Parameter identification method of equivalent circuit model, SOC estimation method and system

Abstract

The invention belongs to the technical field of battery management, and particularly discloses a parameter identification method of an equivalent circuit model and an SOC estimation method and system.The parameter identification method of the equivalent circuit model comprises the steps that the equivalent circuit model in the charging direction and the discharging direction is built; respectively carrying out mixed pulse tests on the equivalent circuit models in the charging direction and the discharging direction of each SOC point to be tested to obtain voltage and current data of the parameter identification test of each SOC point to be tested; and obtaining key parameters of the equivalent circuit model in the charging direction and the discharging direction by using a multiple linear regression method. The SOC estimation method comprises the following steps: determining a battery SOC estimation path according to an open-circuit voltage change rate; and according to the key parameters and the SOC estimation path of the battery, establishing simulation models in the charging direction and the discharging direction, and obtaining SOC estimation values in the charging direction and the discharging direction of the battery by using the simulation models. The accuracy of the equivalent circuit model can be improved, the SOC estimation precision is further improved, and the method is more suitable for actual working conditions.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of battery management, and particularly relates to a parameter identification method, an SOC estimation method and a system of an equivalent circuit model. BACKGROUND

[0002] With the wide application of new energy vehicles and energy storage systems, the battery management system (BMS) plays a key role in battery safety, efficient utilization and life prediction. Among them, the state of charge (SOC) estimation of the battery is one of the core tasks of the BMS, which directly affects the energy management and service life of the battery.

[0003] The existing SOC estimation methods mainly include ampere-hour integration method, open circuit voltage method and model estimation method. Among them, the ampere-hour integration method is widely used due to its simple calculation, but it is easily affected by cumulative error, leading to SOC drift. The open circuit voltage method relies on the static open circuit voltage curve of the battery, and it is difficult to accurately estimate the SOC under dynamic working conditions. The SOC estimation method based on the equivalent circuit model (ECM) can provide more accurate SOC estimation by constructing a dynamic characteristic model of the battery.

[0004] In the model estimation method, Kalman filter (KF) becomes one of the mainstream algorithms for SOC estimation due to its excellent dynamic estimation capability. However, the estimation accuracy of Kalman filter is highly dependent on the accuracy of the equivalent circuit model, and there are many factors affecting the accuracy of the parameters in the existing equivalent circuit model, which is difficult to obtain accurately in real time, and it is difficult to adapt to actual working conditions, which is prone to estimation deviation and lacks real-time adaptive ability. SUMMARY

[0005] The purpose of the present application is to provide a parameter identification method, an SOC estimation method and a system of an equivalent circuit model to improve the accuracy of the equivalent circuit model, thereby improving the SOC estimation accuracy and being more suitable for actual working conditions.

[0006] To achieve the above purpose, the technical scheme is adopted as follows: According to one aspect of the present application, a parameter identification method of an equivalent circuit model is provided, comprising the following steps: constructing equivalent circuit models in charging direction and discharging direction; determining each SOC point to be tested according to the interval width of SOC and the interval of adjacent two SOC points; Conduct mixed pulse tests on the equivalent circuit models of the charging and discharging directions of each SOC point to be tested, and obtain voltage and current data for the parameter identification experiment of each SOC point to be tested; The voltage and current data of the parameter identification experiment at each SOC point to be tested are processed using the multivariate linear regression method to obtain the key parameters of the equivalent circuit model in the charging direction and the discharging direction; the key parameters include open circuit voltage, ohmic internal resistance, polarization internal resistance and polarization capacitance.

[0007] By adopting the above technical solution, the parameters of the equivalent circuit model are identified in the charging direction and the discharging direction, so that different battery parameters can be obtained for the battery charging model and the discharging model, which can not only more accurately describe the battery characteristics, but also reduce the amount of calculation in actual application.

[0008] The hybrid pulse test is used to obtain the battery equivalent parameters at different SOC points, and the multiple linear regression method is used to fit the parameters, which helps to improve the model accuracy.

[0009] Preferably, the equivalent circuit model in the charging direction and the discharging direction is a Rint model, a Thevenin model or a PNGV model.

[0010] Preferably, the equivalent circuit models in the charging direction and the discharging direction are both Thevenin models.

[0011] According to one embodiment of the present invention, a multivariate linear regression method is used to process the voltage and current data of the parameter identification experiment of each SOC point to be tested, and the steps of obtaining the key parameters of the equivalent circuit model in the charging direction and the discharging direction include: Based on the multivariate linear regression fitting principle and circuit relationship formula, the voltage and current data of the parameter identification experiment of each SOC point to be tested are processed using the LINEST function in EXCEL to obtain the key parameters of the equivalent circuit model in the charging direction and discharging direction of each SOC point to be tested.

[0012] According to one embodiment of the present invention, mixed pulse tests are performed on the equivalent circuit models of the charging direction and the discharging direction of each SOC point to be tested, and the steps of obtaining voltage and current data of the parameter identification experiment of each SOC point to be tested include: optimizing the mixed pulse test process by using temperature ranges and rate ranges.

[0013] In this way, the influence of ambient temperature and charge and discharge rate on the equivalent circuit model is eliminated or weakened, further improving the accuracy of the model and making it more adaptable to different working conditions.

[0014] According to one embodiment of the present invention, the step of optimizing the mixed pulse test process using temperature intervals and rate intervals includes: Determine the experimental combination conditions based on the temperature range, rate range and charge and discharge direction; For the equivalent circuit models of the charging direction and discharging direction of each SOC point to be tested, mixed pulse tests are carried out under various experimental combination conditions to obtain voltage and current data.

