Variable-order lithium ion battery model and construction method

Abstract

The invention discloses a construction method of an order-variable lithium ion battery model, and the method comprises the following steps: S1, carrying out a hybrid power pulse characteristic experiment on a battery, and obtaining model parameters; s2, constructing a structural circuit module according to the model parameters; s3, performing a dynamic pressure test on the battery to obtain DST experimental data, and dividing SOC regions; s4, calculating the order of each SOC region according to DST experimental data and an HQIC criterion; s5, constructing a selection module according to the order of each SOC region; and S6, constructing an equivalent circuit model according to the structural circuit module and the selection module, and obtaining a variable-order lithium ion battery model. According to the method, the lithium ion battery model with the variable order is realized, and the model is further selected through the HQIC information criterion and by adding the contrast factor, so that the precision and the complexity of the lithium ion battery model are better balanced.

Description

Technical Field

[0001] The present invention belongs to the technical field of lithium-ion battery equivalent circuit models, and in particular relates to a lithium-ion battery model with a variable order and a construction method thereof. Background Art

[0002] At present, it is often necessary to use equivalent circuit models to analyze lithium-ion batteries. The common equivalent circuit models are Rint model, Thevenin model and PNGV model. Among them, the more common one is to add one or two RC loops to the Thevenin model to form a second-order RC or third-order RC equivalent circuit model. Although these two models have greatly improved the accuracy compared with other previous models, the second-order RC model has a relatively high terminal voltage error in the area where the battery voltage changes drastically; the third-order RC model has a large amount of calculation, which reduces the response rate of the model.

[0003] The current battery model forms are: 1. The whole process adopts the second-order RC model, such as Figure 7 As shown in the figure, the battery model currently used is more common, in which R1 and C1 represent the electrochemical polarization internal resistance and electrochemical polarization capacitance respectively, and R2 and C2 represent the concentration polarization internal resistance and concentration polarization capacitance respectively, which better reflect the changes inside the battery. However, in areas where the battery voltage changes drastically, such as when the SOC is high or low, the accuracy of the second-order RC model changes greatly; 2. There are also third-order RC models used throughout the process, such as Figure 8 As shown in the figure, by adding another RC loop to further represent the internal reaction of the battery, it is easier to get closer to the actual situation of the battery. However, due to the large amount of calculation of the third-order RC model, if the third-order RC model is used throughout the process, it will increase the chip calculation burden, which is not conducive to improving the response speed of the controller; 3. The variable-order RC equivalent circuit model based on the AIC criterion proposes that the third-order RC model should be selected for the characteristics of lithium batteries with rapid voltage changes at the beginning and end of charging and discharging, and the second-order RC model should be selected for the characteristics of stable voltage changes during the platform period of SOC from 20% to 80%; However, the penalty of AIC is relatively light, and the formula is AIC = -2log(L) + 2k, which makes it easy to select a complex model, which may lead to overfitting. ; It is also more inclined to select a complex model. 4. Based on the BIC criterion, the BIC values ​​of the first-order RC model and the second-order RC model are obtained respectively, and then the order of the first-order RC model and the second-order RC model is selected. Finally, it is decided that when the SOC value is high and low, the error of the first-order RC model is large, so the second-order RC model is selected; when the SOC value is medium, the fitting accuracy of the two models does not change much, but the complexity of the second-order model is higher, so the first-order model is selected; but the penalty of BIC is heavier, the formula is BIC = -2log(L) + klog(n), and it is more inclined to simple models in the case of large samples, which may lead to underfitting.

[0004] For example, the Chinese patent with announcement number CN103926538B discloses a variable-order RC equivalent circuit model based on the AIC criterion and an implementation method. The model selects the RC model as the basic model, fully weighs the complexity and accuracy of the model at different SOCs based on the AIC criterion, determines the optimal RC order at each SOC, and implements the variable-order RC model by controlling the switching switch. However, this patent is a variable-order RC equivalent circuit model based on the AIC criterion. Since the penalty of AIC is relatively light, the formula is AIC=-2ln(L)+2k, it is easy to select a complex model, which may lead to overfitting.

