A method to improve battery model accuracy based on variable time domain

Abstract

The present invention discloses a method for improving battery model accuracy based on a variable time domain. Based on the relationship between parameter identification accuracy and the variations in different temperatures and time domains, a coupling relationship between temperature and the optimal time domain is established. The optimal time domain at a certain temperature is determined, replacing the fixed time domain for model parameter identification. The identified model parameters are then substituted into an established basic equivalent circuit model for simulation. Beneficial Effects: The present invention solves the problems of unstable and inaccurate identification accuracy associated with model parameter identification using a fixed time domain, thereby improving the accuracy of the battery model.

Description

Technical Field

[0001] The present invention relates to a method for improving the accuracy of a battery model, and in particular to a method for improving the accuracy of a battery model based on a variable time domain, belonging to the technical field of power batteries. Background Art

[0002] With the rapid growth of the global economy and the rapid development of the automotive industry, the global car population continues to increase. This massive car population has also led to serious fuel shortages and environmental pollution. Therefore, it is imperative to vigorously develop new energy vehicles as an alternative to traditional fuel vehicles to address the current environmental and resource challenges. Pure electric vehicles offer advantages such as high power, low energy consumption, and low pollution. Vigorously developing pure electric vehicles is an effective means of addressing these challenges.

[0003] Lithium-ion batteries have become the mainstream power source for electric vehicles due to their high specific energy density, high specific power, lack of memory effect, and long lifespan. However, automotive battery packs typically consist of hundreds or even thousands of individual cells connected in series and parallel. Variations between these cells can significantly impact the lifespan and safety of the battery pack. To ensure the safety of electric vehicle battery packs, a battery management system (BMS) is required to assess the battery pack's state. Model-driven state estimation methods are a common approach, offering closed-loop estimation and a good balance between computational cost and estimation accuracy. The core of this approach is the establishment of an accurate model and precise model parameters. Therefore, finding an appropriate battery model to describe and accurately identifying battery model parameters has become a key research priority.

[0004] Battery model-based parameter identification is primarily categorized into offline and online methods. Compared to offline methods, online methods offer real-time performance, high accuracy, and robustness, particularly when operating conditions of electric vehicles vary. While online parameter identification has become a mainstream research area, it is often based on optimization algorithms such as recursive least squares (RLS). The accuracy of parameter identification is limited by the number of parameters to be identified and the range within which they must be identified. Simultaneous identification of multiple parameters or the over-range nature of the initial parameters can easily lead to deviations from the correct values ​​of the identified parameters or divergence of the identification algorithm.

[0005] Offline parameter identification, especially the method of offline parameter identification of battery model parameters through hybrid pulse power characteristic test (HPPC) test, can quickly explore the variation law of battery parameters under specific working conditions and its correlation with battery performance. It can also provide a reference for online parameter identification. Therefore, offline parameter identification is still an indispensable method for battery modeling and characteristic research.

[0006] Among existing off-line parameter identification technologies, most use a fixed time domain to identify model polarization parameters. This fixed time domain identification of model parameters has problems such as unstable identification accuracy and poor identification accuracy. Summary of the Invention

[0007] Purpose of the Invention: To address the problems existing in the prior art, the present invention provides a method for improving battery model accuracy based on a variable time domain. Based on the variation between parameter identification accuracy and different temperatures and time domains, the present invention establishes a coupled relationship between temperature and the optimal time domain. This method determines the optimal time domain at a given temperature, replacing the fixed time domain for model parameter identification, thereby improving battery model accuracy.

[0008] Technical solution: A method for improving battery model accuracy based on a variable time domain, comprising the following steps:

[0009] Step 1: Build a basic equivalent circuit model and determine the parameters of the model to be identified based on the basic equivalent circuit model;

[0010] Step 2: Perform a pulse power test on the battery at different set temperatures and at intervals of a certain percentage of the state of charge (SOC) to identify the model parameters to be identified and establish a relationship between the model parameters and the state of charge (SOC);

[0011] Step 3: Select a discharge phase of a set time in each pulse power test of step 2, intercept different time domains in the phase to perform model parameter identification, and establish the relationship between the model parameters and the state of charge (SOC);

[0012] Step 4: Based on the basic equivalent circuit model established in step 1 and the battery model parameters obtained in steps 2 and 3, calculate the error between the simulation and experimental values ​​of each set of model parameters under constant current discharge conditions, and select the time domain corresponding to the smaller error at each set temperature as the optimal time domain to form a "temperature-optimal time domain" binary;

[0013] Step 5. Based on the "temperature-optimal time domain" binary in step 4, establish a coupling relationship curve between temperature and optimal time domain. Determine the optimal time domain at the battery temperature using the collected battery temperature, and use it to identify model parameters instead of the fixed time domain. Then, substitute the identified model parameters into the established basic equivalent circuit model for simulation.