[0015] According to another aspect of the present invention, there is provided a method for estimating SOC, comprising the following steps: Determine key parameters of the equivalent circuit model of the battery to be evaluated; wherein the key parameters include open circuit voltage, ohmic internal resistance, polarization internal resistance, and polarization capacitance; the key parameters are obtained by: constructing equivalent circuit models in the charging direction and the discharging direction; determining each SOC point to be tested based on the SOC interval width and the interval between two adjacent SOC points; performing mixed pulse tests on the equivalent circuit models in the charging direction and the discharging direction of each SOC point to be tested, respectively, to obtain voltage and current data for parameter identification experiments of each SOC point to be tested; applying a multivariate linear regression method to process the voltage and current data of the parameter identification experiments of each SOC point to be tested, to obtain key parameters of the equivalent circuit models in the charging direction and the discharging direction; Get the open circuit voltage change rate; Determine the battery SOC estimation path based on the open circuit voltage change rate; the battery SOC estimation path includes a Kalman filter path, an ampere-hour integration path, and a Kalman filter-ampere-hour integration hybrid path; A simulation model for the charging direction and the discharging direction is established according to key parameters and the battery SOC estimation path, and the SOC estimation value of the battery in the charging direction and the discharging direction is obtained using the simulation model.

[0016] The above technical solution constructs equivalent circuit models for both the charging and discharging directions, making the SOC estimation process more adaptable to actual operating conditions. Combining the directional equivalent circuit models with a Kalman filter path and / or an ampere-hour integration path that can only be switched improves the accuracy of SOC estimation.

[0017] According to one embodiment of the present invention, the step of determining a battery SOC estimation path according to the open circuit voltage change rate includes: When the open circuit voltage change rate is greater than the preset interval, the Kalman filter path is used; When the open circuit voltage change rate is less than a preset interval, the ampere-hour integration path or the Kalman filter-ampere-hour integration hybrid path is adopted.

[0018] According to one embodiment of the present invention, the step of determining a battery SOC estimation path according to the open circuit voltage change rate includes: Adopt intelligent switching mechanism to dynamically select battery SOC estimation path, The intelligent switching mechanism includes: According to the battery electrical parameters of the current working condition, a random forest algorithm is used to obtain an optimal battery SOC estimation path matching the current working condition; An error feedback control is used to monitor the estimation error and error trend of different battery SOC estimation paths in real time, and the optimal battery SOC estimation path matching the current working condition is adjusted according to the estimation error and error trend of different battery SOC estimation paths.

[0019] According to an embodiment of the present application, in the step of obtaining an optimal battery SOC estimation path matching the current working condition according to the battery electrical parameters of the current working condition by using a random forest algorithm, the step includes: Obtaining battery electrical parameters, including open-circuit voltage change rate, current fluctuation rate, charge-discharge rate, internal resistance change rate and temperature; Inputting the battery electrical parameters into the trained random forest algorithm, and outputting the optimal battery SOC estimation path matching the current working condition.

[0020] According to an embodiment of the present application, in the step of using an error feedback control to monitor the estimation error and error trend of different battery SOC estimation paths in real time, and adjusting the optimal battery SOC estimation path matching the current working condition according to the estimation error and error trend of different battery SOC estimation paths, the step includes: Real-time obtaining of the SOC estimation values obtained by the Kalman filtering path and the ampere-hour integration path, and obtaining the difference between the SOC estimation values obtained by the Kalman filtering path and the ampere-hour integration path, and the error change rate between the SOC estimation values obtained by the Kalman filtering path and the ampere-hour integration path and a preset value; If the error change rate between the SOC estimation value obtained by the Kalman filtering path and the preset value exceeds a first preset range, and the difference between the SOC estimation values obtained by the Kalman filtering path and the ampere-hour integration path exceeds a second preset range, then the ampere-hour integration path is used for battery SOC estimation; If the error change rate between the SOC estimation value obtained by the ampere-hour integration path and the preset value exceeds the first preset range, and the difference between the SOC estimation values obtained by the Kalman filtering path and the ampere-hour integration path exceeds the second preset range, then the Kalman filtering path is used for battery SOC estimation; If the error change rate between the SOC estimation value obtained by the Kalman filtering path and the preset value does not exceed the first preset range, and the difference between the SOC estimation values obtained by the Kalman filtering path and the ampere-hour integration path does not exceed the second preset range, then the Kalman filtering path is used, or the ampere-hour integration path is used, or the Kalman filtering path and the ampere-hour integration path are dynamically weighted and fused for battery SOC estimation.

[0021] According to one embodiment of the present invention, the step of obtaining estimated SOC values ​​of the battery in the charging direction and the discharging direction using the simulation model includes: The steps of obtaining the estimated SOC values ​​of the battery in the charging direction and the discharging direction using the simulation model include: Use the simulation model to obtain the initial SOC estimate of the battery in the charging direction and the discharging direction; The obtained initial SOC estimation values ​​of the battery in the charging direction and the discharging direction are input into the trained feedforward neural network model to obtain the SOC correction terms in the charging direction and the discharging direction of the battery; The sum of the initial SOC estimation value in the battery charging direction and the corresponding SOC correction term is the fused estimated value of SOC.

