Lithium battery charge state prediction method and device and medium

Abstract

The invention discloses a method and device for predicting the state of charge of a lithium battery and a medium, and relates to the technical field of new energy, and the method comprises the steps: obtaining the operation data of a to-be-detected lithium battery; inputting the acquired operation data into a constructed state-of-charge prediction model to obtain a predicted state-of-charge of the to-be-detected lithium battery predicted by the state-of-charge prediction model; the state of charge prediction model predicts the state of charge based on an extended Kalman filtering framework, and updates an error covariance matrix of each round of an iteration process in the extended Kalman filtering framework by using a Levenberg-Marquardt method; each round of iteration is carried out through an iteration coefficient which is subjected to numerical adjustment according to the influence generated in iteration by the Levenberg-Marquardt method. According to the method, the state estimation stability and reliability are ensured, and the predicted state of charge obtained by final prediction is effectively prevented from deviating from an actual result.

Description

Technical Field

[0001] The present invention relates to the field of new energy technologies, and particularly to a method, device, and medium for predicting the state of charge of a lithium battery. Background Art

[0002] Lithium-ion batteries exhibit significant advantages in the energy storage field due to their excellent energy density, high power, and long life cycle. To ensure the safe and reliable use of the battery, a battery management system (BMS) is required to monitor the battery state in real time. The state of charge (SOC) of the battery, as a core monitoring parameter, its accurate estimation is of great significance for improving system performance, enhancing reliability, and extending the battery life.

[0003] Currently, SOC prediction techniques mainly include four categories: traditional methods, model-driven methods, filtering estimation methods, and data-driven methods. Among them, traditional methods mainly include the open-circuit voltage method and the ampere-hour integration method. These two methods are easy to operate, but are limited by the sensor accuracy and have relatively large errors. Model-driven methods cover equivalent circuit models, electrochemical models, and electrochemical impedance models. The computational complexity and prediction accuracy of these models are directly affected by the model complexity, and the established models are often common among different types of batteries. Filtering estimation methods, such as Kalman filtering and particle filtering, effectively combine models for SOC prediction, but their prediction accuracy is also limited by the accuracy of the model. Data-driven methods, including support vector machines, Gaussian process regression, neural network algorithms, etc., rely on a large amount of historical data to build network models, have relatively large computational complexity, the prediction accuracy is closely related to the selected training data set, and the generalization performance of the model is difficult to guarantee.

[0004] It can be seen that the existing SOC prediction techniques are difficult to reflect the true state of charge of the lithium battery. Therefore, a new method for predicting the state of charge of the lithium battery is needed. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a method, device, and medium for predicting the state of charge of a lithium battery, thereby solving the problem that the existing SOC prediction techniques are difficult to reflect the true state of charge of the lithium battery.

[0006] According to a first aspect, an embodiment of the present invention provides a method for predicting the state of charge of a lithium battery, the method comprising: Obtaining the operation data of the lithium battery to be detected; Input the obtained operating data into the constructed state of charge prediction model to obtain the predicted state vector of the lithium battery to be detected predicted by the state of charge prediction model; the predicted state vector is a vector composed of the predicted state of charge and the polarization voltage. The state of charge prediction model predicts the state of charge based on the extended Kalman filter framework, and uses the Levenberg-Marquardt method in the extended Kalman filter framework to update the error covariance matrix for each round of the iterative process, and, performs each round of iteration through the iteration coefficient; when the Levenberg-Marquardt method has a positive impact in the iteration, reduce the value of the iteration coefficient in the next round of iteration, and when the Levenberg-Marquardt method has a negative impact in the iteration, increase the value of the iteration coefficient in the next round of iteration.

[0007] Combined with the first aspect, in the first embodiment of the first aspect, the state of charge prediction model is constructed through the following steps: Obtain the actual open-circuit voltage corresponding to each actual state of charge of the sample lithium battery during the discharge process, and determine the state of charge-open circuit voltage curve according to the actual state of charge of the battery and the corresponding actual open-circuit voltage. Use the least squares method to fit the state of charge-open circuit voltage curve of the battery. Construct an extended Kalman filter framework according to the fitted state of charge-open circuit voltage curve of the battery, use the Levenberg-Marquardt method to update the error covariance matrix for each round of the iterative process of the extended Kalman filter framework and perform each round of iteration through the iteration coefficient.

[0008] Combined with the first embodiment of the first aspect, in the second embodiment of the first aspect, the obtaining the actual open-circuit voltage corresponding to each actual state of charge of the sample lithium battery during the discharge process, and determining the state of charge-open circuit voltage curve according to the actual state of charge of the battery and the corresponding actual open-circuit voltage specifically includes: Let the sample lithium battery stand for a preset time to ensure that the inside of the sample lithium battery reaches a stable state. Charge the sample lithium battery with a preset charging current. When the sample lithium battery is charged to the preset state of charge, charge the sample lithium battery with a preset charging voltage until the sample lithium battery is charged to the preset current state; the preset charging current during the charging process is a constant current, and the preset charging voltage during the charging process is a constant voltage. Discharge the sample lithium battery that has entered the preset current state with a constant discharge current, and determine the actual open-circuit voltage value of the sample lithium battery under the actual state of charge of the battery every time it discharges a preset state of charge value until the sample lithium battery discharges completely. Determine the open-circuit voltage value - state of charge curve based on the actual open-circuit voltage value and the actual state of charge corresponding to the actual open-circuit voltage value.

[0009] Combined with the first implementation manner of the first aspect, in the third implementation manner of the first aspect, for the construction of the extended Kalman filter framework, the Levenberg-Marquardt method is used to update the error covariance matrix in each round of the iterative process of the extended Kalman filter framework and perform iteration in each round through the iteration coefficient, specifically including: Construct an extended Kalman filter framework, and solve the initial state vector and the initial covariance matrix according to the fitted state of charge - open-circuit voltage curve; Use the Levenberg-Marquardt method to update the error covariance matrix in each round of the iterative process of the extended Kalman filter framework and perform iteration in each round through the iteration coefficient; Judge whether the iterative phase of the extended Kalman filter meets the iterative convergence condition. In the case of not meeting the iterative convergence condition, adjust the iterative parameters for the next iteration and update until the iterative convergence condition is met and the iteration stops; When the iterative phase of the extended Kalman filter meets the iterative convergence condition, update the Jacobian matrix, update the error covariance matrix at the end of iteration according to the updated Jacobian matrix, and verify the updated error covariance matrix. According to the verification result, use the optimal estimated value at the end of iteration or the optimal estimated value before update as the predicted state vector.