[0005] For example, the Chinese patent with the announcement number CN114200321B discloses a modeling method for a variable-order equivalent circuit model of a lithium-ion battery, including the steps of: establishing a first-order RC model and a second-order RC model of a lithium battery; calculating the terminal voltages of the first-order RC model and the second-order RC model respectively; solving the terminal voltages of the first-order RC model and the second-order RC model using the Bayesian order criterion, and obtaining the BIC values ​​of the first-order RC model and the second-order RC model respectively; taking the RC model with a smaller BIC value as the variable-order equivalent circuit model of the lithium-ion battery. By analyzing the variation law of the frequency band in the impedance spectrum with the battery state of charge and temperature, the Bayesian order criterion is used to select the model order. However, this patent only includes the first-order and second-order changes. At the same time, its variable-order RC equivalent circuit model based on the BIC criterion has a heavy penalty for BIC, and the formula is BIC = -2ln(L) + kln(n). In the case of large samples, it is more inclined to a simple model, which may lead to underfitting. Summary of the invention

[0006] In order to solve the above technical problems and improve the estimation accuracy of the lithium-ion battery equivalent circuit model, the present invention provides a variable-order lithium-ion battery model and a construction method.

[0007] The present invention is achieved through the following technical solutions.

[0008] The present invention provides a method for constructing a variable-order lithium-ion battery model, comprising the following steps:

[0009] S1. Perform a hybrid power pulse characteristic experiment on the battery to obtain model parameters, where the model parameters include first-order RC circuit parameters, second-order RC circuit parameters, and third-order RC circuit parameters;

[0010] S2, constructing a structural circuit module according to the model parameters;

[0011] S3, performing a dynamic pressure test on the battery, obtaining DST experimental data, and dividing the SOC area, wherein the DST experimental data includes terminal voltage and parameter voltage;

[0012] S4. Calculate the order of each SOC region based on DST experimental data and HQIC criteria;

[0013] S5, constructing a selection module according to the order of each SOC region;

[0014] S6. Construct an equivalent circuit model according to the structural circuit module and the selection module to obtain a lithium-ion battery model with a variable order.

[0015] Preferably, the step S1, performing a hybrid power pulse characteristic experiment on the battery to obtain model parameters comprises the following steps:

[0016] S11, performing a hybrid power pulse characteristic experiment on the battery to obtain parameter identification experimental data;

[0017] S12, fitting the parameter identification experimental data to obtain an open circuit voltage curve and parameter fitting values;

[0018] S13, fitting the open circuit voltage curve through a simulation tool to obtain the open circuit voltage;

[0019] S14. Identify the open circuit voltage and the parameter fitting value through a parameter identification estimator to obtain model parameters.

[0020] Preferably, the simulation tool supports MATLAB.

[0021] Preferably, performing a dynamic stress test on the battery to obtain DST experimental data comprises the following steps:

[0022] Perform dynamic pressure test on the battery to obtain the terminal voltage and corresponding current;

[0023] The corresponding current is input into the structural circuit module to obtain the parameter voltage.

[0024] Preferably, the step S4, calculating the order of each SOC region according to the DST experimental data and the HQIC criterion, comprises the following steps:

[0025] S41, calculating the residual sum of squares of each SOC region according to the terminal voltage and the parameter voltage;

[0026] S42, calculating the HQIC value of each SOC area according to the residual square sum of each SOC area;

[0027] S43, obtaining a comparison factor of each SOC area according to the HQIC value of each SOC area;

[0028] S44. Determine and obtain the order of each SOC region according to the HQIC value and the comparison factor of each SOC region.

[0029] Preferably, the step S41, calculating the residual sum of squares according to the terminal voltage and the parameter voltage, comprises the following steps:

[0030] S411, calculating a voltage error according to the terminal voltage and the parameter voltage;

[0031] S412: Calculate the residual square sum of each SOC region according to the voltage error.

[0032] Preferably, the step S44, determining and obtaining the order of each SOC region according to the HQIC value and the comparison factor of each SOC region, comprises the following steps:

[0033] S441, setting a battery stable interval and a contrast factor reference value, and selecting an SOC region as the current region;

[0034] S442, judging whether the HQIC value of the first-order RC model in the current region is the smallest, if so, selecting the first-order RC model in the current region, otherwise proceeding to step S442;

[0035] S443, judging whether the HQIC value of the second-order RC model of the current region is the smallest, if so, the second-order RC model is selected for the current region, otherwise, proceeding to step S443;

[0036] S444, determining whether the current region is within the battery stable range and the contrast factor is less than the contrast factor reference value, if so, selecting the model corresponding to the second smallest HQIC value for the current region, otherwise selecting the third-order RC model for the current region;

[0037] S445. Select an unselected SOC area as the current area, and return to step S442 until all SOC areas have been selected.