[0022] According to one aspect of the present invention, there is provided a SOC estimation device based on an equivalent circuit model, comprising: Model building module, used to build equivalent circuit models in charging and discharging directions; A parameter identification module is used to determine each SOC point to be tested based on the SOC interval width and the interval between two adjacent SOC points; perform mixed pulse tests on the equivalent circuit models in the charging direction and the discharging direction of each SOC point to be tested, respectively, to obtain voltage and current data for the parameter identification experiment of each SOC point to be tested; apply a multivariate linear regression method to process the voltage and current data of the parameter identification experiment of each SOC point to be tested, and obtain key parameters of the equivalent circuit model in the charging direction and the discharging direction; the key parameters include open circuit voltage, ohmic internal resistance, polarization internal resistance and polarization capacitance; A path selection module is used to obtain the open circuit voltage change rate; determine the battery SOC estimation path based on the open circuit voltage change rate; the battery SOC estimation path includes a Kalman filter path, an ampere-hour integration path, and a Kalman filter-ampere-hour integration hybrid path; The simulation estimation module is used to establish a simulation model for the charging direction and the discharging direction according to key parameters and the battery SOC estimation path, and use the simulation model to obtain the SOC estimation value of the battery in the charging direction and the discharging direction.

[0023] According to one aspect of the present invention, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the SOC estimation method based on the equivalent circuit model of any one of the above-mentioned embodiments is implemented.

[0024] According to one aspect of the present invention, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, the SOC estimation method based on the equivalent circuit model of any one of the above embodiments is implemented.

[0025] Compared with the prior art, the present invention has at least the following beneficial effects: 1. The present invention identifies the parameters of the equivalent circuit model in the charging direction and the discharging direction, thereby obtaining different battery parameters for the battery charging model and the discharging model, which can not only more accurately describe the battery characteristics but also reduce the amount of calculation in practical applications.

[0026] 2. This invention uses a hybrid pulse power characteristic test (HPPC) to obtain battery equivalent parameters at different SOC points and employs a multivariate linear regression method for parameter fitting. The parameter identification process can be further optimized for different battery types to improve model accuracy. The hybrid pulse test process is optimized using temperature and rate ranges to eliminate or mitigate the effects of ambient temperature and charge / discharge rate on the equivalent circuit model, further improving model accuracy and adapting it to different operating conditions.

[0027] 3. This invention utilizes an intelligent switching mechanism to select between the Kalman filter path and the ampere-hour integration path. This mechanism leverages the strengths of each method and effectively mitigates issues such as divergence in the Kalman filter method and drift in the ampere-hour integration method. This further improves SOC estimation accuracy, enables dynamic adjustment of the SOC estimation method, and enhances its adaptability. This intelligent switching mechanism offers excellent real-time performance and accuracy, making it suitable for power battery management systems subject to dynamic loads or drastic temperature fluctuations. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The accompanying drawings, which constitute part of the present invention, are intended to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are intended to explain the present invention and do not constitute an undue limitation of the present invention. In the accompanying drawings: Figure 1 Schematic diagram of the structure of the Thevenin model, an equivalent circuit model of Example 1 of the present invention; Figure 2 is the open circuit voltage fitting curve in the discharge direction in Example 1 of the present invention; Figure 3 : This is the ohmic internal resistance fitting curve in the discharge direction in Example 1 of the present invention; Figure 4 This is the open circuit voltage fitting curve in the charging direction in Example 1 of the present invention; Figure 5 This is the ohmic internal resistance fitting curve in the charging direction in Example 1 of the present invention; Figure 6 Schematic diagram of the Kalman filter recursive process in Example 1 of the present invention; Figure 7 Schematic diagram of a simulation model for estimating SOC using a Kalman filter path in Example 1 of the present invention; Figure 8This is a schematic diagram of using a feedforward neural network for correction fitting in Example 1 of the present invention; Figure 9 Schematic diagram of the structure of the SOC estimation device in Example 3 of the present invention; Figure 10 This is a schematic diagram of the structure of an electronic device in Example 4 of the present invention.

[0029] Reference numerals: electronic device 100 ; processor 102 ; computer program 103 ; communication bus 104 . DETAILED DESCRIPTION

[0030] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that, in the absence of conflict, the embodiments and features of the embodiments of the present invention can be combined with each other.

[0031] The following detailed description is an exemplary description and is intended to provide further detailed description of the present invention. Unless otherwise indicated, all technical terms used in the present invention have the same meaning as those generally understood by those skilled in the art to which the present invention belongs. The terms used in the present invention are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention.

[0032] Example 1 This embodiment provides a parameter identification method for an equivalent circuit model and an SOC estimation method based on the parameter identification method for the equivalent circuit model.

[0033] The parameter identification method of the equivalent circuit model includes the following steps: Construct equivalent circuit models for charging and discharging directions; Determine each SOC point to be tested according to the SOC interval width and the interval between two adjacent SOC points; Conduct mixed pulse tests on the equivalent circuit models of the charging and discharging directions of each SOC point to be tested, and obtain voltage and current data for the parameter identification experiment of each SOC point to be tested; The voltage and current data of the parameter identification experiment of each SOC point to be tested are processed by the multivariate linear regression method to obtain the key parameters of the equivalent circuit model in the charging direction and the discharging direction; the key parameters include the open circuit voltage U oc 、Ohmic internal resistance R o , polarization internal resistance R p and polarization capacitance C p .

[0034] By adopting the above technical solution, the parameters of the equivalent circuit model are identified in the charging direction and the discharging direction, so that different battery parameters can be obtained for the battery charging model and the discharging model, which can not only more accurately describe the battery characteristics, but also reduce the amount of calculation in actual application.

[0035] Among the three commonly used equivalent circuit models: the Rint model, the Thevenin model, or the PNGV model, the Rint model has poor simulation accuracy, the Thevenin model has higher accuracy under constant current charging conditions, and the PNGV model has higher accuracy under complex conditions with large charge and discharge alternations. However, the PNGV model has many parameters, and the computational complexity of designing an SOC Kalman filter algorithm based on it will be large, which is not conducive to its application in real-time battery management systems. Therefore, in this embodiment, the equivalent circuit models for both the charging and discharging directions are the Thevenin model. This embodiment engineers the Thevenin model, identifies model parameters based on different charging and discharging directions, and uses different battery parameters for the battery charging and discharging models. Ultimately, this can both more accurately describe battery characteristics and reduce the computational complexity in actual applications.