[0010] Combined with the third implementation manner of the first aspect, in the fourth implementation manner of the first aspect, for the construction of the extended Kalman filter framework and the solution of the initial state vector and the initial covariance matrix according to the fitted state of charge - open-circuit voltage curve, specifically including: Based on the ampere-hour integration method calculation formula and the discrete state space expression of the first-order resistor-capacitor equivalent circuit model, construct the total state expression and the total measurement expression, and solve the total state expression and the total measurement expression according to the fitted state of charge - open-circuit voltage curve; Initialize the parameters of the extended Kalman filter framework; Determine the initial state vector and the initial covariance matrix; The discrete state space expression of the first-order resistor-capacitor equivalent circuit model is obtained through the following steps: Construct a first-order resistor-capacitor equivalent circuit, and based on the first-order resistor-capacitor equivalent circuit, form a first-order resistor-capacitor equivalent circuit model; the first-order resistor-capacitor equivalent circuit model includes a voltage source of an open-circuit voltage, an ohmic internal resistance, a transfer internal resistance, and a polarization capacitor. The voltage source, the ohmic internal resistance, the transfer internal resistance, and the polarization capacitor are connected in parallel with each other and jointly describe the transient response characteristics of the lithium battery. The first-order resistor-capacitor equivalent circuit model is discretized to obtain a discrete state expression, and a discrete state space expression is determined based on the discrete state expression.

[0011] Combined with the fourth embodiment of the first aspect, in the fifth embodiment of the first aspect, when using the Levenberg-Marquardt method to update the error covariance matrix in each round of the extended Kalman filter framework iteration process and performing iteration in each round through an iteration coefficient, it specifically includes: Perform iterative update of the extended Kalman filter framework according to the initial covariance matrix and the iteration coefficient, and use the Levenberg-Marquardt method to correct the error covariance matrix in each round of the extended Kalman filter iteration stage; Determine the Kalman gain in each round of the extended Kalman filter iteration stage according to the error covariance matrix in each round; Update the optimal estimated value of the state of charge and the error covariance matrix corresponding to the iteration round according to the Kalman gain in each round.

[0012] Combined with the third embodiment of the first aspect, in the sixth embodiment of the first aspect, when determining whether the extended Kalman filter iteration stage satisfies the iteration convergence condition, and in the case of not satisfying the iteration convergence condition, adjust the iteration parameters for the next iteration and update until the iteration convergence condition is satisfied and the iteration stops, it specifically includes: Determine whether the extended Kalman filter iteration stage satisfies the iteration convergence condition; In the case of determining that the iteration convergence condition is not satisfied, obtain the loss function values of the current iteration round and the previous iteration round; When determining that the loss function value of the current iteration round is less than the loss function value of the previous iteration round, reduce the iteration parameters and perform the next round of iterative update until the iteration convergence condition is satisfied and the iteration stops; When determining that the loss function value of the current iteration round is not less than the loss function value of the previous iteration round, increase the iteration parameters and perform the next round of iterative update until the iteration convergence condition is satisfied and the iteration stops.

[0013] Combined with the third embodiment of the first aspect, in the seventh embodiment of the first aspect, when the extended Kalman filter iteration stage satisfies the iteration convergence condition, update the Jacobian matrix, update the error covariance matrix at the end of iteration according to the updated Jacobian matrix, and verify the updated error covariance matrix. According to the verification result, use the optimal estimated value at the end of iteration or the optimal estimated value before update as the predicted state vector, it specifically includes: When the extended Kalman filter iteration stage satisfies the iteration convergence condition, update the Jacobian matrix through the re-approximation method; Update the error covariance matrix at the iteration stop according to the updated Jacobian matrix; Obtain the traces of the error covariance matrices before and after the update. When it is determined that the trace of the updated error covariance matrix is less than the trace of the error covariance matrix before the update, use the updated optimal estimated value as the predicted state vector of the sample lithium battery; When it is determined that the trace of the updated error covariance matrix is not less than the trace of the error covariance matrix before the update, use the optimal estimated value at the iteration stop as the predicted state vector of the sample lithium battery.

[0014] According to a second aspect, an embodiment of the present invention further provides a state of charge prediction device for a lithium battery, and the device includes: A data acquisition module, configured to acquire the operation data of the lithium battery to be detected; A state prediction module, configured to input the acquired operation data into the constructed state of charge prediction model to obtain the predicted state vector of the lithium battery to be detected predicted by the state of charge prediction model; the predicted state vector is a vector composed of a predicted state of charge and a polarization voltage; The state of charge prediction model predicts the state of charge based on the extended Kalman filter framework, and uses the Levenberg-Marquardt method to update the error covariance matrix in each round of the iteration process in the extended Kalman filter framework, and performs each round of iteration through an iteration coefficient; when the Levenberg-Marquardt method has a positive impact on the iteration, reduce the value of the iteration coefficient in the next round of iteration, and when the Levenberg-Marquardt method has a negative impact on the iteration, increase the value of the iteration coefficient in the next round of iteration.

[0015] According to a third aspect, an embodiment of the present invention further provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the steps of the state of charge prediction method for a lithium battery as described in any one of the above.

[0016] According to a fourth aspect, an embodiment of the present invention further provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the state of charge prediction method for a lithium battery as described in any one of the above.

[0017] The method, device and medium for predicting the state of charge of a lithium battery according to the present invention predict the predicted state of charge of the lithium battery to be detected through the established state of charge prediction model. The state of charge prediction model predicts the state of charge based on the extended Kalman filter framework, and uses the L-M method in the extended Kalman filter framework to update the error covariance matrix in each round of the iterative process. Moreover, each round of iteration is performed through an iteration coefficient that adjusts numerically according to the influence generated by the L-M method during iteration. More specifically, when the L-M method has a positive influence during iteration, that is, the iteration direction is consistent with the convergence direction, the value of the iteration coefficient is reduced in the next round of iteration. When the L-M method has a negative influence during iteration, that is, the iteration direction is opposite to the convergence direction, the value of the iteration coefficient is increased in the next round of iteration. Introducing the L-M algorithm to iterate the extended Kalman filter can ensure global convergence, and dynamically adjusting the value of the parameter iteration coefficient avoids the local optimal solution trap that traditional methods may encounter. Compared with the traditional iterative extended Kalman filter, the improved extended Kalman filter framework used by the state of charge prediction model solves the convergence problem, pursues accurate estimation of the covariance matrix, and pursues a better Kalman gain. While ensuring the stability and reliability of the state estimation, it effectively prevents the predicted state of charge finally predicted from deviating from the actual result. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The features and advantages of the present invention will be more clearly understood by referring to the accompanying drawings. The drawings are schematic and should not be construed as imposing any limitation on the present invention. In the drawings: Figure 1 Shows a schematic flow chart of the method for predicting the state of charge of a lithium battery provided by the present invention; Figure 2 Shows a schematic structural diagram of the first-order RC equivalent circuit model constructed in the method for predicting the state of charge of a lithium battery provided by the present invention; Figure 3 Shows a schematic diagram of the OCV-SOC curve fitted in the method for predicting the state of charge of a lithium battery provided by the present invention; Figure 4 Shows a current data curve diagram under the simulated dynamic stability test (DST) working condition in the method for predicting the state of charge of a lithium battery provided by the present invention; Figure 5 Shows a voltage data curve diagram under the simulated dynamic stability test (DST) working condition in the method for predicting the state of charge of a lithium battery provided by the present invention; Figure 6 Shows an experimental result diagram under the simulated dynamic stability test (DST) working condition in the method for predicting the state of charge of a lithium battery provided by the present invention; Figure 7Shows the experimental result diagram under the simulated US06 working condition in the method for predicting the state of charge of the lithium battery provided by the present invention; Figure 8 Shows the structural schematic diagram of the device for predicting the state of charge of the lithium battery provided by the present invention; Figure 9 Shows the hardware structural schematic diagram of the electronic device provided by the present invention. Detailed implementation manners

