Battery model offline identification method based on time-frequency domain combination

Abstract

The invention belongs to the field of battery management systems, and particularly relates to a battery model off-line identification method based on time-frequency domain combination, which comprises the following steps: acquiring an impedance spectrum through EIS test, extracting omega max, and directly calculating R1 and C1; performing a short-time direct-current pulse test, applying a short-time constant-current pulse to the battery, and collecting voltage response data; taking the frequency domain parameter as a constraint, and fitting a large time parameter by using time domain data; and model verification: integrating parameters and verifying the dynamic response precision of the full time scale. Small time constants are accurately captured through frequency domain analysis, and large time parameters are optimized through time domain fitting, so that dynamic errors of the model are reduced; frequency domain rapid decoupling reduces the time domain test duration, and the comprehensive test time is shortened; high-frequency and high-precision time domain acquisition equipment is not needed, and only a conventional EIS instrument and a pulse testing device are needed; the method is suitable for electric vehicles, energy storage systems and other scenes with high requirements for real-time performance and precision.

Description

Technical Field

[0001] The present invention belongs to the field of battery management systems, and in particular relates to an offline identification method for a battery model based on the combination of time and frequency domains. Background Art

[0002] Among existing battery dynamic characteristics modeling methods, lumped parameter models (such as equivalent circuit models) are widely used due to their simple structure and clear physical meaning. Common models include the Rint model, the Thevenin model (first-order RC model), and higher-order RC models. Parameter identification typically uses time-domain methods (such as pulse testing) or frequency-domain methods (such as electrochemical impedance spectroscopy (EIS)).

[0003] In the time domain, the voltage response is obtained through DC pulse testing, and the parameters are determined using least squares fitting. However, the identification accuracy of high-frequency dynamic characteristics (small time constants) is insufficient. In the frequency domain, the impedance spectrum characteristics are extracted through EIS, but this traditional method only focuses on a single time scale and has difficulty in simultaneously characterizing the fast dynamics (such as polarization response) and slow dynamics (such as diffusion processes) of the battery.

[0004] In the existing technology, existing methods do not effectively distinguish between different time scales of battery dynamic characteristics (such as millisecond-level polarization and minute-level diffusion), resulting in poor model adaptability. When the time domain or frequency domain is identified separately, coupling errors are easily generated between parameters. Especially under small time constants, frequency domain characteristics are easily masked by large time scale parameters. Traditional time domain methods require long-term testing, while frequency domain methods have high requirements for equipment. Moreover, the combination of the two is insufficient, making it difficult to strike a balance between efficiency and accuracy. Summary of the Invention

[0005] To solve the above technical problems, the present invention provides a battery model offline identification method based on the combination of time and frequency domains, comprising:

[0006] S1: Obtain EIS electrochemical impedance spectroscopy based on the first-order RC model and extract the relevant parameter semicircle vertex frequency ω max , calculate the frequency domain parameters r1 and R1 / C1; where r1 is the small time constant, R1 is the resistance, and C1 is the capacitance;

[0007] S2: Perform short-duration DC pulse testing in the time domain to collect voltage response data;

[0008] S3: Using the frequency domain parameters as initial values, perform constrained least squares fitting on the time domain data to determine the large time scale parameters;

[0009] S4: Integrate parameters and verify the accuracy of the model’s dynamic response over the full time range.

[0010] Beneficial effects of the present invention:

[0011] The present invention accurately captures small time constants through frequency domain analysis and optimizes large time parameters through time domain fitting, thereby reducing the dynamic error of the model. Rapid frequency domain decoupling reduces the time domain test duration and shortens the comprehensive test time. No high-frequency and high-precision time domain acquisition equipment is required, only a conventional EIS instrument and pulse test device are required. The invention is suitable for scenarios with high real-time and precision requirements, such as electric vehicles and energy storage systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 This is a flow chart of a battery model offline identification method based on the combination of time and frequency domains of the present invention;

[0013] Figure 2 Schematic diagram of the first-order RC model structure of the present invention;

[0014] Figure 3 This is a schematic diagram of the semicircular characteristic frequency of the electrochemical impedance spectroscopy of the present invention;

[0015] Figure 4 Schematic diagram of the second-order RC model structure of the present invention. DETAILED DESCRIPTION

[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0017] A battery model offline identification method based on the combination of time and frequency domains, such as Figure 1 Shown, including:

[0018] S1: Obtain EIS electrochemical impedance spectroscopy based on the first-order RC model and extract the relevant parameter semicircle vertex frequency ω max , calculate the frequency domain parameters r1 and R1 / C1; where r1 is the small time constant, R1 is the resistance, and C1 is the capacitance;

[0019] S2: Perform short-duration DC pulse testing in the time domain to collect voltage response data;

[0020] S3: Using the frequency domain parameters as initial values, perform constrained least squares fitting on the time domain data to determine the large time scale parameters;

[0021] S4: Integrate parameters and verify the accuracy of the model’s dynamic response over the full time range.

[0022] Figure 2It is a structural diagram of the first-order RC equivalent circuit model; it shows the structure of the first-order RC equivalent circuit model, including the ohmic internal resistance R0 and the polarization branch (R1 and C1 in parallel). This model is used to describe the high-frequency dynamic characteristics (small time scale) of the battery. Its frequency domain impedance expression is:

[0023]

[0024] Where r1=R1C1 is the time constant.

[0025] Substituting the real and imaginary parts into the circle equation:

[0026]

[0027] This equation describes a semicircle.

[0028] Figure 3 This is a schematic diagram of the semicircular characteristic frequency of the electrochemical impedance spectrum; the figure shows the semicircular characteristics of the electrochemical impedance spectrum, and the horizontal axis is the real impedance Z Re , the vertical axis is the imaginary impedance Z Im The extreme value of the imaginary impedance (the semicircle vertex) corresponds to the angular frequency ω max , then satisfy:

[0029]

[0030] Therefore, the small time constant can be directly obtained by r1 = 1 / ω max calculate.