[0036] A hybrid pulse test (HPPC test) was used to obtain the equivalent parameters of batteries at different SOC points for parameter identification experiments. After obtaining the voltage and current data for the parameter identification experiments at each SOC point, the Thevenin model parameters can be identified. Each SOC point corresponds to a set of model parameters. Based on the principle of multivariate linear regression fitting and circuit relationship, the LINEST function in Excel is used to analyze the parameters of each SOC point. The parameter identification principle is shown in Equations (1) and (2). Table 1 shows the identified model parameters for the charging direction, and Table 2 shows the identified model parameters for the discharge direction.

[0037] (1) (2) in, i For the i Sampling moments; U L.i For the i The battery terminal voltage at a moment; I L.i For the i The load current at a moment; R o is the ohmic internal resistance; U oc is the open circuit voltage; R p is the polarization internal resistance; I p.jTo indicate the j The current flowing through the polarization branch in the time period ( R p The current is related to the polarization voltage U p (capacitance C p upper voltage); Δ t is the sampling time interval, in this embodiment, Δ t The unit is seconds; is the time constant, calculated according to formula (3): (3) in, Rp is the polarization internal resistance, Cp For polarized capacitors, this article analyzes the characteristics of the battery used. The value is selected as 4s.

[0038] Table 1 Model parameters of the Thevenin model in the charging direction

[0039] Table 2 Model parameters of the Thevenin model discharge direction

[0040] Furthermore, the temperature range and rate range are used to optimize the mixed pulse test process. Specifically, Determine the experimental combination conditions based on the temperature range, rate range and charge and discharge direction; For the equivalent circuit models of the charging direction and discharging direction of each SOC point to be tested, mixed pulse tests are carried out under various experimental combination conditions to obtain voltage and current data.

[0041] In this embodiment, in order to improve the versatility and estimation accuracy of the model under complex working conditions, a more refined model parameter library was established, and a mixed pulse test was carried out under the following experimental combination conditions: (1) Temperature range: 10°C (low temperature), 25°C (normal temperature), 40°C (high temperature); (2) Charge and discharge rate: 1C and 5C (reflecting standard rate conditions and high rate conditions respectively); (3) SOC point selection: The SOC interval is [0.1-0.9], and an SOC point is set every 0.1 interval; the 9 SOC points are tested in the charging direction and the discharging direction respectively (i.e., the relevant parameters of the charging direction and the discharging direction are measured separately under each set of temperature-rate conditions).

[0042] Finally, we constructed 3 (temperature zones) × 2 (ratios) × 2 (directions) = 12 groups of directional partition parameter models, each group containing 4 main model parameters for 9 SOC points. (1) The influence of temperature on parameters (taking SOC=0.5 as an example): At 10℃, R o Significantly increased (average 22% higher than normal temperature), R p The decrease is obvious, C p decreases, which is reflected in the increase of internal resistance and the slowdown of electrochemical reaction; at 40℃, U oc The average rise is 0.9V, R o It decreased by 12%, the voltage response was more sensitive, but the polarization characteristics were enhanced at high rates.

[0043] (2) The influence of magnification on parameters: From 1C to 5C, R o The average increase was 14%, R p Increased by 18%, especially in the low SOC segment (such as SOC = 0.2), R o The peak value increased by 21%, indicating that the internal resistance sensitive area is particularly affected by the rate; the polarized capacitance C p It drops significantly at 5C, which is manifested as slower polarization recovery at high rates.

[0044] (3) Model fitting accuracy evaluation: When the temperature / rate matching model is used to estimate the SOC for the corresponding working conditions, the average error is within ±1.2%; if the normal temperature 1C model is used across regions (that is, the key parameters identified under normal temperature and 1C rate conditions are directly applied to other working conditions), the error is expanded to ±3.5%~4.0%, and the fluctuation is more significant in the middle SOC range (such as SOC=0.4-0.6).

[0045] The Kalman filter method is used in the non-operating range, where the open-circuit voltage changes rapidly, while other methods such as ampere-hour integration are used to estimate the battery SOC in the operating range, where the open-circuit voltage changes slowly. For lithium iron phosphate batteries, for example, the non-operating range can be set to SOC < 20 and SOC > 90, while the operating range is set to 20 ≤ SOC ≤ 90.

[0046] After obtaining the key parameters of the equivalent circuit model in the charging and discharging directions, the SOC can be estimated using the equivalent circuit model after parameter identification. The specific methods include: Get the open circuit voltage change rate; Determine the battery SOC estimation path based on the open circuit voltage change rate; the battery SOC estimation path includes a Kalman filter path, an ampere-hour integration path, and a Kalman filter-ampere-hour integration hybrid path; A simulation model for the charging direction and the discharging direction is established according to key parameters and the battery SOC estimation path, and the SOC estimation value of the battery in the charging direction and the discharging direction is obtained using the simulation model.

[0047] Generally, when determining the battery SOC estimation path based on the open-circuit voltage rate of change, when the open-circuit voltage rate of change is greater than a preset range, it is considered a non-operating range, and the Kalman filter path is used. When the open-circuit voltage rate of change is less than the preset range, it is considered an operating range, and the ampere-hour integration path or a hybrid Kalman filter-ampere-hour integration path is used. The preset range for the open-circuit voltage rate of change can be adjusted based on experience and actual conditions. In the hybrid Kalman filter-ampere-hour integration path, SOC estimation is performed by weighted fusion of the Kalman filter path and the ampere-hour integration path.