[0038] A lithium-ion battery model with a variable order, constructed by the above construction method, comprises a structural circuit module and a selection module, wherein the order of the structural circuit module is variable, and the selection module controls the order of the structural circuit module;

[0039] The structural circuit module includes a first-order RC circuit, a second-order RC circuit, a third-order RC circuit, a second-order switch and a third-order switch. The first-order RC circuit is connected to the second-order RC circuit, the second-order RC circuit is connected to the third-order RC circuit, the second-order switch controls the second-order RC circuit, and the third-order switch controls the third-order RC circuit.

[0040] Preferably, the selection module controls the order of the structural circuit module according to the SOC area.

[0041] Preferably, it further comprises a calculation module, wherein the calculation module is used to set parameters of the selection module;

[0042] The calculation module includes a region division submodule, an HQIC calculation submodule, a contrast factor calculation submodule and a judgment submodule;

[0043] The area division submodule is used to divide the SOC area;

[0044] The HQIC calculation submodule is used to calculate the HQIC value of each SOC area in the structural circuit module;

[0045] The contrast factor calculation submodule is used to calculate the contrast factor of each SOC area according to the HQIC value of each SOC area;

[0046] The judgment submodule determines and obtains the order of each SOC region according to the HQIC value and the comparison factor of each SOC region, and sets the parameters of the selection module according to the order of each SOC region.

[0047] The beneficial effects of the present invention are:

[0048] A variable-order lithium-ion battery model is implemented. The HQIC information criterion is used and a comparison factor is added to further select the model, which reduces the amount of calculation of the lithium-ion battery model and improves the accuracy, achieving a better balance between accuracy and complexity. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 is a flow chart of a method provided by an embodiment of the present invention;

[0050] Figure 2 is a schematic diagram of a structural circuit module provided by an embodiment of the present invention;

[0051] Figure 3 is a schematic diagram of a structural circuit module provided by an embodiment of the present invention;

[0052] Figure 4 It is a schematic diagram of the HPPC fitting of various order models provided by an embodiment of the present invention;

[0053] Figure 5 is a schematic diagram of HQIC values ​​of various SOC regions provided by an embodiment of the present invention;

[0054] Figure 6 is a schematic diagram of simulation verification results provided by an embodiment of the present invention;

[0055] Figure 7 It is a schematic diagram of a second-order RC battery model of the prior art;

[0056] Figure 8 It is a schematic diagram of a third-order RC battery model in the prior art. DETAILED DESCRIPTION

[0057] The technical solution of the present invention is further described below, but the scope of protection claimed is not limited to the description.

[0058] like Figure 1 As shown, a method for constructing a lithium-ion battery model with a variable order includes the following steps:

[0059] S1. Perform a hybrid power pulse characteristic experiment on the battery to obtain model parameters, where the model parameters include first-order RC circuit parameters, second-order RC circuit parameters, and third-order RC circuit parameters;

[0060] HPPC is the hybrid power pulse characteristic experiment.

[0061] The first-order RC circuit parameters are the resistance and capacitance values ​​in the first-order RC circuit, the second-order RC circuit parameters are the resistance and capacitance values ​​in the second-order RC circuit, and the third-order RC circuit parameters are the resistance and capacitance values ​​in the third-order RC circuit.

[0062] S1, performing a hybrid power pulse characteristic experiment on the battery to obtain model parameters includes the following steps:

[0063] S11, performing a hybrid power pulse characteristic experiment on the battery to obtain parameter identification experimental data;

[0064] S12, fitting the parameter identification experimental data to obtain an open circuit voltage curve and parameter fitting values;

[0065] S13, fitting the open circuit voltage curve through a simulation tool to obtain the open circuit voltage;

[0066] The simulation tool supports MATLAB.

