less than the upper limit threshold of the state of charge and greater than the lower limit threshold of the state of charge, then the battery is determined to be currently in the platform's operating range.

[0221] or,

[0222] Obtain the local slope of the state-of-charge-open-circuit voltage curve corresponding to the current state-of-charge estimate;

[0223] The local slope is compared with a preset slope threshold;

[0224] If the local slope is greater than the preset slope threshold, then the battery is determined to be currently in the non-platform working range;

[0225] If the local slope is less than the preset slope threshold, then the battery is determined to be currently in the platform's working range.

[0226] In one possible implementation, the energy storage system state of charge estimation device further includes a building module for:

[0227] The construction process of the lightweight battery model includes:

[0228] The battery is tested within the non-platform operating range to collect test data; wherein, the test data includes measured voltage, measured current, and temperature;

[0229] Based on the test data and the preset second-order RC equivalent circuit model, the output model predicts the voltage.

[0230] An objective function is constructed with the goal of minimizing the error between the measured voltage and the model-predicted voltage.

[0231] The objective function is iteratively optimized using the nonlinear least squares method to identify the parameters of the second-order RC equivalent circuit model within the non-platform operating range, thereby constructing the lightweight battery model.

[0232] In one possible implementation, the filtering algorithm is an extended Kalman filter algorithm; the parameters of the second-order RC equivalent circuit model in the non-platform operating range include: ohmic internal resistance R0, first polarization resistor R1, first polarization capacitor C1, second polarization resistor R2, and second polarization capacitor C2, and the first polarization resistor R1 and the first polarization capacitor C1 form a first RC network, and the second polarization resistor R2 and the second polarization capacitor C2 form a second RC network;

[0233] The construction process of the extended Kalman filter algorithm includes:

[0234] The state of charge, electrochemical polarization voltage, and concentration polarization voltage to be estimated are used as state vectors; wherein, the electrochemical polarization voltage is obtained based on the first polarization resistor R1 and the first polarization capacitor C1; and the concentration polarization voltage is obtained based on the second polarization resistor R2 and the second polarization capacitor C2.

[0235] Based on the dynamic characteristics of the lightweight battery model, a state equation f(·) is constructed to describe the update relationship of the state vector with the measured current and sampling time.

[0236] Based on the aforementioned lightweight battery model, observation equations are constructed to establish the relationship between the state vector and the measured terminal voltage z. k The mapping relationship.

[0237] In one possible implementation, the first estimation module is used for:

[0238] Using the state equation f(·), the state vector updated from the previous time step is obtained. Given the input current at time k, calculate the predicted state vector at time k.

[0239] Based on the observation equation h(·), the predicted state vector Calculate the predicted observations

[0240] Calculate the Kalman gain K k ;

[0241] Based on the predicted state vector The Kalman gain K k The measured terminal voltage z k and the predicted observations Obtain the updated state vector at time k. Among them, in the updated state vector In this study, the component of the state of charge is used as the estimated value of the battery's state of charge.

[0242] In one possible implementation, the real-time operating parameters include at least one of the following: measured terminal voltage z. k Measured current, cell temperature, ambient temperature, and charge / discharge rate;

[0243] The noise covariance parameter includes at least one of the following: process noise covariance and measurement noise covariance.

[0244] In one possible implementation, the energy storage system state of charge estimation device further includes an adjustment module for:

[0245] Based on the state of charge estimate obtained by the first estimation algorithm within the non-platform operating range, the initial value of the second estimation algorithm is adjusted.

[0246] It should be noted that the information interaction and execution process between the above modules are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, which will not be repeated here.

[0247] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0248] This application also provides a computer device, which includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement the steps in any of the above method embodiments.

[0249] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0250] This application provides a computer program product that, when run on a computer device, enables the computer device to execute the steps described in the various method embodiments above.

[0251] Figure 7 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Figure 7 As shown, the computer device of this embodiment includes: at least one processor 70 ( Figure 7 (Only one is shown in the diagram), memory 71, and computer program 72 stored in the memory 71 and executable on the at least one processor 70, wherein the processor 70 executes the computer program 72 to implement the steps in any of the above embodiments of the energy storage system state of charge estimation method.

[0252] The computer device may include, but is not limited to, a processor 70 and a memory 71. Those skilled in the art will understand that... Figure 7The examples of computer devices are merely examples and do not constitute a limitation on computer devices. They may include more or fewer components than shown in the illustration, or combinations of certain components, or different components. For example, they may also include input / output devices, network access devices, etc.

[0253] The processor 70 may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0254] In some embodiments, the memory 71 may be an internal storage unit of the computer device, such as a hard disk or memory. In other embodiments, the memory 71 may be an external storage device of the computer device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc. Furthermore, the memory 71 may include both internal and external storage units of the computer device. The memory 71 is used to store the operating system, applications, bootloader, data, and other programs, such as the program code of the computer program. The memory 71 can also be used to temporarily store data that has been output or will be output.