[0048] The ampere-hour integration path, select SOC and the improved Thevenin model upper capacitor C P Voltage U P is the state variable, and the state equation after linearization is shown in Equation (4). The battery terminal voltage obtained by detection is selected as the observation quantity, and the observation equation shown in Equation (5) is obtained.

[0049] (4) (5) Upper middle, k is the current sampling moment, T s is the sampling time, U p is the polarization voltage; C A is the rated capacity of the battery, R p is the polarization resistance, w 1,k 、 w 2,k is the system noise at the current moment, η is the charge and discharge efficiency factor, v k is the observation noise, U l is the load voltage, U oc is the voltage source voltage, I k For the k Current at the moment.

[0050] Using the Kalman filter path to estimate SOC requires that the estimated state SOC and the system output vector must be established in the observation equation U lThis embodiment fits the improved Thevenin model in the charging direction and the discharging direction according to the parameter identification data. P 、U oc The functional relationship with SOC further defines the matrix required for Kalman filtering as follows: (6) (7) (8) (9) in, ; ; Select quadratic and cubic polynomials for fitting comparison respectively, and analyze the more appropriate functional relationship to fit U oc 、R o Functional relationship with SOC.

[0051] Quadratic fitting: discharge direction U oc With SOC, R o The functional relationship with SOC is shown in Equations 10 and 11. The charging direction U oc With SOC, R o The functional relationship with SOC is shown in Equations 12 and 13.

[0052] (10) (11) (12) (13) Among them, U oc_F is the open circuit voltage in the discharge direction; R o_F is the ohmic internal resistance in the discharge direction; U oc_C is the open circuit voltage in the charging direction; R o_C is the ohmic internal resistance in the charging direction.

[0053] Cubic fitting: discharge direction U oc With SOC, R o The functional relationship with SOC is shown in Equations 14 and 15. The charging direction U oc With SOC, R o The functional relationship with SOC is shown in Equations 16 and 17.

[0054] (14) (15) (16) (17) Figure 2 and Figure 4 They are the open circuit voltage and fitting curve of the discharge direction and charge direction models, Figure 3 and Figure 5 The ohmic internal resistance of the model in the discharge direction and the charging direction and its fitting curve are shown in the above figures. oc 、R o The functional relationship between SOC and its quadratic and cubic fitting curves is shown in the figure. The cubic fitting curve is obviously better than the quadratic one, so the cubic function is selected as the U in the observation equation. oc 、R o Functional relationship with SOC.

[0055] The discharge direction is obtained:

[0056] Charging direction:

[0057] With the above data and function preparation, we can get the Kalman filter process shown in equations (19) to (23), where U l,k is the battery terminal voltage calculated according to the model, Y k is the measured voltage of the battery pack.

[0058] (18) (19) (20) (twenty one) (twenty two) (twenty three) The Kalman filter first k -1 moment filtering result X k-1 / k-1 From formula (18), we can get k The predicted value X at time k / k-1 , obtained from formula (19) k The predicted value P of the mean square error matrix at the moment k / k-1 , obtained from formula (20) k The filter gain matrix K at time k , Equation (21) is the battery terminal voltage U calculated according to the model l,k , formula (22) is based on the observed value Y k For the predicted state vector X k / k-1Perform filtering to obtain the filtered value of the state vector, and use Equation (23) to calculate the filtered value of the mean square error matrix. In the above equations, Q and R are the system noise vectors. w ( k ) and the observation noise vector v ( k ), is the interference matrix. The entire filtering algorithm is composed of k =1, 2, ... to loop through equations (18) to (23), making the SOC in the state variable closer to the true value. In each loop, the matrix A k 、B k 、C k The R you need p 、R o 、U oc The parameter values ​​are determined by the parameter identification number k The SOC filter value at time -1 is obtained by looking up the table and interpolating. The recursive algorithm of Kalman filter can be used Figure 6 In the processing of model parameters, this embodiment obtains the parameters of the Thevenin circuit model in two directions: charging and discharging, which is an improvement and optimization of the battery model.

[0059] After identifying the relevant parameters and clarifying the algorithm, the simulation model of charging direction and discharging direction can be established in Simulink, such as Figure 7 As shown. Read the battery current from the actual battery test conditions, such as Figure 7 The "From File" module in the ampere-hour integration module then sends the current signal to the battery module and the ampere-hour integration module. In the ampere-hour integration module, the initial ampere-hour integral value of the ampere-hour integration path is set based on the actual battery conditions. In this embodiment, the battery test conditions are all conducted at room temperature, so the temperature input is 20°C. The ampere-hour integration path in the ampere-hour integration module calculates the SOC based on the known initial value and the charge-discharge efficiency derived from the temperature and current. This SOC value can be considered the true SOC value for a short period of time. This SOC value is input to the battery module, which calculates the battery terminal voltage and current based on the current and SOC value. These values ​​are then fed into the Kalman filter calculation path as input to the Kalman filter module. In the Kalman filter module, the Kalman filter calculation path does not determine the true SOC starting point value. With an arbitrary SOC starting point value, the SOC value is calculated according to the Kalman filter calculation path. The SOC value obtained by the Kalman filter calculation path is then compared with the SOC value obtained by the ampere-hour integration path to verify the reliability of the battery SOC estimation path.

[0060] Tables 3 to 5 show the error statistics when different initial SOC values ​​are set when the above method is used for SOC estimation during the 6A constant current charge and discharge test, capacity consumption test, and capacity balance test.