[0019] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0020] With the increasing maturity of related technologies, new energy vehicles represented by pure electric vehicles, hybrid electric vehicles and fuel cell vehicles have developed vigorously. Lithium-ion batteries show significant advantages in the energy storage field due to their excellent energy density, high power and long life cycle. To ensure the safe and reliable use of the battery, the BMS needs to monitor the battery state in real time. During the actual use of the lithium battery, the SOC will continuously change with characteristic parameters such as the battery capacity, internal resistance-capacitance parameters, temperature, discharge rate and aging degree. As the core monitoring parameter, the accurate estimation of the SOC is of great significance for improving the system performance, enhancing the reliability and extending the battery life.

[0021] Currently, SOC prediction technologies mainly include four categories: traditional methods, model-driven methods, filtering estimation methods and data-driven methods.

[0022] Among them, traditional methods mainly include the open-circuit voltage method and the ampere-hour integration method. These two methods are easy to operate, but are limited by the sensor accuracy and have relatively large errors. The model-driven methods cover the equivalent circuit model, the electrochemical model, and the electrochemical impedance model. The equivalent circuit model is used to describe and simulate the dynamic characteristics of the battery. It regards the battery as a two-port network and uses devices such as voltage sources, resistors, and capacitors to form a circuit to simulate the internal characteristics of the battery. The electrochemical model calculates the terminal voltage and SOC of the battery according to the electrochemical reaction process and is a battery model based on the porous electrode and solution concentration theory. The electrochemical model mainly reflects the internal chemical reaction mechanism of the battery, with high model accuracy, but it is difficult to determine all parameters and has huge computational complexity and time consumption. The electrochemical impedance model can accurately describe the battery characteristics, but the matching process in practical applications is difficult, complex, and not intuitive, and the impedance model is only useful at specific SOCs and temperatures and cannot predict the DC reaction and battery operation time. The computational amount and prediction accuracy of these models are directly affected by the model complexity, and the established models are often general among different types of batteries. Filtering estimation methods, such as Kalman filtering and particle filtering, effectively combine models for SOC prediction, but their prediction accuracy is also limited by the accuracy of the models. Data-driven methods, including support vector machines, Gaussian process regression, neural network algorithms, etc., do not need to consider the complex chemical reaction mechanism inside the battery for SOC prediction based on data-driven methods. Instead, based on a large number of battery experimental test data, a mapping relationship model between external characteristic parameters such as current, voltage, and temperature and SOC is established and trained, which relies on a large amount of historical data to construct a network model and has a relatively large computational amount. Taking the neural network model as a typical representative, this method has extremely high fitting ability while ignoring the details of the internal chemical reactions of the battery, is applicable to the SOC estimation of various power batteries, and has high estimation accuracy. However, it requires a large amount of data for training and has a large computational amount. In practical applications, high-performance chips must be equipped, which increases the cost of the BMS. The prediction accuracy is closely related to the selected training data set, and it is difficult to guarantee the generalization performance of the model.

[0023] To achieve dynamic SOC estimation, filters and observers are often combined with the battery model to form a model-based SOC prediction method for SOC prediction. Commonly used filters and observers include Kalman filters, particle filters, H∞ filters, and other state observers (such as sliding mode observers). Such an estimation method uses a closed-loop structure. By continuously correcting the SOC estimation value, the SOC estimation value continuously approaches the true value, thereby making the algorithm have a certain robustness.

[0024] However, the prediction accuracy of the SOC prediction method that combines filters and observers with the battery model is often restricted by the covariance matrix estimation and is difficult to reflect the true state of charge of the lithium battery.

[0025] It can be seen that the existing SOC prediction technologies are difficult to reflect the true state of charge of lithium batteries. In view of the above problems, a new method for predicting the state of charge of lithium batteries is needed.

[0026] To solve the above problems, in this embodiment, a method for predicting the state of charge of a lithium battery is provided, aiming to obtain a better Kalman gain and reflect the true state of the lithium battery, thereby improving the accuracy of the state of charge prediction. The method for predicting the state of charge of a lithium battery according to the embodiment of the present invention can be used in electronic devices, and the electronic devices include but are not limited to computers, mobile terminals, etc. Figure 1 It is a schematic flowchart of the method for predicting the state of charge of a lithium battery according to the embodiment of the present invention, as Figure 1 shown, the method may include the following steps: S10. Obtain the operation data of the lithium battery to be detected. In this embodiment, the operation data includes but is not limited to the open-circuit voltage, rated capacity, terminal voltage, flowing current, sampling time interval, etc. of the lithium battery to be detected.

[0027] S20. Input the obtained operation data into the constructed state of charge prediction model to obtain the predicted state vector of the lithium battery to be detected predicted by the state of charge prediction model. The predicted state vector is a vector composed of the predicted state of charge and the polarization voltage. Specifically, the predicted state vector is a two-row and one-column matrix composed of the predicted state of charge and the polarization voltage.

[0028] Among them, the state of charge prediction model predicts the state of charge based on the extended Kalman filter framework, and uses the Levenberg-Marquardt (L-M) method in the extended Kalman filter framework to update the error covariance matrix for each round of the iterative process, and, performs each round of iteration through the iterative coefficient that numerically adjusts according to the influence generated by the L-M method in the iteration. More specifically, when the L-M method has a positive influence in the iteration, that is, the iteration direction is consistent with the convergence direction, the value of the iterative coefficient is reduced in the next round of iteration. When the L-M method has a negative influence in the iteration, that is, the iteration direction is opposite to the convergence direction, the value of the iterative coefficient is increased in the next round of iteration.

[0029] Compared with the traditional iterative extended Kalman filter, the improved extended Kalman filter framework used in the state of charge prediction model solves the convergence problem. Introducing the L-M algorithm to iterate the extended Kalman filter can ensure global convergence to pursue accurate estimation of the covariance matrix. Moreover, by using a dynamically changing iteration coefficient, a better Kalman gain is pursued, and the value of the parameter iteration coefficient is dynamically adjusted to avoid the local optimal solution trap that the traditional method may encounter. While ensuring the stability and reliability of state estimation, it effectively prevents the predicted state of charge obtained finally from deviating from the actual result.

[0030] In this embodiment, the state of charge prediction model is constructed through the following steps: S30. Obtain the actual open circuit voltage corresponding to each actual state of charge of the sample lithium battery during the discharge process, and determine the state of charge-open circuit voltage curve (SOC-OCV curve) according to the actual state of charge of the battery and the corresponding actual open circuit voltage.

[0031] In this embodiment, step S30 specifically includes: S31. Let the sample lithium battery stand for a preset time to ensure that the inside of the sample lithium battery reaches a stable state. This step is to let the sample lithium battery stand.