[0067] In this embodiment, the Curve fitting Toolbox tool of MATLAB is used to fit the open circuit voltage curve to obtain the open circuit voltage expression:

[0068] OCV(SOC)=-52.07×SOC^7+195×SOC^6-286.7×SOC^5

[0069] +206.4×SOC^4-72.17×SOC^3+9.285×SOC^2

[0070] +1.191×SOC+3.208

[0071] Among them, OCV is the open circuit voltage and SOC is the remaining battery power state.

[0072] S14. Identify the open circuit voltage and the parameter fitting value through a parameter identification estimator to obtain model parameters.

[0073] In this embodiment, the first-order RC circuit parameters include R0, R1, and C1, the second-order RC circuit parameters include R2 and C2, and the third-order circuit parameters include R3 and C3.

[0074] The obtained model parameter fitting degree is as follows Figure 4 shown.

[0075] S2, constructing a structural circuit module according to the model parameters;

[0076] In this embodiment, the structure circuit module constructed is as follows Figure 2 shown.

[0077] S3, performing a dynamic pressure test on the battery, obtaining DST experimental data, and dividing the SOC area, wherein the DST experimental data includes terminal voltage and parameter voltage;

[0078] The main reason for choosing the DST condition is that it is mainly used to evaluate the comprehensive performance of the battery under actual use conditions. It simulates various dynamic load changes that the vehicle may encounter during driving, such as acceleration, deceleration, idling and other conditions, so it can better reflect the actual experimental conditions of the battery.

[0079] In this embodiment, the SOC area is divided into 10 sections, namely 0-10%, 10-20%, 20-30%, 30-40%, 40-50%, 50-60%, 60-70%, 70-80%, 80-90%, 90-100%. The total number of samples in the experiment is 24478, and the number of samples in each SOC area is about 2400, which belongs to the category of large samples.

[0080] The method of performing a dynamic pressure test on the battery to obtain DST experimental data includes the following steps:

[0081] Perform dynamic pressure test on the battery to obtain the terminal voltage and corresponding current;

[0082] The corresponding current is input into the structural circuit module to obtain the parameter voltage.

[0083] S4. Calculate the order of each SOC region based on DST experimental data and HQIC criteria;

[0084] HQIC information criterion was proposed by Edward Hannan and Brian Quinn in 1979. HQIC criterion is similar to Akaike information criterion and Bayesian information criterion, namely AIC and BIC. The main difference between HQIC and AIC and BIC is the form of penalty term. The penalty term of HQIC is 2kln(ln(n)), which is between 2k of AIC and kln(n) of BIC when the sample size is large.

[0085] The step S4, calculating the order of each SOC region according to the DST experimental data and the HQIC criterion, comprises the following steps:

[0086] S41, calculating the residual sum of squares of each SOC region according to the terminal voltage and the parameter voltage;

[0087] The step S41, calculating the residual sum of squares according to the terminal voltage and the parameter voltage, comprises the following steps:

[0088] S411, calculating a voltage error according to the terminal voltage and the parameter voltage;

[0089] S412: Calculate the residual square sum of each SOC region according to the voltage error.

[0090] S42, calculating the HQIC value of each SOC area according to the residual square sum of each SOC area;

[0091] In this embodiment, HQIC can be expressed as:

[0092] HQIC=n×ln(RSS / n)+2×k×ln(ln(n))

[0093] Among them, n is the number of samples, k is the number of parameters of the model, and RSS is the residual sum of squares.

[0094] The HQIC result is as follows: Figure 5 shown.

[0095] S43, obtaining a comparison factor of each SOC area according to the HQIC value of each SOC area;

[0096] In this embodiment, the contrast factor expression is:

[0097] Q=|HQIC2-HQIC3|

[0098] Among them, HQIC2 is the second smallest HQIC value in the SOC area, and HQIC3 is the third smallest HQIC value in the SOC area.

[0099] S44. Determine and obtain the order of each SOC region according to the HQIC value and the comparison factor of each SOC region.

[0100] The step S44, determining and obtaining the order of each SOC region according to the HQIC value and the comparison factor of each SOC region, comprises the following steps:

[0101] S441, setting a battery stable interval and a contrast factor reference value, and selecting an SOC region as the current region;

[0102] In this embodiment, the battery stable interval is set to 20-80%, and the contrast factor reference value is set to 260.