[0014] Preferably, the basic equivalent circuit model constructed in step 1 is a first-order Thevenin equivalent circuit model, and the equation of the model is as follows:

[0015] U=U OC -IR0-U1 (1)

[0016]

[0017] Where, U represents the terminal voltage, U OC is the open circuit voltage, R0 is the ohmic internal resistance, I is the load current, U1 is the polarization voltage, C1 is the polarization capacitance, and R1 is the polarization internal resistance.

[0018] Preferably, the model parameters to be identified in step 1 are: ohmic internal resistance R0, polarization resistance R1, polarization capacitance C1, open circuit voltage U OC .

[0019] Preferably, the pulse power test in step 2 is a hybrid pulse power characteristic test (HPPC).

[0020] Preferably, the method for identifying the model parameters to be identified in step 2 is as follows:

[0021] Ohmic internal resistance identification:

[0022] According to the pulse power test charge and discharge test data, R0 is calculated using the ohmic internal resistance calculation formula, which is as follows:

[0023]

[0024] Where U A is the starting voltage under discharge pulse excitation at a set time point in the cycle step voltage variation diagram, U B The end voltage under discharge pulse excitation at a set time point in the voltage variation diagram of the cycle step;

[0025] Open circuit voltage identification:

[0026] According to the pulse power test charge and discharge test data, the battery voltage after a certain period of rest after a pulse power test is completed is regarded as the open circuit voltage U OC (Open Circuit Voltage, OCV).

[0027] Preferably, the model parameters in step 3 are polarization parameters, and the identification method of the model parameters is as follows:

[0028] Performing time domain analysis on the circuit, the time domain relationship equation of the basic equivalent circuit model is obtained as follows:

[0029]

[0030] Where t is time, τ1 is time constant, and τ1=R1C1;

[0031] Formula (4) and formula (5) are described as exponential function relationships as follows:

[0032]

[0033] By comparing equations (4), (5), and (6), the parameters of the model to be identified can be determined:

[0034] R1=A1 / I (7)

[0035] C1=t1 / R1 (8)

[0036] Where A1 and y0 are the coefficients to be identified, and t1 is the time constant corresponding to τ1.

[0037] Preferably, the coupling relationship in step 5 is obtained by fitting the Arrhenius equation, which is as follows:

[0038]

[0039] Where, T measure is the actual collected battery temperature, t optimize is the optimal time domain under the battery temperature, and A, B, and C are the coefficients to be identified.

[0040] Beneficial effects: The present invention establishes a commonly used battery equivalent circuit model, extracts model parameters through a hybrid pulse power characteristic test (HPPC) experiment, calculates the error between the simulation value and the test value of each group of model parameters under constant current discharge conditions, selects the time domain corresponding to the smaller error at each temperature as the optimal time domain, and establishes a coupling relationship curve between temperature and the optimal time domain by fitting the Arrhenius equation. Then, the coupling relationship is used to obtain the optimal time domain at a certain temperature, and the optimal time domain is used instead of the fixed time domain for model parameter identification. The identified model parameters are then substituted into the established battery equivalent circuit model for simulation, thereby solving the problems of unstable and inaccurate identification accuracy of model parameters identified by using a fixed time domain. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A flow chart of a method for improving battery model accuracy based on a variable time domain;

[0042] Figure 2 This is the equivalent circuit diagram of the Thevenin battery;

[0043] Figure 3 The current and voltage change curves of the HPPC cycle steps at 25°C and a state of charge (SOC) of 0.6;

[0044] Figure 4 The OCV-SOC curves of the linear interpolation fitting at 5℃, 25℃ and 55℃ are shown;