[0031] Example 1: Dynamic parameter identification based on lithium-ion batteries:

[0032] Step 1: Frequency domain parameter extraction. Use electrochemical workstation to perform EIS test, the frequency range is 10 -2 ~10 4 Hz. Then perform data processing to extract the semicircle vertex frequency ω of the impedance spectrum max =100rad / s; calculate the minimum time constant r1 = 1 / ω max＝ 0.01s; according to the formula R1=2*Z Re (ω max ) and C1 = r1 / R1, we get R1 = 0.05Ω, C1 = 200F.

[0033] Step 2: Time-domain pulse testing. Test conditions: Apply a 10-second constant current pulse (charge and discharge current I = 1A) to the battery and record the voltage response. For data acquisition, use a data acquisition card at a 1kHz sampling rate to acquire the voltage-time curve.

[0034] Step 3: Joint parameter optimization. Fix the frequency domain parameters R1 and C1 to known values and perform constrained least squares fitting on the time domain data to obtain the diffusion branch parameters R2 = 0.1Ω and C2 = 5000F.

[0035] Step 4: Model verification. Simulation comparison: Substitute the parameters into the lumped model and compare the simulation results with the measured data.

[0036] The present invention adopts the strategy of "differential scale identification and combination of time and frequency domains" to divide parameter identification into two parts: small time scale (high frequency dynamics) and large time scale (low frequency dynamics). Small time scale parameters are determined by frequency domain impedance spectrum analysis, using characteristic frequency and model semicircular characteristics to decouple parameters; large time scale parameters are determined by time domain pulse testing (HPPC) combined with frequency domain constrained least squares fitting to reduce parameter coupling errors. The key innovations include: First, frequency domain characteristic frequency extraction, based on the impedance real-imaginary part relationship of the first-order RC model through the EIS semicircle vertex angular frequency ω max Directly calculate the small time constant r1=1 / ω max Achieve rapid decoupling; second, the time domain and frequency domain joint optimization takes the small time parameters extracted from the frequency domain as constraints, substitutes them into the time domain pulse response data, and uses the least squares algorithm to fit the large time parameters to avoid parameter cross-interference; third, multi-scale model integration, constructs a lumped parameter model including R0 (ohmic internal resistance), R1-C1 (polarization branch) and diffusion branch, covering the full time scale dynamic characteristics.

[0037] Example 2: Extension to multi-order RC model:

[0038] For higher precision requirements, it can be expanded to a second-order RC model such as Figure 4 As shown, the two semicircle vertex frequencies ω are separated by frequency domain max1 and ω max2 , calculate R1-C1 and R2-C2 respectively, and then fit the diffusion branch parameters based on the time domain data.

[0039] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A battery model offline identification method based on the combination of time and frequency domains, characterized in that: include: S1: Obtain EIS electrochemical impedance spectroscopy based on the first-order RC model and extract the relevant parameter semicircle vertex frequency ω max , calculate the frequency domain parameters r1 and R1 / C1; where r1 is the small time constant, R1 is the resistance, and C1 is the capacitance; S2: Perform short-duration DC pulse testing in the time domain to collect voltage response data; S3: Using the frequency domain parameters as initial values, perform constrained least squares fitting on the time domain data to determine the large time scale parameters; S4: Integrate parameters and verify the accuracy of the model’s dynamic response over the full time range.

2. The method for offline identification of battery models based on the combination of time and frequency domains according to claim 1, characterized in that: The first-order RC model consists of an ohmic internal resistance R0 and a polarization branch connected in series, and the polarization branch consists of a resistor R1 and a capacitor C1 connected in parallel; the first-order RC model is used to describe the high-frequency dynamic characteristics of the battery.

3. The offline identification method of battery model based on time-frequency domain combination according to claim 1 is characterized in that: Obtain EIS electrochemical impedance spectroscopy based on the first-order RC model and extract the relevant parameter semicircle vertex frequency ω max , calculate r1 and R1 / C1, including: The EIS test was performed using an electrochemical workstation with a frequency range of 10 -2 ~10 4 Hz; Perform data processing and extract the impedance spectrum semicircle vertex frequency ω max =100rad / s; Calculate the minimum time constant r1 = 1 / ω max＝ 0.01s; according to the formula R1=2*Z Re (ω max ) and C1=r1 / R1, we get R1=0.05Ω, C1=200F; where R1 is the resistor, C1 is the capacitor, Z Re is the real impedance.

4. The method for offline identification of battery models based on the combination of time and frequency domains according to claim 1, characterized in that: Perform short-duration DC pulse testing in the time domain to collect voltage response data, including: A constant current pulse of 10 seconds was applied to the battery, wherein the charge and discharge current of the constant current pulse was I=1 A, and the voltage response was recorded. For data acquisition, a voltage-time curve was obtained using a data acquisition card at a sampling rate of 1 kHz.

5. The offline identification method of battery model based on time-frequency domain combination according to claim 1 is characterized in that: Using the frequency domain parameters as initial values, constrained least squares fitting is performed on the time domain data to determine the large time scale parameters, including: The frequency domain parameters R1 and C1 are fixed to known values, and a constrained least squares fit is performed on the voltage response data to obtain the diffusion branch parameters R2 = 0.1Ω and C2 = 5000F.

6. The method for offline identification of battery models based on the combination of time and frequency domains according to claim 1, characterized in that: Integrate parameters and verify the accuracy of the model's dynamic response over the full time range, including: Substitute the parameters into the first-order equivalent circuit model for simulation testing, and compare the simulation results with the measured data.

Images