[0103] S442, judging whether the HQIC value of the first-order RC model in the current region is the smallest, if so, selecting the first-order RC model in the current region, otherwise proceeding to step S442;

[0104] S443, judging whether the HQIC value of the second-order RC model of the current region is the smallest, if so, the second-order RC model is selected for the current region, otherwise, proceeding to step S443;

[0105] S444, determining whether the current region is within the battery stable range and the contrast factor is less than the contrast factor reference value, if so, selecting the model corresponding to the second smallest HQIC value for the current region, otherwise selecting the third-order RC model for the current region;

[0106] In the stable range of the battery, the internal reaction of the battery is relatively stable, and a lower-order battery model can be used to express the internal reaction, thereby reducing the computational complexity.

[0107] The contrast factor further balances the computational effort and accuracy of the model.

[0108] S445. Select an unselected SOC area as the current area, and return to step S442 until all SOC areas have been selected.

[0109] In this embodiment, the order of each SOC region obtained is shown in the following table:

[0110]

[0111] S5, constructing a selection module according to the order of each SOC region;

[0112] S6. Construct an equivalent circuit model according to the structural circuit module and the selection module to obtain a lithium-ion battery model with a variable order.

[0113] like Figure 3 As shown, a lithium-ion battery model with a variable order is constructed by the above construction method, including a structural circuit module and a selection module, the order of the structural circuit module is variable, and the selection module controls the order of the structural circuit module;

[0114] The structural circuit module includes a first-order RC circuit, a second-order RC circuit, a third-order RC circuit, a second-order switch and a third-order switch. The first-order RC circuit is connected to the second-order RC circuit, the second-order RC circuit is connected to the third-order RC circuit, the second-order switch controls the second-order RC circuit, and the third-order switch controls the third-order RC circuit.

[0115] like Figure 2As shown, in this embodiment, R0, R1 and C1 are first-order RC circuits, R2 and C2 are second-order RC circuits, R3 and C3 are third-order RC circuits, S1 is a second-order switch, and S2 is a third-order switch. In this embodiment, when S1S2 are both closed, the lithium-ion battery model is a first-order RC model; when S1 is disconnected and S2 is closed, it is a second-order RC model; when S1S2 are both disconnected, it is a third-order RC model.

[0116] The selection module controls the order of the structure circuit module according to the SOC area.

[0117] Also included is a calculation module, the calculation module is used to set parameters of the selection module;

[0118] The calculation module includes a region division submodule, an HQIC calculation submodule, a contrast factor calculation submodule and a judgment submodule;

[0119] The area division submodule is used to divide the SOC area;

[0120] The HQIC calculation submodule is used to calculate the HQIC value of each SOC area in the structural circuit module;

[0121] The contrast factor calculation submodule is used to calculate the contrast factor of each SOC area according to the HQIC value of each SOC area;

[0122] The judgment submodule determines and obtains the order of each SOC region according to the HQIC value and the comparison factor of each SOC region, and sets the parameters of the selection module according to the order of each SOC region.

[0123] In this embodiment, a simulation verification is finally performed by building a variable-order lithium-ion battery model in simulink and verifying it using the DST working condition. The verification results are as follows: Figure 6 shown.

[0124] The analysis and verification results are as follows:

[0125]

[0126] The verification results show that the variable-order lithium-ion battery model of the present invention is superior to the first-order RC model, the second-order RC model and the third-order RC model.

[0127] The present invention realizes a lithium-ion battery model with a variable order, and further selects the model by using the HQIC information criterion and adding a contrast factor, so that the calculation amount of the lithium-ion battery model is reduced and the accuracy is higher, and the accuracy and complexity are better balanced.

Claims

1. A method for constructing a lithium-ion battery model with a variable order, characterized in that: The following steps are involved: S1. Perform a hybrid power pulse characteristic experiment on the battery to obtain model parameters, where the model parameters include first-order RC circuit parameters, second-order RC circuit parameters, and third-order RC circuit parameters; S2, constructing a structural circuit module according to the model parameters; S3, performing a dynamic pressure test on the battery, obtaining DST experimental data, and dividing the SOC area, wherein the DST experimental data includes terminal voltage and parameter voltage; S4. Calculate the order of each SOC region based on DST experimental data and HQIC criteria; S5, constructing a selection module according to the order of each SOC region; S6. Construct an equivalent circuit model according to the structural circuit module and the selection module to obtain a lithium-ion battery model with a variable order.