[0061] Table 3 Error statistics of the simulation model at different SOC initial values ​​in the 6A constant current charge and discharge test

[0062] Table 4 Error statistics of the simulation model at different SOC initial values ​​in the capacity consumption test

[0063] Table 5 Error statistics of the simulation model at different SOC initial values ​​in the capacity balance test

[0064] In actual battery management systems, voltage detection may not be particularly accurate due to factors such as sensors and the actual electromagnetic environment, and errors may occur. To further improve the robustness and accuracy of the SOC estimation method in this embodiment under complex working conditions, based on the above-mentioned Thevenin equivalent circuit model for the charging point direction and the discharging direction, a feedforward neural network is introduced to correct the errors between the initial SOC estimates (i.e., the physical model estimates) obtained by the simulation model and the actual measured values ​​in the charging and discharging directions. Figure 8 .

[0065] In the specific implementation process, first, according to the above method, the simulation model is used to obtain the initial SOC estimation value SOC in the charging direction and the discharging direction of the battery. model ; and obtain the real-time terminal voltage ( U l ), current ( I ), temperature (T), internal resistance change rate (dR o / dt) etc.; The obtained initial SOC estimation value of the battery charging direction and discharging direction and the real-time terminal voltage ( U l ), current ( I ), temperature (T), internal resistance change rate (dR o Parameters such as ΔSOC (ΔSOC) and ΔSOC (Δt) are input into the trained feedforward neural network model to obtain the SOC correction term ΔSOC in the battery charging and discharging directions. The sum of the initial SOC estimate value and the corresponding SOC correction term in the battery charging direction and discharging direction is the fusion estimate SOC fused The final fused estimate is: .

[0066] The feedforward neural network model is trained on three types of test data: constant current, pulse and capacity balance. When training the feedforward neural network model, the training set comes from the test results of the mixed pulse test, including the physical model SOC estimation value (SOC_model), the real-time terminal voltage ( U l ), current ( I ), temperature (T), internal resistance change rate (dR o During training, the weights, biases, and hyperparameters (such as the learning rate) in the feedforward neural network model are optimized. The goal is to minimize the mean squared error (MSE) of the SOC estimate. Verification results show that using the feedforward neural network model to correct the initial SOC estimate can reduce the estimation error by approximately 45% in the mid-range SOC range, keeping the average error within ±1.0%.

[0067] The SOC estimation method based on equivalent circuit parameter identification provided in this embodiment improves the equivalent circuit model by using temperature intervals and rate intervals, and integrates a feedforward neural network model. It has both interpretability and adaptability and is suitable for deployment in a BMS system, especially for operating scenarios with obvious nonlinear fluctuations and frequent high-rate operating conditions.

[0068] Example 2 This embodiment provides a parameter identification method for an equivalent circuit model and an SOC estimation method based on the parameter identification method for the equivalent circuit model. The difference from the first embodiment is that: The steps of determining the battery SOC estimation path based on the open circuit voltage change rate include: Adopt intelligent switching mechanism to dynamically select battery SOC estimation path, The intelligent switching mechanism includes: The random forest algorithm is used to obtain the optimal battery SOC estimation path matching the current operating conditions based on the battery electrical parameters of the current operating conditions. Error feedback control is used to monitor the estimation errors and error trends of different battery SOC estimation paths in real time, and the optimal battery SOC estimation path matching the current operating conditions is adjusted according to the estimation errors and error trends of different battery SOC estimation paths.

[0069] The steps of using the random forest algorithm to obtain the battery SOC estimation path that matches the current operating conditions based on the battery electrical parameters of the current operating conditions include: Obtaining battery electrical parameters, including open circuit voltage change rate, current fluctuation rate, charge and discharge rate, internal resistance change rate, and temperature; The battery electrical parameters are input into the trained random forest algorithm, and the optimal battery SOC estimation path that matches the current operating conditions is output.

[0070] During training, the random forest algorithm inputs the historical battery electrical parameters, environmental parameters, and the battery historical status corresponding to the historical battery electrical parameters and environmental parameters; the battery electrical parameters include voltage change rate, current fluctuation rate, discharge rate, internal resistance change rate, etc., the environmental parameters include temperature, temperature change rate, and the battery historical status includes SOC estimation fluctuations, etc.

[0071] After inputting the battery electrical parameters, the trained random forest algorithm obtains the optimal SOC estimation path under the current operating conditions, which includes classification labels (such as discrete choices) and probability distribution (soft decisions).

[0072] Among them, the classification tags include: 0: prioritize the Kalman filter path (KF); 1: Prioritize the ampere-hour integration path (Ah); 2: Weighted fusion of the two (e.g., KF accounts for 70% and Ah accounts for 30%).

[0073] Probability distributions include: Output the probability of each path (such as KF: 0.8, Ah: 0.1, KF and Ah weighted fusion: 0.1) for use in subsequent dynamic weighted processing.

[0074] Error feedback control is used to monitor the estimation error and error trend of different battery SOC estimation paths in real time, and the optimal battery SOC estimation path matching the current operating conditions is adjusted according to the estimation error and error trend of different battery SOC estimation paths, including: Acquire in real time the SOC estimation values ​​obtained by the Kalman filter path and the ampere-hour integration path, obtain the difference between the SOC estimation values ​​obtained by the Kalman filter path and the ampere-hour integration path, and the error change rate between the SOC estimation values ​​obtained by the Kalman filter path and the ampere-hour integration path and a preset value; If the error change rate between the SOC estimate obtained by the Kalman filter path and the preset value exceeds a first preset range, and the difference between the SOC estimate values ​​obtained by the Kalman filter path and the ampere-hour integration path exceeds a second preset range, the ampere-hour integration path is used to estimate the battery SOC; If the error change rate between the SOC estimate obtained by the ampere-hour integration path and the preset value exceeds a first preset range, and the difference between the SOC estimate values ​​obtained by the Kalman filter path and the ampere-hour integration path exceeds a second preset range, the Kalman filter path is used to estimate the battery SOC; If the error change rate between the SOC estimation value obtained by the Kalman filtering path and the preset value does not exceed the first preset range, and the difference between the SOC estimation values obtained by the Kalman filtering path and the ampere-hour integration path does not exceed the second preset range, the battery SOC is estimated by using the Kalman filtering path, or by using the ampere-hour integration path, or by dynamically weighting and fusing the Kalman filtering path and the ampere-hour integration path.