[0032] In this embodiment, the preset time can be configured by the user. For example, the preset time is set to 1h.

[0033] S32. Charge the sample lithium battery with a preset charging current. When the sample lithium battery is charged to the preset state of charge, charge the sample lithium battery with a preset charging voltage until the sample lithium battery is charged to the preset current state. This step is to charge the sample lithium battery.

[0034] In this embodiment, the preset state of charge is the state when the lithium battery is close to full charge, such as 90% SOC; the preset current state is that the current of the lithium battery drops to a certain threshold to ensure that the battery is fully charged.

[0035] To ensure the controllability and stability during the initial charging process, the preset charging current during the charging process is a constant current, and the preset charging voltage during the charging process is a constant voltage. That is, in this step, the sample lithium battery is first charged with a preset constant current, and then switched to be charged with a preset constant voltage.

[0036] S33. Discharge the sample lithium battery that has entered the preset current state with a constant discharge current, and determine the actual open circuit voltage value of the sample lithium battery under the actual state of charge every time a preset state of charge value is discharged, that is , until the discharge of the sample lithium battery ends. This step is to discharge the sample lithium battery.

[0037] In this embodiment, discharging is carried out at a constant current, and the corresponding actual open-circuit voltage value is recorded every time the sample lithium battery discharges to a preset state of charge value, such as 2% SOC. In this way, every time the SOC of the sample lithium battery reaches an actual SOC value during the discharging process, the corresponding actual open-circuit voltage value can be recorded. Considering the calculation accuracy, the decrease in SOC is accurately calculated by the ampere-hour integration method during the discharging process of the sample lithium battery.

[0038] S34. Determine the open-circuit voltage value - state of charge curve according to the actual open-circuit voltage value and the actual state of charge of the battery corresponding to the actual open-circuit voltage value. In this step, the recorded OCV values and the corresponding SOC values are plotted to form a state of charge - open-circuit voltage curve (SOC-OCV curve).

[0039] In this embodiment, in order to ensure that the drawn OCV-SOC curve can be applied to the subsequent prediction of the SOC of the battery, the processes of standing, charging, discharging and recording are repeated multiple times. And after each discharging, the lithium battery to be tested is allowed to stand for a long enough time again, and then the open-circuit voltage value is measured, that is, steps S31 to S33 are repeated.

[0040] S40. Fit the state of charge - open-circuit voltage curve (SOC-OCV curve) by the least squares method.

[0041] Among them, in the recorded SOC-OCV curve, the x-axis is the SOC value and the y-axis is the OCV value. The functional relationship between SOC and OCV can be obtained by the data fitting method.

[0042] Since the fitting effect of the SOC-OCV curve is affected by the polynomial order, after the fitting is completed, the root mean squared error (RMSE), coefficient of determination (R-square), and sum of squares due to error (SSE) are used to evaluate the quality of the curve fitting effect; if the fitting effect is not good, the curve fitting effect is improved by increasing the order of the polynomial until the evaluation index results do not improve significantly or exceed the set threshold. The evaluation of the fitting effect is shown in Table 1.

[0043] Table 1 Polynomial fitting results

[0044] The power of the function has a positive correlation mapping relationship with the fitting accuracy. However, the larger the power, the greater the computational complexity. When the order is low, the curve cannot better reflect the relationship between SOC and OCV. But when the order is too high, the curve will oscillate, thus affecting the accuracy. Please refer to Figure 2 , in this embodiment, a ninth-degree polynomial is used to fit the SOC-OCV curve. Correspondingly, the functional relationship between the obtained SOC and OCV is: OCV(SOC)=2952.557SOC 9 -13998.9416SOC 8 +28345.1537SSOC 7 -32005.7919SOC 6 +22068.5577SOC 5 -9558.2579SOC 4 +2573.9902SOC 3 -411.0355SOC 2 +35.3359SOC+1.9273 S50. Construct an extended Kalman filter framework, use the L-M method to update the error covariance matrix for each round of the iterative process of the extended Kalman filter framework, and perform iteration for each round through the iteration coefficient.

[0045] In this embodiment, step S50 specifically includes: S51. Construct an extended Kalman filter framework, and solve the initial state vector and the initial covariance matrix according to the fitted battery state of charge-open circuit voltage curve, laying a foundation for solving the optimal estimated value of the state of charge later.

[0046] In this embodiment, step S51 specifically includes: S511. Based on the ampere-hour integration method calculation formula and the discrete state-space expression of the first-order resistor-capacitance (Resistor-Capacitance) equivalent circuit model, construct the total state expression That is, the total state expression equation (calculation formula) and the total measurement expression That is, the total measurement equation (calculation formula), and solve the total state expression according to the fitted SOC-OCV curve and the total measurement expression , specifically:

[0047] Among them, represents the state of charge of the sample lithium battery at time represents The open-circuit voltage corresponding to the SOC value of the lithium-ion battery at a moment sample; denote The polarization voltage of the lithium-ion battery at a moment sample; denote the rated capacity of the sample lithium-ion battery; denote the transfer internal resistance; denote the polarization capacitance; denote the ohmic internal resistance; denote the preset sampling time interval, that is, the corresponding sampling interval of voltage and current in time; denote the Coulomb efficiency, which can be used to represent the ratio of the actual available charge to the theoretical charge during the charge and discharge process of the sample lithium-ion battery ( 1); denote , that is, the exponential function with as the base; denote The terminal voltage of the lithium-ion battery at a moment sample; denote The dynamic current of the lithium-ion battery at a moment (that is, a corresponding discrete time), the positive value of which represents that the lithium-ion battery is in the charging state, and the negative value represents that the lithium-ion battery is discharging; denote the measurement noise; denote the process noise.

[0048] Construct the above total state expression and the total measurement expression When constructing, various variable parameters will also be constructed. Specifically:

[0049]

[0050]

[0051]

[0052] In this process, the state variable is selected as ; denote the preset state transition function (matrix); denote the preset control input matrix; denote the slope of the OCV-SOC curve at the predicted SOC; denote the partial derivative of the terminal voltage with respect to the polarization voltage, and take the sensitivity of the terminal voltage to the polarization voltage as a fixed value ; denote at moment for the Jacobian matrix predicted at the moment, and this parameter is predicted from the total measurement expression ; is Always The predicted state vector obtained by moment prediction is expressed by the total state Predicted.

[0053] In this embodiment, the discrete state space expression of the first-order RC equivalent circuit model is obtained by the following steps: A10. First, build a first-order resistor-capacitance equivalent circuit. Based on the first-order RC equivalent circuit, a first-order RC equivalent circuit model is constructed by an RC network structure and a resistor in series. The core components of the first-order RC equivalent circuit model include an open-circuit voltage source. , one ohm internal resistance , a transfer internal resistance and a polarized capacitor ,These components are connected to each other in parallel and together describe the transient response characteristics of the lithium battery.

