Battery state estimation method based on general MCU platform

Abstract

The invention discloses a battery state estimation method based on a general MCU platform, and the method comprises the steps: data collection and processing: employing a voltage sensor, a current sensor and a temperature sensor to collect the operation data of a battery, and enabling the current sensor and the voltage sensor to automatically carry out the zero calibration every five minutes, the temperature signal is subjected to three-point sampling and median filtering processing; sOC estimation comprises the following steps: a) static state estimation: estimating by adopting an open-circuit voltage and SOC relation curve OCV-SOC table look-up mode, and introducing temperature compensation correction to obtain SOC; b) dynamic charge and discharge state estimation: # imgabs0 # real-time error correction: constructing a voltage monitoring time window with the length of N for calculating the deviation delta V between the actual voltage and the estimated voltage in real time, when the delta V exceeds a set threshold, triggering an extended Kalman filter (EKF) algorithm to correct the battery state, the state variables of the estimated voltage including the state of charge SOC and the internal resistance Rinter; and dynamically updating an initial value SOC0 in the ampere-hour integral method through a filtering correction result.

Description

Technical Field

[0001] The present invention relates to the technical field of battery management systems, and in particular to a method for accurately estimating the battery state applicable to scenarios such as electric vehicles and energy storage systems, focusing on solving the problem of real-time and high-precision estimation of the state of charge and health. Background Art

[0002] With the widespread adoption of new energy vehicles and energy storage systems, battery management systems (BMS) play a crucial role in ensuring battery safety, extending battery life, and improving vehicle performance. Accurately estimating the battery's state of charge (SOC) and state of health (SOH) is a core function of BMS.

[0003] However, existing technologies still have many shortcomings in SOC and SOH estimation:

[0004] Low SOC estimation accuracy: The traditional ampere-hour integration method suffers from current sensor zero drift error and cumulative error, which can easily cause SOC estimation offset. While the open-circuit voltage (OCV) method has high theoretical accuracy, its application is limited under dynamic conditions, with an overall error generally exceeding 3%.

[0005] Single dimension of SOH estimation: Current SOH estimation is mostly based on a single parameter, such as the number of cycles or capacity decay, without considering multiple influencing factors such as temperature and internal resistance, resulting in large deviations in the estimation results.

[0006] Insufficient dynamic adaptability: Under dynamic operating conditions such as battery charge and discharge switching and drastic temperature fluctuations, existing estimation methods are difficult to accurately track in real time and are prone to jump errors;

[0007] Strong hardware dependence: Some high-precision estimation methods rely on dedicated computing chips to perform complex matrix operations, which increases the system hardware cost and is not conducive to promotion on mid- and low-end devices or general-purpose MCU platforms.

[0008] Due to the above problems, the risk of battery overcharging or over-discharging in the existing BMS system during use has increased significantly by about 40%, while the battery life has been shortened by about 15%-20%, seriously restricting the performance and reliability improvement of new energy vehicles and energy storage equipment.

[0009] Therefore, there is an urgent need for a battery state estimation method and system with high-precision estimation capability, good dynamic response capability, and low hardware dependence to improve the overall battery management level. Summary of the Invention

[0010] The purpose of the present invention is to provide a battery state estimation method based on a universal MCU platform. The method can achieve high-precision, real-time and stable battery state of charge (SOC) and state of health (SOH) estimation under resource-constrained universal microcontroller (MCU) hardware conditions, so as to improve the comprehensive performance of the battery management system (BMS) in terms of cost, adaptability, estimation accuracy and operational stability.

[0011] The technical solution adopted by the present invention to solve the above technical problems is:

[0012] A battery state estimation method based on a general MCU platform includes the following steps:

[0013] S1. Data acquisition and processing: Battery operation data is collected using voltage sensors, current sensors, and temperature sensors. The current sensors and voltage sensors use a dynamic calibration mechanism and automatically perform zero-point calibration every 5 minutes. The temperature signal is sampled at three points and processed using a median filter. The collected data is divided into static, charge and discharge, and transient fluctuation data types according to the working status, and used for different estimation models respectively. The accuracy of the voltage sensor and current sensor is ±0.1%.

[0014] Collect reliable, high-quality data (voltage, current, temperature) and perform pre-processing (calibration, filtering, and state classification). This provides a clean and clearly classified data source for the subsequent SOC estimation in step S2, error correction in step S3, and SOH assessment in step S4.

[0015] S2, SOC estimation, including:

[0016] a) Static state estimation: When the battery current is detected to be continuously less than 0.05C and the duration is greater than 10 minutes, the open circuit voltage and SOC relationship curve OCV-SOC table is used for estimation, and temperature compensation is introduced to obtain the SOC. The temperature compensation correction formula is as follows:

[0017]

[0018] Where, f(OCV) is the open circuit voltage lookup function, T is the current temperature, k T is the temperature compensation coefficient, k T =0.05% / ℃.