[0075] In the process of dynamically selecting the battery SOC estimation path by using the intelligent switching mechanism, the random forest algorithm is responsible for pre-judging the most suitable battery SOC estimation path according to the current working condition (such as voltage change, current fluctuation, temperature, etc.), and the error feedback control is used to monitor the estimation difference and error trend of the two methods (the Kalman filtering path and the ampere-hour integration path) in real time. If it is found that the error of the current path determined by the random forest algorithm exceeds the threshold value (for example, the SOC deviation is greater than 2.5%), the system will automatically switch to a more reliable algorithm or dynamically adjust the weights of the two, to ensure the stability and accuracy of the estimation result. The two cooperate to realize high precision and strong adaptability of SOC estimation.

[0076] Embodiment 3 As shown in Figure 9 Based on the same inventive concept as Embodiment 1, this embodiment provides an SOC estimation device based on equivalent circuit model parameter identification, comprising: a model construction module for constructing equivalent circuit models in charging and discharging directions; a parameter identification module for determining each to-be-tested SOC point according to the interval width of the SOC and the interval of the adjacent two SOC points; performing a hybrid pulse test on the equivalent circuit models in the charging and discharging directions for each to-be-tested SOC point, respectively, to obtain the voltage and current data of the parameter identification experiment of each to-be-tested SOC point; and applying a multivariate linear regression method to process the voltage and current data of the parameter identification experiment of each to-be-tested SOC point, to obtain the key parameters of the equivalent circuit models in the charging and discharging directions; the key parameters include open circuit voltage, ohmic resistance, polarization resistance and polarization capacitance; a path selection module for obtaining the open circuit voltage change rate; determining the battery SOC estimation path according to the open circuit voltage change rate; the battery SOC estimation path includes the Kalman filtering path, the ampere-hour integration path and the Kalman filtering-ampere-hour integration mixed path; a simulation estimation module for establishing simulation models in the charging and discharging directions according to the key parameters and the battery SOC estimation path, and obtaining the SOC estimation values of the battery in the charging and discharging directions by using the simulation models.

[0077] Embodiment 4 As shown in Figure 10As shown, the present invention also provides an electronic device 100 for implementing the SOC estimation method based on equivalent circuit model parameter identification of embodiment 1; The electronic device 100 includes a memory 101 , at least one processor 102 , a computer program 103 stored in the memory 101 and executable on the at least one processor 102 , and at least one communication bus 104 .

[0078] The memory 101 may be used to store a computer program 103 . The processor 102 implements the steps of the SOC estimation method in Example 1 by running or executing the computer program stored in the memory 101 and calling data stored in the memory 101 .

[0079] The memory 101 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data (such as audio data) created according to the use of the electronic device 100. In addition, the memory 101 may include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device.

[0080] The at least one processor 102 may be a central processing unit (CPU), or may be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor 102 may be a microprocessor or any conventional processor. The processor 102 is the control center of the electronic device 100 and connects various parts of the entire electronic device 100 using various interfaces and lines.

[0081] The memory 101 in the electronic device 100 stores a plurality of instructions to implement a SOC estimation method based on a parameter identification method of an equivalent circuit model. The processor 102 can execute the plurality of instructions to implement: Construct equivalent circuit models for charging and discharging directions; Determine each to-be-tested SOC point according to the interval width of the SOC and the interval of two adjacent SOC points; Perform a hybrid pulse test on the equivalent circuit model of the charging direction and the discharging direction of each to-be-tested SOC point respectively, and obtain the voltage and current data of the parameter identification experiment of each to-be-tested SOC point; Apply a multivariate linear regression method to process the voltage and current data of the parameter identification experiment of each to-be-tested SOC point, and obtain the key parameters of the equivalent circuit model of the charging direction and the discharging direction; the key parameters include the open-circuit voltage U oc , the ohmic internal resistance R o , the polarization internal resistance R p , and the polarization capacitance C p .

[0082] Embodiment 5 The modules / units integrated by the electronic device 100, if realized in the form of a software function unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on such understanding, all or part of the processes in the above-mentioned embodiment methods of the present application can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer-readable storage medium. The computer program can implement the steps of each method embodiment described above when executed by a processor. The computer program includes computer program code, which can be in the form of source code, object code, an executable file, or some intermediate form. The computer-readable medium can include any entity or device capable of carrying computer program code, a recording medium, a U disk, a mobile hard disk, a magnetic disk, an optical disk, a computer memory, and a read-only memory (ROM).

[0083] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0084] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0085] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0086] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0087] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A parameter identification method for an equivalent circuit model, characterized in that: include: Construct equivalent circuit models for charging and discharging directions; Determine each SOC point to be tested according to the SOC interval width and the interval between two adjacent SOC points; Conduct mixed pulse tests on the equivalent circuit models of the charging and discharging directions of each SOC point to be tested, and obtain voltage and current data for the parameter identification experiment of each SOC point to be tested; The voltage and current data of the parameter identification experiment at each SOC point to be tested are processed using the multivariate linear regression method to obtain the key parameters of the equivalent circuit model in the charging direction and the discharging direction; the key parameters include open circuit voltage, ohmic internal resistance, polarization internal resistance and polarization capacitance.