[0054] like Figure 3 As shown, in this first-order RC equivalent circuit model, express Terminal voltage at any moment (this parameter reflects the battery voltage state at the moment), represents the ohmic internal resistance, represents the internal transmission resistance, Represents polarized capacitance. According to Kirchhoff's Voltage Law (KVL) and Kirchhoff's Current Law (KCL).

[0055] At this time, for the sample lithium battery, Terminal voltage at time and dynamic current It can be expressed as:

[0056]

[0057] in, express The open circuit voltage at the moment; It represents the voltage value of the lithium battery when no current flows. This parameter is usually expressed as a complex nonlinear function.

[0058] A20. For a lithium battery, the state of charge (SOC) of the battery is numerically calculated as the ratio of the remaining battery charge to the rated battery charge. Based on this, discretization processing is performed on the first-order RC equivalent circuit model composed of the first-order RC equivalent circuit, and a discrete state expression can be obtained. Based on the discrete state expression, a discrete state space expression is determined. Among them, the discrete state expression, that is, the discrete state equation (calculation formula) is:

[0059] The discrete state space expression, that is, the discrete state equation (calculation formula) is:

[0060] The voltage mutation caused by the transient change of current in the lithium battery is caused by the ohmic internal resistance of the lithium battery Specifically:

[0061] Among them, represents the voltage value at an instantaneous moment before the start of lithium battery discharge; represents the voltage value when the lithium battery starts to discharge; represents the voltage value when the lithium battery discharge ends; represents the voltage value at an instantaneous moment after the end of lithium battery discharge; represents the discharge current during the lithium battery discharge process.

[0062] During the discharge process, the voltage mutation of the lithium battery is caused by the ohmic internal resistance and is manifested as an instantaneous voltage drop. The slow decrease of the voltage is dominated by the polarization effect, and its dynamic characteristics are related to the parallel link composed of the transfer internal resistance and the polarization capacitance in the first-order RC equivalent circuit (i.e., the response under a step current excitation); after the discharge ends, the battery polarization effect disappears, and the voltage of the lithium battery gradually rises, which is related to the parallel link composed of the transfer internal resistance and the polarization capacitance in the first-order RC equivalent circuit. At this time, the lithium battery is in a zero-input response state.

[0063] According to the exponential fitting of the voltage rise process at the end of discharge, the specific values of the two parameters of the transfer internal resistance and the polarization capacitance in the first-order RC equivalent circuit can be obtained. Specifically, the fitting forms of these two parameters are:

[0064] Among them, represents the time constant, which is used to reflect the attenuation rate of the polarization voltage, ; Assume the initial value of the polarization voltage , transfer internal resistance Calculate through the initial value of the polarization voltage and the discharge current . Specifically:

[0065] The time constant obtained by fitting and the transfer internal resistance are used to calculate the polarization capacitance . Specifically:

[0066] S512. Initialize the parameters of the extended Kalman filter framework, that is, set the Kalman filter framework parameters , , , and to their initial values. Among them, for the parameters , , and , please refer to the foregoing. represents the initial covariance matrix. These parameters are all involved in the extended Kalman filtering process, and an effective initial value is set for the state of charge prediction model to avoid filter divergence.

[0067] S513. Determine the initial state vector and the initial covariance matrix. In this embodiment, the state vector is updated through the following calculation formula to predict the initial state vector of the iterative process. Specifically:

[0068] where represents the initial state vector, and is also the predicted state vector at time for time . It should be noted that the initial time of the iterative process is usually set to time 0, so the initial state vector is ; represents the predicted state vector at time

[0069] After the previous steps, the specific values of the parameters and can be obtained, which will not be elaborated here.

[0070] After that, the initial covariance matrix corresponding to the initial state vector is updated through the following calculation formula . Specifically:

[0071] Among them, represents the initial covariance matrix, and is also the error covariance matrix at time It should be noted that the initial time of the iterative process is usually set to time 0, so the initial covariance matrix is ; represents the error covariance matrix at time (the starting point of recursion is time represents the transpose of the preset state transition function (matrix); represents the covariance of the process noise.

[0072] S52. Use the L-M method to update the error covariance matrix in each round of the iterative process of the extended Kalman filter framework and perform iteration in each round through the iteration coefficient.

[0073] In this embodiment, step S52 specifically includes: S521. Perform iterative update of the extended Kalman filter framework according to the initial covariance matrix and the iteration coefficient, and use the L-M method to correct the error covariance matrix in each round of the iterative stage of the extended Kalman filter, that is, the prediction stage. Specifically, update the predicted state vector and the error covariance matrix in each round of the iterative process through the following calculation formula:

[0074] Among them, represents the predicted state vector at time in the -th round of the iterative process for time; represents the error covariance matrix of the predicted state vector at time in the -th round of the iterative process for time.

[0075] The correction process of the error covariance matrix is as follows:

[0076] Among them, represents the corrected error covariance matrix in the -th round of the iterative process, that is, the result after the error covariance matrix is corrected; represents the iteration coefficient. The iteration coefficient is introduced in the correction process of the error covariance matrix. The introduction of this parameter can ensure the global convergence of the extended Kalman filter.

[0077] When the value is large, it means that the iterative result of the extended Kalman filter introducing the L-M method is closer to the result of gradient descent. On the contrary, When the value is small, it means that the iterative result of the extended Kalman filter introducing the L-M method is closer to the calculation result obtained by the Gauss-Newton method.

[0078] In this embodiment, during the actual calculation process, When the value is set small, the iterative result will be consistent with the convergence direction. For example, the parameter can be set to 0.15 for the initial value.

[0079] S522. Determine the Kalman gain for each round in the extended Kalman filter iteration stage according to the error covariance matrix of each round. For example, determine the Kalman gain for the first round in the extended Kalman filter iteration stage according to the first-round, that is, the initial covariance matrix. Specifically:

[0080]

[0081]

[0082] Among them, represents the covariance between the predicted state vector at time in the -th round of iteration and the observed predicted value at time ; represents the predicted state vector used in the calculation process; represents the observed predicted value used in the calculation process, represents the covariance of the observed predicted value itself at time in the -th round of iteration; represents the covariance of the observed predicted value itself at time represents the Jacobian matrix at time with respect to ; represents the transpose of the Jacobian matrix ; represents the Kalman gain in the i -th round of iteration; represents the noise variance of the Jacobian matrix, that is, the noise variance of the measurement value obtained from the overall measurement expression which is the noise variance of the measurement value, that is, the observed data, obtained from the expression.

[0083] In this embodiment, the observation function is the overall measurement expression after removing the noise part, that is, ; Substitute the predicted state vector into the observation function The obtained observation prediction value corresponding to the predicted state vector at that moment.

[0084] It can be understood that, similar to the predicted state vector, the true state vector is a two-row and one-column matrix composed of the true state of charge and the polarization voltage.

[0085] S523. Update the optimal estimated value of the state of charge and the error covariance matrix corresponding to each iteration round according to the Kalman gain of each round. Specifically:

[0086]

[0087] Among them, represents the optimal estimated value in the round of iteration process; represents the actual observed value; represents the above-mentioned total measurement expression; represents at the moment of the Jacobian matrix at the moment, which is the current moment.