[0019] b) Dynamic charge and discharge state estimation: SOC estimation is performed using the improved ampere-hour integration method. The integration calculation takes into account the coulomb efficiency η, and η is dynamically adjusted according to the current temperature and charge and discharge rate to obtain the current SOC(t) estimate:

[0020]

[0021] Among them, SOC0 is the initial state of charge, which comes from the OCV estimation in the last static state or the last EKF correction result; under static conditions, the SOC value is obtained by looking up the table using the open-circuit voltage method, and after temperature compensation, it is set as the current ampere-hour integral initial value SOC0, which is used as the dynamic SOC calculation benchmark; I(τ) represents the current at time τ, which is collected in real time by the current sensor and is measured in amperes; t0 is the starting time of the SOC estimation cycle, corresponding to the time point of the initial SOC0, and t is the time corresponding to the current SOC; Cn represents the rated capacity of the battery, in ampere-hours; η varies between 0.95 and 1.0 and is stored as a function in the MCU internal lookup table module.

[0022] Specifically, when the temperature T is less than 0°C and the battery is in the charging state, η takes the value of 0.97;

[0023] When the temperature T is less than 0°C and the device is in the discharge state, η takes the value of 0.98;

[0024] When the temperature is 0℃≤T≤45℃ and in the charging state, η is 0.99;

[0025] When the temperature is 0℃≤T≤45℃ and in the discharge state, η takes the value of 0.995;

[0026] When the temperature T>45℃ and the battery is in the charging state, η is 0.95;

[0027] When the temperature T>45℃ and it is in the discharge state, η takes the value of 0.9.

[0028] Estimate battery state of charge (SOC) based on different operating conditions (stationary / dynamic).

[0029] Static state: use OCV to look up the table, with high accuracy;

[0030] Dynamic state: Use the improved ampere-hour integration method, which is suitable for actual work scenarios.

[0031] S3, real-time error correction: Construct a voltage monitoring time window of length N (initial setting N=60), collect voltage, current and other data of the last N moments, and use it to calculate the deviation ΔV between the actual voltage and the estimated voltage in real time. When ΔV exceeds the set threshold, such as ΔV>50mV, the extended Kalman filter EKF algorithm is triggered to correct the battery state. The state variables of the estimated voltage include the state of charge SOC and the internal resistance R internal ; Dynamically update the initial value SOC0 in the ampere-hour integration method through the filtering correction result.

[0032] The extended Kalman filter (EKF) algorithm uses the following state space model to construct state variables and observation equations:

[0033] State variables: ;

[0034] Among them, SOC k represents the battery state of charge at the kth moment, R internal,k represents the internal resistance of the battery at the kth moment;

[0035] Observation Equation: Estimating Voltage ;

[0036] Among them, OCV (SOC k ) is the functional relationship between open circuit voltage and SOC, I k is the current at the kth moment;

[0037] Calculate the difference between the actual voltage and the estimated voltage

[0038] Among them, V measured,k is the actual terminal voltage collected at the kth moment;

[0039] When the voltage deviation ΔV k When the preset threshold is exceeded, the EKF algorithm calculates the state variable SOC k With R internal,k Make corrections and update the SOC k Used to calibrate the initial state of charge SOC0 of the ampere-hour integration method.

[0040] Dynamically adjust the window length:

[0041] The initial length of the voltage monitoring time window is N = 60s, and it is dynamically adjusted based on the temperature change rate and current fluctuation:

[0042] When the temperature change rate is greater than 1°C / min, the time window length is automatically shortened to N=30s;

[0043] When the current fluctuation is less than 5%, the recovery time window length is N=60s.

[0044] Correct the estimation error introduced in step 2 due to complex operating conditions, and use the extended Kalman filter (EKF) to dynamically correct the SOC and internal resistance to improve the system's robustness and estimation stability under non-ideal operating conditions (such as sudden temperature changes, charge and discharge switching, etc.).

[0045] S4, SOH estimation: including capacity attenuation factor SOH C , internal resistance growth factor SOH R and temperature stress factor SOH T Calculation of , and the comprehensive SOH result based on the weighted fusion model: .

[0046] Capacity attenuation factor SOH C Obtained through the following methods:

[0047] After completing 10 complete charge and discharge cycles, the actual capacity C recorded during the discharge process is counted. measured and with rated capacity C rated The ratio calculation yields:

[0048]

[0049] The actual discharge capacity can be obtained in the following two ways:

[0050] The first method is direct current integration, which uses the sensor data from step S1 to record the battery current over time and integrate it to obtain the discharge capacity:

[0051]

[0052] I(t): Current collected in real time (positive value indicates discharge);

[0053] η(t): Coulombic efficiency, derived from the dynamic efficiency adjustment set in step S2;

[0054] t start , t end : The time points of discharge start and end are determined by the current state and are derived from step S1.