[0045] Figure 5 This is a comparison chart of the polarization parameters of the optimal time-domain identification model for the 6-minute discharge phase of the HPPC test at 5°C, and the model simulation values ​​and experimental values ​​under the 0.5C / 1C constant current discharge condition;

[0046] Figure 6 This is a comparison chart of the polarization parameters of the optimal time-domain identification model for the 6-minute discharge phase of the HPPC test at 25°C, and the model simulation values ​​and experimental values ​​under the 0.5C / 1C constant current discharge condition;

[0047] Figure 7 This is a comparison chart of the polarization parameters of the optimal time-domain identification model for the 6-minute discharge phase of the HPPC test at 55°C, and the model simulation values ​​and experimental values ​​under the 0.5C / 1C constant current discharge condition;

[0048] Figure 8 This is a curve diagram of the temperature and optimal time domain coupling relationship fitted based on the Arrhenius equation. DETAILED DESCRIPTION

[0049] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but the protection scope of the present invention is not limited thereto.

[0050] like Figure 1 As shown, a method for improving the accuracy of a battery model based on a variable time domain includes the following steps:

[0051] Step 1: Build a basic equivalent circuit model and determine the parameters of the model to be identified based on the basic equivalent circuit model;

[0052] Among existing lithium battery equivalent circuit models, the Thevenin equivalent circuit model can, to a certain extent, compensate for the shortcomings of the internal resistance model in characterizing battery dynamic characteristics. This model takes into account the sudden and gradual changes in battery voltage under the stimulation of charge and discharge currents, using R0 to simulate the resistance characteristics of voltage mutations. The RC circuit can effectively simulate the polarization effects during battery charge and discharge and characterize the chemical reactions within the battery. Therefore, the Thevenin equivalent circuit model is selected in this invention, but it is certainly not limited to this.

[0053] like Figure 2 As shown, the basic equivalent circuit model built is a first-order Thevenin equivalent circuit model, and the equation of the model is as follows:

[0054] U=U OC -IR0-U1 (1)

[0055]

[0056] Where, U represents the terminal voltage, U OC is the open circuit voltage, R0 is the ohmic internal resistance, I is the load current, U1 is the polarization voltage, C1 is the polarization capacitance, and R1 is the polarization internal resistance.

[0057] According to the above expression, the model parameters to be identified of the first-order Thevenin equivalent circuit model of the battery are: ohmic internal resistance R0, polarization resistance R1, polarization capacitance C1, open circuit voltage U OC .

[0058] Step 2: Perform a pulse power test on the battery at different set temperatures and at intervals of a certain percentage of the state of charge (SOC) to identify the model parameters to be identified and establish a relationship between the model parameters and the state of charge (SOC);

[0059] HPPC tests were performed on the battery using a battery testing system and a battery temperature test chamber at 5°C, 25°C, and 55°C at intervals of 5% state of charge (SOC). The battery used in the test was the US18650VTC6 lithium-ion battery produced by Sony, with a rated capacity of 3.0Ah and a nominal voltage of 3.6V.

[0060] Collect the battery operating voltage and current data over a period of time during the HPPC cycle test to obtain the cycle step voltage and current change diagram at intervals of 5% state of charge (SOC). Figure 3 The curve of voltage and current changes in the cycle step when the state of charge (SOC) is 0.6 at 25°C.

[0061] Every 5% state of charge (SOC), a parameter model ohmic resistance R0 and a parameter model open circuit voltage U are calculated and identified. OC , establish the relationship between model parameters and state of charge (SOC), the specific process is as follows:

[0062] First, identify the parameter model ohmic resistance R0. Figure 3 For example, the A to B stage is suddenly stimulated by a 10s discharge pulse, and the voltage quickly increases from U A Descend to U BThis phenomenon is caused by the ohmic internal resistance R0. According to this stage, the ohmic internal resistance R0 can be identified and the calculation formula is as follows:

[0063]

[0064] Where I is the load current.

[0065] Secondly, the open circuit voltage U of the identification parameter model OC Each complete pulse stage of the HPPC test has a 6-minute discharge process at a constant current of 0.5C, during which the state of charge (SOC) drops by about 5%. The battery is then left to stand for 2 hours to allow the battery voltage to reach a relatively stable state. The battery voltage after 2 hours is considered the open circuit voltage U OC , and then get the state of charge (SOC) and open circuit voltage U OC The corresponding data, such as Figure 4 Shown is the OCV-SOC curve using linear interpolation fitting at ambient temperatures of 5°C, 25°C, and 55°C.