[0088] Actually, is the actual observed value obtained by substituting the true state vector into the total measurement expression .

[0089] S53. Determine whether the extended Kalman filter iteration stage meets the iteration convergence condition. If the iteration convergence condition is not met, adjust the iteration parameters for the next iteration and update until the iteration convergence condition is met and the iteration stops.

[0090] In this embodiment, the iteration convergence condition is that the iteration error between two adjacent iterations is less than the preset error or the maximum number of iteration rounds is reached. The preset error represents the error between two adjacent iterations. If this value is less than a certain value, it can be considered that the iteration result has converged to the extreme point, and at this time, the iteration can be terminated. For example, set the value of the parameter to 0.0001; or the current iteration round has reached the maximum number of iterations, and the maximum number of iterations can be set to 20.

[0091] The calculation method of the iteration error is:

[0092] Secondly, by comparing the magnitudes of the loss function values after two adjacent rounds of iteration, it can be determined whether the L-M method has a positive impact on the iteration during the iteration. The calculation formula of the loss function is:

[0093] Among them, indicates the loss function value of the round iteration.

[0094] The loss function values after two adjacent iterations correspond to and respectively. Next, compare and for the values of the two parameters. When , it indicates that the iteration is convergent, that is, the iteration direction is correct. At this time, reduce the value and use the reduced as the new iteration parameter, and then perform the next round of iterative update with the new iteration parameter. When , it indicates that the iteration direction is opposite to the convergence trend. At this time, increase the value and use the increased as the new iteration parameter, and then perform the next round of iterative update with the new iteration parameter. That is, when a positive effect is produced, reduce the iteration parameter and perform the next round of iterative update. When an opposite effect is produced, increase the iteration parameter and perform the next round of iterative update.

[0095] Preferably, when reducing the value, reduce to 0.5 , and when increasing the value, increase to 4 . By dynamically adjusting the value of the parameter , the trap of local optimal solutions that may be encountered in traditional methods is avoided.

[0096] In this embodiment, step S53 specifically includes: S531. Determine whether the extended Kalman filter iteration stage satisfies the iteration convergence condition.

[0097] S532. In the case of determining that the iteration convergence condition is not satisfied, obtain the loss function values of the current iteration round and the previous iteration round, that is, obtain and .

[0098] S533. When determining that the loss function value of the current iteration round is less than the loss function value of the previous iteration round, reduce the iteration parameter and perform the next round of iterative update until the iteration convergence condition is satisfied and the iteration stops.

[0099] S534. When determining that the loss function value of the current iteration round is not less than the loss function value of the previous iteration round, increase the iteration parameter and perform the next round of iterative update until the iteration convergence condition is satisfied and the iteration stops.

[0100] S54. When the extended Kalman filter iteration stage meets the iteration convergence condition, update the Jacobian matrix, update the error covariance matrix at the end of iteration according to the updated Jacobian matrix, and verify the updated error covariance matrix. According to the verification result, use the optimal estimated value at the end of iteration or the optimal estimated value before update as the predicted state vector.

[0101] Through recalibration of the error covariance matrix and verification of the recalibration result, re-approximation processing can be performed after updating the state vector. In this process, by comparing the observed data and the model prediction result, existing biases are accurately identified and corrected, and then the covariance matrix estimation is updated and optimized. Verification of the recalibration result of the error covariance matrix further ensures the convergence of the state of charge prediction model, avoids divergence, and thus improves the accuracy of predicting the state of charge.

[0102] S541. When the extended Kalman filter iteration stage meets the iteration convergence condition, update and calibrate the Jacobian matrix by the re-approximation method. Specifically:

[0103] where denotes the total measurement expression at time the partial derivative of the predicted Jacobian matrix with respect to the total state expression at time the predicted state vector obtained by prediction.

[0104] S542. Update the error covariance matrix at the end of iteration according to the updated Jacobian matrix. Specifically:

[0105]

[0106]

[0107] where denotes the covariance between the predicted state vector and the observed predicted value at the end of iteration, i.e., at time ; denotes the transpose of the Jacobian matrix at the end of iteration, i.e., at time ; denotes the covariance of the observed predicted value itself at the end of iteration, i.e., at time ; denotes the transpose of ; ; denotes the Kalman gain at the end of iteration; denotes The transpose of; Denotes the updated error covariance matrix, that is The error covariance matrix of the predicted state vector at time, it can be seen that From the one before the update Obtained, Denotes the error covariance matrix when the iteration stops, that is, when the iteration stops At time for The error covariance matrix of the predicted state vector at time.

[0108] S543. Obtain the traces of the error covariance matrices before and after the update. When it is determined that the trace of the updated error covariance matrix is less than the trace of the error covariance matrix before the update, use the updated optimal estimate value as the predicted state vector of the sample lithium battery.

[0109] In this embodiment, first solve the traces of the error covariance matrices before and after the update. Specifically, if , then The one obtained after the update The corresponding ( Denotes the updated optimal estimate value) as the predicted state vector of the sample lithium battery.

[0110] S544. When it is determined that the trace of the updated error covariance matrix is not less than the trace of the error covariance matrix before the update, use the optimal estimate value when the iteration stops as the predicted state vector of the sample lithium battery.

[0111] Specifically, if , then The corresponding ( Denotes the optimal estimate value when the iteration stops) as the predicted state of charge of the sample lithium battery. Where Denotes the operation of solving the unscented part, that is, finding the trace.

[0112] In this embodiment, after updating the predicted state vector, judge whether to use the optimal estimate value before or after the update as the predicted state vector of the sample lithium battery according to the trace of the error covariance matrix, so as to perform recalibration of the predicted state vector, realize accurate estimation of the state of charge of the battery, further ensure the convergence of the model, avoid divergence, and thus improve the accuracy of the predicted state vector obtained by prediction The method for predicting the state of charge of a lithium battery according to the present invention obtains the state of charge of the lithium battery to be detected through the constructed state of charge prediction model. The state of charge prediction model predicts the state of charge based on the extended Kalman filter framework, and uses the L-M method in the extended Kalman filter framework to update the error covariance matrix in each round of the iterative process. Moreover, each round of iteration is performed through an iterative coefficient that adjusts the value numerically according to the influence generated by the L-M method during the iteration. More specifically, when the L-M method has a positive influence during the iteration, that is, the iteration direction is consistent with the convergence direction, the value of the iterative coefficient is reduced in the next round of iteration. When the L-M method has a negative influence during the iteration, that is, the iteration direction is opposite to the convergence direction, the value of the iterative coefficient is increased in the next round of iteration. Introducing the L-M algorithm to iterate the extended Kalman filter can ensure global convergence, and dynamically adjusting the value of the iterative coefficient avoids the local optimal solution trap that traditional methods may encounter. Compared with the traditional iterative extended Kalman filter, the improved extended Kalman filter framework used by the state of charge prediction model solves the convergence problem, pursues an accurate estimation of the covariance matrix, and also pursues a better Kalman gain. While ensuring the stability and reliability of the state estimation, it effectively prevents the predicted state of charge obtained by the final prediction from deviating from the actual result.