2. The parameter identification method of the equivalent circuit model according to claim 1, characterized in that: The step of applying the multivariate linear regression method to process the voltage and current data of the parameter identification experiment at each SOC point to be tested to obtain the key parameters of the equivalent circuit model in the charging direction and the discharging direction includes: Based on the multivariate linear regression fitting principle and circuit relationship formula, the voltage and current data of the parameter identification experiment of each SOC point to be tested are processed using the LINEST function in EXCEL to obtain the key parameters of the equivalent circuit model in the charging direction and discharging direction of each SOC point to be tested.

3. The parameter identification method of the equivalent circuit model according to claim 1, characterized in that: The step of performing mixed pulse tests on the equivalent circuit models in the charging direction and the discharging direction of each SOC point to be tested, and obtaining voltage and current data for the parameter identification experiment of each SOC point to be tested, includes: The temperature range and rate range are used to optimize the mixed pulse test process.

4. The parameter identification method of the equivalent circuit model according to claim 3, characterized in that: The step of optimizing the mixed pulse test process using the temperature range and the rate range includes: Determine the experimental combination conditions based on the temperature range, rate range and charge and discharge direction; For the equivalent circuit models of the charging direction and discharging direction of each SOC point to be tested, mixed pulse tests are carried out under various experimental combination conditions to obtain voltage and current data.

5. A SOC estimation method, characterized in that: The following steps are involved: Determine key parameters of the equivalent circuit model of the battery to be evaluated; wherein the key parameters include open circuit voltage, ohmic internal resistance, polarization internal resistance, and polarization capacitance; the key parameters are obtained by: constructing equivalent circuit models in the charging direction and the discharging direction; determining each SOC point to be tested based on the SOC interval width and the interval between two adjacent SOC points; performing mixed pulse tests on the equivalent circuit models in the charging direction and the discharging direction of each SOC point to be tested, respectively, to obtain voltage and current data for parameter identification experiments of each SOC point to be tested; applying a multivariate linear regression method to process the voltage and current data of the parameter identification experiments of each SOC point to be tested, to obtain key parameters of the equivalent circuit models in the charging direction and the discharging direction; Get the open circuit voltage change rate; Determine the battery SOC estimation path based on the open circuit voltage change rate; The battery SOC estimation path includes the Kalman filter path, the ampere-hour integration path, and the Kalman filter-ampere-hour integration hybrid path; A simulation model for the charging direction and the discharging direction is established according to key parameters and the battery SOC estimation path, and the SOC estimation value of the battery in the charging direction and the discharging direction is obtained using the simulation model.

6. The SOC estimation method according to claim 5, characterized in that: The step of determining the battery SOC estimation path according to the open circuit voltage change rate includes: When the open circuit voltage change rate is greater than the preset interval, the Kalman filter path is used; When the open circuit voltage change rate is less than a preset interval, the ampere-hour integration path or the Kalman filter-ampere-hour integration hybrid path is adopted.

7. The SOC estimation method according to claim 5, characterized in that: The step of determining the battery SOC estimation path according to the open circuit voltage change rate includes: Adopt intelligent switching mechanism to dynamically select battery SOC estimation path, The intelligent switching mechanism includes: The random forest algorithm is used to obtain the optimal battery SOC estimation path matching the current operating conditions based on the battery electrical parameters of the current operating conditions. Error feedback control is used to monitor the estimation errors and error trends of different battery SOC estimation paths in real time, and the optimal battery SOC estimation path matching the current operating conditions is adjusted according to the estimation errors and error trends of different battery SOC estimation paths.

8. The SOC estimation method according to claim 5, characterized in that: The step of obtaining the estimated SOC values ​​of the battery in the charging direction and the discharging direction using the simulation model includes: Use the simulation model to obtain the initial SOC estimate of the battery in the charging direction and the discharging direction; The obtained initial SOC estimation values ​​of the battery in the charging direction and the discharging direction are input into the trained feedforward neural network model to obtain the SOC correction terms in the charging direction and the discharging direction of the battery; The sum of the initial SOC estimation value in the battery charging direction and the corresponding SOC correction term is the fused estimated value of SOC.

9. An SOC estimation device based on an equivalent circuit model, characterized in that: include: Model building module, used to build equivalent circuit models in charging and discharging directions; A parameter identification module is used to determine each SOC point to be tested based on the SOC interval width and the interval between two adjacent SOC points; perform mixed pulse tests on the equivalent circuit models in the charging direction and the discharging direction of each SOC point to be tested, respectively, to obtain voltage and current data for the parameter identification experiment of each SOC point to be tested; apply a multivariate linear regression method to process the voltage and current data of the parameter identification experiment of each SOC point to be tested, and obtain key parameters of the equivalent circuit model in the charging direction and the discharging direction; the key parameters include open circuit voltage, ohmic internal resistance, polarization internal resistance and polarization capacitance; A path selection module is used to obtain the open circuit voltage change rate; determine the battery SOC estimation path based on the open circuit voltage change rate; the battery SOC estimation path includes a Kalman filter path, an ampere-hour integration path, and a Kalman filter-ampere-hour integration hybrid path; The simulation estimation module is used to establish a simulation model for the charging direction and the discharging direction according to key parameters and the battery SOC estimation path, and use the simulation model to obtain the SOC estimation value of the battery in the charging direction and the discharging direction.

10. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the processor is configured to execute a computer program stored in the memory to implement the SOC estimation method according to any one of claims 5 to 8.

Images