[0066] Step 3: Select a discharge phase of a set time in each pulse power test of step 2, intercept different time domains in the phase to perform model parameter identification, and establish the relationship between the model parameters and the state of charge (SOC);

[0067] like Figure 3 As shown, B to C is the pulse discharge stage. Due to the polarization characteristics of the battery, the terminal voltage changes from U B Slowly descend to U C The process of the discharge current charging the polarized capacitor is the zero-state response stage of the RC circuit. By analyzing the circuit in the time domain, the functional relationship between the terminal voltage U and time t in this stage can be obtained as follows:

[0068]

[0069] Where t is time, τ1 is time constant, and τ1=R1C1;

[0070] like Figure 3 As shown in the figure, D to E is the static state after the pulse discharge - the "rebound rise" stage. The terminal voltage rises rapidly and then gradually tends to a stable state, which is the "rebound characteristic" of the battery. The process of polarized capacitor discharge is the zero input response stage of the RC circuit. At this time, the functional relationship between the terminal voltage U and time t is:

[0071]

[0072] And formula (4) and formula (5) conform to the exponential function relationship, which can be described as:

[0073]

[0074] By comparing equations (4), (5), and (6), we can obtain:

[0075] R1=A1 / I (7)

[0076] C1=t1 / R1 (8)

[0077] Where A1 and y0 are the coefficients to be identified, and t1 is the time constant corresponding to τ1.

[0078] Based on the above time domain analysis results, the 20s, 40s, and 60s time domains of the 6-min discharge phase of the HPPC test were selected at 5°C, 25°C, and 55°C for model parameter identification. A parameter model polarization capacitor C1 and a parameter model polarization resistor R1 were identified every 5% state of charge (SOC), and the relationship between the above model parameters and the state of charge (SOC) was established.

[0079] Step 4: Based on the basic equivalent circuit model established in step 1 and the battery model parameters obtained in steps 2 and 3, calculate the error between the simulation and experimental values ​​of each set of model parameters under constant current discharge conditions, and select the time domain corresponding to the smaller error at each set temperature as the optimal time domain to form a "temperature-optimal time domain" binary;

[0080] Substitute the battery model parameters obtained in steps 2 and 3 into the battery equivalent circuit model established in step 1, perform terminal voltage tests of the model and the experiment respectively under 0.5C / 1C constant current discharge conditions, compare the measured simulation values ​​with the experimental values, and use the average relative error and maximum relative error as error reference standards to calculate the error between the simulation and experimental values ​​of each set of model parameters under constant current discharge conditions.

[0081] According to the above error calculation, the time domain corresponding to the smaller error at each temperature is selected as the optimal time domain. In this embodiment, 60s, 20s and 10s are selected as the optimal time domains at 5°C, 25°C and 55°C respectively, forming a "temperature-optimal time domain" binary. The simulation and experimental results are compared. Figure 5-Figure 7 shown.

[0082] Step 5. Based on the "temperature-optimal time domain" binary in step 4, establish a coupling relationship curve between temperature and optimal time domain. Determine the optimal time domain at the battery temperature using the collected battery temperature, and use it to identify model parameters instead of the fixed time domain. Then, substitute the identified model parameters into the established basic equivalent circuit model for simulation.

[0083] Based on the Arrhenius equation, the influence of temperature on the optimal time domain is determined, and a coupling relationship curve between temperature and optimal time domain is established. Then, the coupling relationship is used to obtain the optimal time domain at a certain temperature, replacing the fixed time domain for model parameter identification, thereby improving the model accuracy. The fitting result of the "temperature-optimal time domain" binary in step 4 is as follows: Figure 8 As shown, according to Figure 8 For the fitting curve, the Arrhenius equation can be expressed as follows:

[0084]

[0085] Where, T measure is the actual collected battery temperature, t optimize is the optimal time domain at this temperature.