[0113] Based on the already constructed state of charge prediction model, charge and discharge verification is carried out on battery model A in a laboratory environment. Specifically, please refer to Figure 4 and Figure 5 . In these two figures, the unit of the abscissa is seconds (S), and the units of the ordinate are amperes (A) and volts (V) respectively. The data set used in the verification process is the battery voltage and current data obtained by the laboratory simulating the DST and US06 working conditions on battery model A; please refer to Figure 6 and Figure 7 . Among them, Figure 6 shows the experimental result diagram under the simulated dynamic stability test (DST) working condition in the method for predicting the state of charge of the lithium battery provided by the present invention, Figure 7 shows the experimental result diagram under the simulated US06 working condition in the method for predicting the state of charge of the lithium battery provided by the present invention. In these two figures, the unit of the abscissa is seconds (S), and the unit of the ordinate is the percentage of the specific SOC. It can be seen that there is a certain deviation between the SOC calculated by the ampere-hour integration method at the second level and the SOC calculated by the ampere-hour integration method at the millisecond level. In order to evaluate more accurately, assuming that the calculation result of the ampere-hour integration at the millisecond level is closer to the true value, on this premise, the SOC value estimated by the improved extended Kalman filter algorithm shows a high consistency with this assumed true value, and its error range is relatively small. This result indicates that the already constructed state of charge prediction model has significant advantages in improving the accuracy of SOC estimation.

[0114] The battery state of charge prediction device for a lithium battery provided by the embodiment of the present invention will be described below. The battery state of charge prediction device for a lithium battery described below can be correspondingly referred to the battery state of charge prediction method for a lithium battery described above.

[0115] To solve the above problems, in this embodiment, a battery state of charge prediction device for a lithium battery is provided, aiming to obtain a better Kalman gain, and at the same time reflect the real state of the lithium battery, so as to improve the accuracy of battery state of charge prediction. Figure 8 It is a schematic structural diagram of the battery state of charge prediction method for a lithium battery according to the embodiment of the present invention. As Figure 8 shown, this device may include: A data acquisition module 10, configured to acquire the operation data of the lithium battery to be detected. In this embodiment, the operation data includes but is not limited to the polarization voltage, rated capacity, terminal voltage, flowing current, sampling time interval, etc. of the lithium battery to be detected.

[0116] A state prediction module 20, configured to input the acquired operation data into the constructed state of charge prediction model, and obtain a predicted state vector of the lithium battery to be detected predicted by the state of charge prediction model. The predicted state vector is a vector composed of a predicted state of charge and a polarization voltage. Specifically, the predicted state vector is a two-row and one-column matrix composed of a predicted state of charge and a polarization voltage.

[0117] The battery state of charge prediction device for a lithium battery of the present invention predicts the predicted state of charge of the lithium battery to be detected through the constructed state of charge prediction model. The state of charge prediction model predicts the state of charge based on the extended Kalman filter framework, and uses the L-M method to update the error covariance matrix in each round of the iterative process in the extended Kalman filter framework. In addition, each round of iteration is performed through an iterative coefficient that adjusts numerically according to the influence generated by the L-M method in the iteration. More specifically, when the L-M method generates a positive influence in the iteration, that is, the iteration direction is consistent with the convergence direction, the value of the iterative coefficient is reduced in the next round of iteration. When the L-M method generates a negative influence in the iteration, that is, the iteration direction is opposite to the convergence direction, the value of the iterative coefficient is increased in the next round of iteration. Introducing the L-M algorithm to iterate the extended Kalman filter can ensure global convergence, and dynamically adjusting the value of the iterative coefficient avoids the local optimal solution trap that may be encountered in the traditional method. Compared with the traditional iterative extended Kalman filter, the improved extended Kalman filter framework used by the state of charge prediction model solves the convergence problem, pursues accurate estimation of the covariance matrix, and pursues a better Kalman gain. While ensuring the stability and reliability of state estimation, it effectively prevents the predicted state of charge finally predicted from deviating from the actual result.

[0118] Figure 9 An entity structure schematic diagram of an electronic device is illustrated, such as Figure 9 shown. The electronic device may include: a processor 910, a communications interface 920, a memory 930, and a communication bus 940. Among them, the processor 910, the communications interface 920, and the memory 930 complete communication with each other through the communication bus 940. The processor 910 may call logical commands in the memory 930 to execute a method for predicting the state of charge of a lithium battery. The method includes: Obtaining operation data of the lithium battery to be detected; Inputting the obtained operation data into a constructed state of charge prediction model to obtain a predicted state vector of the lithium battery to be detected predicted by the state of charge prediction model; the predicted state vector is a vector composed of a predicted state of charge and a polarization voltage; The state of charge prediction model predicts the state of charge based on an extended Kalman filter framework, and uses the Levenberg-Marquardt method in the extended Kalman filter framework to update the error covariance matrix for each round of the iterative process, and, performs each round of iteration through an iteration coefficient; when the Levenberg-Marquardt method has a positive impact on the iteration, the value of the iteration coefficient is reduced in the next round of iteration, and when the Levenberg-Marquardt method has a negative impact on the iteration, the value of the iteration coefficient is increased in the next round of iteration.

[0119] In addition, when the logical instructions in the above-mentioned memory 930 are implemented in the form of software functional units and sold or used as an independent product, they may be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, may be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. And the aforementioned storage medium includes: various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memor), a random access memory (RAM, Random Access Memory), a magnetic disk, or an optical disc that can store program codes.

[0120] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for predicting the state of charge of a lithium battery, characterized in that, The method includes: Obtaining the operating data of the lithium battery to be detected; Inputting the obtained operating data into the constructed state of charge prediction model to obtain the predicted state vector of the lithium battery to be detected predicted by the state of charge prediction model; the predicted state vector is a vector composed of the predicted state of charge and the polarization voltage; The state of charge prediction model predicts the state of charge based on the extended Kalman filter framework, and uses the Levenberg-Marquardt method to update the error covariance matrix of each round in the iterative process of the extended Kalman filter framework, and performs each round of iteration through the iteration coefficient; when the Levenberg-Marquardt method has a positive impact in the iteration, the value of the iteration coefficient is reduced in the next round of iteration, and when the Levenberg-Marquardt method has a negative impact in the iteration, the value of the iteration coefficient is increased in the next round of iteration.

2. The method for predicting the state of charge of a lithium battery according to claim 1, wherein The state of charge prediction model is constructed through the following steps: Obtaining the actual open-circuit voltage corresponding to each actual state of charge of the sample lithium battery during the discharge process, and determining the state of charge-open circuit voltage curve according to the actual state of charge of the battery and the corresponding actual open-circuit voltage; Fitting the state of charge-open circuit voltage curve of the battery using the least squares method; Constructing an extended Kalman filter framework according to the fitted state of charge-open circuit voltage curve of the battery, using the Levenberg-Marquardt method to update the error covariance matrix of each round in the iterative process of the extended Kalman filter framework and performing each round of iteration through the iteration coefficient.