2. The construction method according to claim 1, characterized in that: S1, performing a hybrid power pulse characteristic experiment on the battery to obtain model parameters includes the following steps: S11, performing a hybrid power pulse characteristic experiment on the battery to obtain parameter identification experimental data; S12, fitting the parameter identification experimental data to obtain an open circuit voltage curve and parameter fitting values; S13, fitting the open circuit voltage curve through a simulation tool to obtain the open circuit voltage; S14. Identify the open circuit voltage and the parameter fitting value through a parameter identification estimator to obtain model parameters.

3. The construction method according to claim 2, characterized in that: The simulation tool supports MATLAB.

4. The construction method according to claim 1, characterized in that: The method of performing a dynamic pressure test on the battery to obtain DST experimental data includes the following steps: Perform dynamic pressure test on the battery to obtain the terminal voltage and corresponding current; The corresponding current is input into the structural circuit module to obtain the parameter voltage.

5. The construction method according to claim 1, characterized in that: The step S4, calculating the order of each SOC region according to the DST experimental data and the HQIC criterion, comprises the following steps: S41, calculating the residual sum of squares of each SOC region according to the terminal voltage and the parameter voltage; S42, calculating the HQIC value of each SOC area according to the residual square sum of each SOC area; S43, obtaining a comparison factor of each SOC area according to the HQIC value of each SOC area; S44. Determine and obtain the order of each SOC region according to the HQIC value and the comparison factor of each SOC region.

6. The construction method according to claim 5, characterized in that: The step S41, calculating the residual sum of squares according to the terminal voltage and the parameter voltage, comprises the following steps: S411, calculating a voltage error according to the terminal voltage and the parameter voltage; S412: Calculate the residual square sum of each SOC region according to the voltage error.

7. The construction method according to claim 5, characterized in that: The step S44, determining and obtaining the order of each SOC region according to the HQIC value and the comparison factor of each SOC region, comprises the following steps: S441, setting a battery stable interval and a contrast factor reference value, and selecting an SOC region as the current region; S442, judging whether the HQIC value of the first-order RC model in the current region is the smallest, if so, selecting the first-order RC model in the current region, otherwise proceeding to step S442; S443, judging whether the HQIC value of the second-order RC model of the current region is the smallest, if so, the second-order RC model is selected for the current region, otherwise, proceeding to step S443; S444, determining whether the current region is within the battery stable range and the contrast factor is less than the contrast factor reference value, if so, selecting the model corresponding to the second smallest HQIC value for the current region, otherwise selecting the third-order RC model for the current region; S445. Select an unselected SOC area as the current area, and return to step S442 until all SOC areas have been selected.

8. A variable-order lithium-ion battery model, characterized in that: Constructed by the construction method according to any one of claims 1 to 7, comprising a structural circuit module and a selection module, wherein the order of the structural circuit module is variable, and the selection module controls the order of the structural circuit module; The structural circuit module includes a first-order RC circuit, a second-order RC circuit, a third-order RC circuit, a second-order switch and a third-order switch. The first-order RC circuit is connected to the second-order RC circuit, the second-order RC circuit is connected to the third-order RC circuit, the second-order switch controls the second-order RC circuit, and the third-order switch controls the third-order RC circuit.

9. The lithium-ion battery model according to claim 8, characterized in that: The selection module controls the order of the structure circuit module according to the SOC area.

10. The lithium-ion battery model according to claim 9, characterized in that: Also includes a calculation module, the calculation module is used to set the parameters of the selection module; The calculation module includes a region division submodule, an HQIC calculation submodule, a contrast factor calculation submodule and a judgment submodule; The area division submodule is used to divide the SOC area; The HQIC calculation submodule is used to calculate the HQIC value of each SOC area in the structural circuit module; The contrast factor calculation submodule is used to calculate the contrast factor of each SOC area according to the HQIC value of each SOC area; The judgment submodule determines and obtains the order of each SOC region according to the HQIC value and the comparison factor of each SOC region, and sets the parameters of the selection module according to the order of each SOC region.

Images