[0255] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a device / computer equipment, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0256] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0257] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0258] In the embodiments provided in this application, it should be understood that the disclosed apparatus / computer devices and methods can be implemented in other ways. For example, the apparatus / computer device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings or direct couplings or communication connections may be through some interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0259] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0260] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for estimating the state of charge of an energy storage system, characterized in that, The method includes: The state of charge of the battery is estimated using a first estimation algorithm; wherein the first estimation algorithm is a filtering algorithm based on a lightweight battery model, the lightweight battery model being identified based on test data collected from the non-platform operating range of the battery; Furthermore, the noise covariance parameter in the first estimation algorithm is adaptively adjusted based on the real-time operating parameters of the battery.

2. The method for estimating the state of charge of an energy storage system as described in claim 1, characterized in that, The method further includes: Obtain the current state parameters of the battery, and determine the current working range of the battery based on the current state parameters; When the battery is in a non-platform operating range, the step of estimating the state of charge of the battery using the first estimation algorithm is executed; When the battery is in the platform operating range, the state of charge of the battery is estimated by a second estimation algorithm based on the integration of the battery current.

3. The method for estimating the state of charge of an energy storage system as described in claim 2, characterized in that, Determining the current operating range of the battery based on the current state parameters includes: Compare the current state of charge estimate with the preset upper limit threshold and lower limit threshold of the state of charge. If the current state of charge estimate is less than the lower limit threshold of the state of charge, or greater than the upper limit threshold of the state of charge, then the battery is determined to be currently in the non-platform operating range. If the current state of charge estimate is less than the upper limit threshold of the state of charge and greater than the lower limit threshold of the state of charge, then the battery is determined to be currently in the platform's operating range. or, Obtain the local slope of the state-of-charge-open-circuit voltage curve corresponding to the current state-of-charge estimate; The local slope is compared with a preset slope threshold; If the local slope is greater than the preset slope threshold, then the battery is determined to be currently in the non-platform working range; If the local slope is less than the preset slope threshold, then the battery is determined to be currently in the platform's working range.

4. The method for estimating the state of charge of an energy storage system as described in claim 2, characterized in that, The construction process of the lightweight battery model includes: The battery is tested within the non-platform operating range to collect test data; wherein, the test data includes measured voltage, measured current, and temperature; Based on the test data and the preset second-order RC equivalent circuit model, the output model predicts the voltage. An objective function is constructed with the goal of minimizing the error between the measured voltage and the model-predicted voltage. The objective function is iteratively optimized using the nonlinear least squares method to identify the parameters of the second-order RC equivalent circuit model within the non-platform operating range, thereby constructing the lightweight battery model.

5. The method for estimating the state of charge of an energy storage system as described in claim 4, characterized in that, The filtering algorithm is the extended Kalman filter algorithm; the parameters of the second-order RC equivalent circuit model in the non-platform operating range include: ohmic internal resistance R0, first polarization resistor R1, first polarization capacitor C1, second polarization resistor R2, and second polarization capacitor C2, and the first polarization resistor R1 and the first polarization capacitor C1 form a first RC network, and the second polarization resistor R2 and the second polarization capacitor C2 form a second RC network; The construction process of the extended Kalman filter algorithm includes: The state of charge, electrochemical polarization voltage, and concentration polarization voltage to be estimated are used as state vectors; wherein, the electrochemical polarization voltage is obtained based on the first polarization resistor R1 and the first polarization capacitor C1; and the concentration polarization voltage is obtained based on the second polarization resistor R2 and the second polarization capacitor C2. Based on the dynamic characteristics of the lightweight battery model, a state equation f(·) is constructed to describe the update relationship of the state vector with the measured current and sampling time. Based on the aforementioned lightweight battery model, observation equations are constructed to establish the relationship between the state vector and the measured terminal voltage z. k The mapping relationship.

6. The method for estimating the state of charge of an energy storage system as described in claim 5, characterized in that, The estimation of the battery's state of charge using the first estimation algorithm includes: Using the state equation f(·), the state vector updated from the previous time step is obtained. Given the input current at time k, calculate the predicted state vector at time k. Based on the observation equation h(·), the predicted state vector Calculate the predicted observations Calculate the Kalman gain K k ; Based on the predicted state vector The Kalman gain K k The measured terminal voltage z k and the predicted observations Obtain the updated state vector at time k. Among them, in the updated state vector In this study, the component of the state of charge is used as the estimated value of the battery's state of charge.

7. The method for estimating the state of charge of an energy storage system as described in claim 1, characterized in that, The real-time operating parameters include at least one of the following: measured terminal voltage z k Measured current, cell temperature, ambient temperature, and charge / discharge rate; The noise covariance parameter includes at least one of the following: process noise covariance and measurement noise covariance.

8. The method for estimating the state of charge of an energy storage system as described in claim 2, characterized in that, The method further includes: Based on the state of charge estimate obtained by the first estimation algorithm within the non-platform operating range, the initial value of the second estimation algorithm is adjusted.

9. A state of charge estimation device for an energy storage system, characterized in that, The energy storage system state of charge estimation device includes: The first estimation module is used to estimate the state of charge of the battery using a first estimation algorithm; wherein the first estimation algorithm is a filtering algorithm based on a lightweight battery model, and the lightweight battery model is identified based on test data collected from the non-platform operating range of the battery; An adaptive adjustment module is used to adaptively adjust the noise covariance parameter in the first estimation algorithm based on the real-time operating parameters of the battery.

10. A computer program product, characterized in that, It includes a computer program that, when run, implements the energy storage system state of charge estimation method as described in any one of claims 1 to 8.

Images