3. The method for predicting the state of charge of a lithium battery according to claim 2, wherein, The obtaining the actual open-circuit voltage corresponding to each actual state of charge of the sample lithium battery during the discharge process, and determining the state of charge-open circuit voltage curve according to the actual state of charge of the battery and the corresponding actual open-circuit voltage specifically includes: Letting the sample lithium battery stand for a preset time to ensure that the inside of the sample lithium battery reaches a stable state; Charging the sample lithium battery with a preset charging current. When the sample lithium battery is charged to the preset state of charge, charging the sample lithium battery with a preset charging voltage until the sample lithium battery is charged to the preset current state; the preset charging current during the charging process is a constant current, and the preset charging voltage during the charging process is a constant voltage; Discharging the sample lithium battery that has entered the preset current state with a constant discharge current, and determining the actual open-circuit voltage value of the sample lithium battery in the actual state of charge of the battery every time a preset state of charge value is discharged until the sample lithium battery finishes discharging; Determining the open-circuit voltage value-state of charge curve according to the actual open-circuit voltage value and the corresponding actual state of charge of the battery.

4. The method for predicting the state of charge of a lithium battery according to claim 2, wherein The constructing the extended Kalman filter framework, using the Levenberg-Marquardt method to update the error covariance matrix of each round in the iterative process of the extended Kalman filter framework and performing each round of iteration through the iteration coefficient specifically includes: Constructing an extended Kalman filter framework, and solving the initial state vector and the initial covariance matrix according to the fitted state of charge-open circuit voltage curve of the battery; Update the error covariance matrix for each round of the iterative process of the extended Kalman filter framework using the Levenberg-Marquardt method and perform each round of iteration through the iteration coefficient; Judge whether the iterative stage of the extended Kalman filter meets the iterative convergence condition. If the iterative convergence condition is not met, adjust the iterative parameters for the next iteration and update until the iterative convergence condition is met and the iteration stops; When the iterative stage of the extended Kalman filter meets the iterative convergence condition, update the Jacobian matrix, update the error covariance matrix at the end of iteration according to the updated Jacobian matrix, and verify the updated error covariance matrix. According to the verification result, the optimal estimated value at the end of iteration or the optimal estimated value before update will be used as the predicted state vector.

5. The method for predicting the state of charge of a lithium battery according to claim 4, wherein The construction of the extended Kalman filter framework and the solution of the initial state vector and the initial covariance matrix according to the fitted state of charge-open circuit voltage curve specifically include: Construct the total state expression and the total measurement expression based on the ampere-hour integration method calculation formula and the discrete state space expression of the first-order resistor-capacitor equivalent circuit model, and solve the total state expression and the total measurement expression according to the fitted state of charge-open circuit voltage curve; Initialize the parameters of the extended Kalman filter framework; Determine the initial state vector and the initial covariance matrix; The discrete state space expression of the first-order resistor-capacitor equivalent circuit model is obtained through the following steps: Construct a first-order resistor-capacitor equivalent circuit, and form a first-order resistor-capacitor equivalent circuit model based on the first-order resistor-capacitor equivalent circuit; the first-order resistor-capacitor equivalent circuit model includes a voltage source of an open-circuit voltage, an ohmic internal resistance, a transfer internal resistance, and a polarization capacitor. The voltage source, the ohmic internal resistance, the transfer internal resistance, and the polarization capacitor are connected in parallel to jointly describe the transient response characteristics of the lithium battery; Perform discretization processing on the first-order resistor-capacitor equivalent circuit model to obtain a discrete state expression, and determine the discrete state space expression based on the discrete state expression.

6. The method for predicting the state of charge of a lithium battery according to claim 5, characterized in that, The use of the Levenberg-Marquardt method to update the error covariance matrix for each round of the iterative process of the extended Kalman filter framework and perform each round of iteration through the iteration coefficient specifically includes: Perform iterative update of the extended Kalman filter framework according to the initial covariance matrix and the iteration coefficient, and use the Levenberg-Marquardt method to correct the error covariance matrix for each round of the iterative stage of the extended Kalman filter; Determine the Kalman gain for each round of the iterative stage of the extended Kalman filter according to the error covariance matrix for each round; Update the optimal estimated value of the state of charge and the error covariance matrix for the corresponding iteration round according to the Kalman gain for each round.

7. The method for predicting the state of charge of a lithium battery according to claim 4, wherein The judgment of whether the iterative stage of the extended Kalman filter meets the iterative convergence condition, and when the iterative convergence condition is not met, adjust the iterative parameters for the next iteration and update until the iterative convergence condition is met and the iteration stops, specifically includes: Judge whether the iterative stage of the extended Kalman filter meets the iterative convergence condition; When it is determined that the iterative convergence condition is not satisfied, obtain the loss function values of the current iteration round and the previous iteration round; When it is determined that the loss function value of the current iteration round is less than that of the previous iteration round, shrink the iteration parameter and perform the next round of iterative update until the iterative convergence condition is satisfied and the iteration stops; When it is determined that the loss function value of the current iteration round is not less than that of the previous iteration round, expand the iteration parameter and perform the next round of iterative update until the iterative convergence condition is satisfied and the iteration stops.

8. The method for predicting the state of charge of a lithium battery according to claim 4, wherein When the extended Kalman filter iteration stage satisfies the iterative convergence condition, update the Jacobian matrix, update the error covariance matrix at the end of iteration according to the updated Jacobian matrix, and verify the updated error covariance matrix. According to the verification result, use the optimal estimated value at the end of iteration or the optimal estimated value before update as the predicted state of charge. Specifically, it includes: When the extended Kalman filter iteration stage satisfies the iterative convergence condition, update the Jacobian matrix by the re-approximation method; Update the error covariance matrix at the end of iteration according to the updated Jacobian matrix; Obtain the traces of the error covariance matrices before and after update. When it is determined that the trace of the updated error covariance matrix is less than that of the error covariance matrix before update, use the updated optimal estimated value as the predicted state vector of the sample lithium battery; When it is determined that the trace of the updated error covariance matrix is not less than that of the error covariance matrix before update, use the optimal estimated value at the end of iteration as the predicted state vector of the sample lithium battery.

9. A state of charge prediction device for a lithium battery, characterized in that, The device includes: A data acquisition module, configured to acquire the operation data of the lithium battery to be detected; A state prediction module, configured to input the acquired operation data into the constructed state of charge prediction model to obtain the predicted state vector of the lithium battery to be detected predicted by the state of charge prediction model; the predicted state vector is a vector composed of the predicted state of charge and the polarization voltage; The state of charge prediction model predicts the state of charge based on the extended Kalman filter framework, and uses the Levenberg-Marquardt method to update the error covariance matrix of each round in the iterative process in the extended Kalman filter framework, and performs each round of iteration through the iteration coefficient; when the Levenberg-Marquardt method has a positive impact on the iteration, reduce the value of the iteration coefficient in the next round of iteration, and when the Levenberg-Marquardt method has a negative impact on the iteration, increase the value of the iteration coefficient in the next round of iteration.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for predicting the state of charge of the lithium battery according to any one of claims 1 to 8.

Images