[0086] In actual application, the battery temperature T measure , substitute it into the above temperature and optimal time domain coupling relationship curve, and then get the optimal time domain t at this temperature optimize , with this optimal time domain t optimize Instead of the traditional fixed time domain, refer to the specific implementation method of identifying the model polarization parameters in step 3, and use the t optimize The model polarization parameters are identified in the time domain, and the identified model parameters are then substituted into the established battery first-order Thevenin equivalent circuit model for simulation to improve the accuracy of the battery model.

[0087] The embodiments described are preferred implementations of the present invention, but the present invention is not limited to the above implementations. Any obvious improvements, substitutions or modifications that can be made by those skilled in the art without departing from the essence of the present invention are within the scope of protection of the present invention.

Claims

1. A method for improving battery model accuracy based on a variable time domain, characterized in that: The following steps are involved: Step 1: Build a basic equivalent circuit model and determine the parameters of the model to be identified based on the basic equivalent circuit model; Step 2: Perform a pulse power test on the battery at different set temperatures and at intervals of a certain percentage of the state of charge (SOC) to identify the model parameters to be identified and establish a relationship between the model parameters and the state of charge (SOC); Step 3: Select a discharge phase of a set time in each pulse power test of step 2, intercept different time domains in the phase to perform model parameter identification, and establish the relationship between the model parameters and the state of charge (SOC); Step 4: Based on the basic equivalent circuit model established in step 1 and the battery model parameters obtained in steps 2 and 3, calculate the error between the simulation and experimental values ​​of each set of model parameters under constant current discharge conditions, and select the time domain corresponding to the smaller error at each set temperature as the optimal time domain to form a "temperature-optimal time domain" binary; Step 5: Based on the "temperature-optimal time domain" binary in step 4, establish a coupled relationship curve between temperature and optimal time domain. Determine the optimal time domain at the battery temperature using the collected battery temperature, and use this to identify model parameters instead of the fixed time domain. Substitute the identified model parameters into the established basic equivalent circuit model for simulation.

2. The method for improving battery model accuracy based on a variable time domain according to claim 1, characterized in that: The basic equivalent circuit model constructed in step 1 is a first-order Thevenin equivalent circuit model, and the equation of the model is as follows: （1） （2） Where, Indicates the terminal voltage, is the open circuit voltage, represents the ohmic internal resistance, represents the load current, is the polarization voltage, represents the polarized capacitance, Indicates polarization internal resistance.

3. The method for improving battery model accuracy based on a variable time domain according to claim 2, characterized in that: The model parameters to be identified in step 1 are: ohmic internal resistance , polarization resistance , polarized capacitance , open circuit voltage .

4. The method for improving battery model accuracy based on a variable time domain according to claim 2, characterized in that: In the step 2, the pulse power test selects the hybrid pulse power characteristic test (Hybrid Pulse Power Characteristic, HPPC).

5. The method for improving battery model accuracy based on variable time domain according to claim 4, characterized in that: The identification method for the model parameters to be identified in step 2 is as follows: Ohmic internal resistance identification: According to the pulse power test charge and discharge test data, the ohmic internal resistance calculation formula is used to calculate , the formula is as follows: （3） Where, is the starting voltage under discharge pulse excitation at a set time point in the cycle step voltage variation diagram, The ending voltage under discharge pulse excitation at a set time point in the voltage variation diagram of the cycle step; Open circuit voltage identification: According to the pulse power test charge and discharge test data, the battery voltage after a certain period of rest after a pulse power test is completed is regarded as the open circuit voltage. (Open Circuit Voltage, OCV).

6. The method for improving battery model accuracy based on a variable time domain according to claim 2, characterized in that: The model parameters in step 3 are polarization parameters, and the identification method of the model parameters is as follows: Performing time domain analysis on the circuit, the time domain relationship equation of the basic equivalent circuit model is obtained as follows: （4） （5） Where, For time, is the time constant, where ; Formula (4) and formula (5) are described as exponential function relationships as follows: （6） By comparing equations (4), (5), and (6), the parameters of the model to be identified can be determined: （7） （8） Where, 、 is the coefficient to be identified, For The corresponding time constant.

7. The method for improving battery model accuracy based on a variable time domain according to claim 1, characterized in that: The coupling relationship in step 5 is obtained by fitting the Arrhenius equation, and the formula is as follows: （9） Where, is the actual collected battery temperature, is the optimal time domain under the battery temperature, and A, B, and C are the coefficients to be identified.

Images