[0055] The second method: indirect back-calculation based on SOC change value:

[0056] In a complete charge / discharge cycle, the discharge capacity can be inferred from the SOC change:

[0057] ;

[0058] Among them, SOC start , SOC end SOC estimation from step S2;

[0059] After these SOC values ​​are dynamically corrected by the extended Kalman filter (EKF) in step S3, they can effectively avoid cumulative errors and improve the accuracy of ΔSOC.

[0060] In practical applications, the first and second calculation results can verify each other and improve SOH C reliability.

[0061] Internal resistance growth factor SOH R During the battery pulse discharge phase, the current internal resistance is calculated based on the voltage drop ΔV1 and the current change ΔI1:

[0062] And based on this estimate:

[0063] Among them, R initial Reference internal resistance value when calibrating a new battery;

[0064] R endlife The reference internal resistance value for the set end of life.

[0065] The temperature stress factor is calculated based on the accumulated time at high temperature. When the battery temperature exceeds 45°C, the time starts to count. T The calculation is as follows:

[0066]

[0067] Among them, α is the high temperature aging coefficient, α=0.001% / h, It is the cumulative high temperature exposure time in hours.

[0068] The dynamic correction mechanism in step S1 is a built-in timer that triggers a zero-point calibration every 5 minutes. Current sensor zero-point calibration: When the current is detected as zero, such as when the MCU determines that the system is in a static state or all loads are disconnected, it automatically uses the current current sensor output as the zero-current reference and updates the bias value for subsequent SOC integration and error correction.

[0069] Voltage sensor zero-point calibration: When there's no external charger or load connected to the battery, and the voltage is stable at ΔV / Δt below a set threshold, the current measured voltage is automatically used as the zero-point offset to correct for measurement errors caused by sensor drift. This calibration mechanism effectively offsets long-term sensor drift due to temperature drift and aging, improving the reliability of SOC estimation and EKF filter inputs, and enhancing the stability of the estimation model.

[0070] Regarding the temperature signal, the system collects multiple sets of temperature data at the battery end plate, busbar and cover plate. Each set of data is recorded as T1, T2, T3, and is processed as the current effective temperature T after median filtering. m This method is used for subsequent SOC estimation and temperature compensation calculations. It can effectively eliminate occasional spike interference, improve the reliability of temperature data and the robustness of model input.

[0071] Compared with the prior art, the advantages of the present invention are:

[0072] 1. The battery state estimation method described in the present invention consists of four steps: step 1 is intelligent data acquisition, step 2 is initial SOC estimation, step 3 is dynamic error correction, and step 4 is multi-dimensional SOH evaluation. Although each step can be run independently, they are highly coupled and interdependent, forming a complete closed-loop estimation system. The core of the system is to use the data generated by the previous steps and the intermediate state variables as the basis for subsequent evaluation to achieve system-level optimization.

[0073] 2. The present invention adopts a hierarchical and progressive estimation structure to systematically integrate SOC and SOH modeling and estimation, avoiding local errors caused by isolated estimation. All SOH calculations are based on real-time and effective data support and dynamic correction mechanisms, which have higher timeliness and accuracy than traditional periodic offline calculations. The overall system can run on a general-purpose MCU, has the advantages of high adaptability and low hardware cost, and meets automotive-grade estimation accuracy standards. DETAILED DESCRIPTION

[0074] The following will be described clearly and completely in conjunction with the technical solutions of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments.

[0075] The present invention provides a battery state estimation method that integrates multi-source data acquisition, dynamic SOC estimation, online error correction and multi-dimensional SOH evaluation. It can not only significantly improve the real-time and accuracy of SOC estimation, but also realize dynamic comprehensive judgment of SOH, solving the problems of poor state estimation accuracy, severe lag and inability to correct during operation in the existing technology. It has good engineering practicality and commercial promotion value.

[0076] This embodiment builds a test platform on a 12V lithium iron phosphate battery, uses an STM32F103C8T6 MCU as the main control chip, and is supplemented by an INA226 voltage / current acquisition module and an NTC thermistor sensor to monitor the three parameters in real time.

[0077] Example 1

[0078] This embodiment uses a 1C rate discharge process as an application scenario to illustrate the specific execution process of the battery state estimation method based on a general MCU platform proposed in the present invention.

[0079] In this embodiment, the system first collects voltage, current, and temperature signals. The current sensor detects a continuous discharge current, indicating that the system is in a dynamic discharge state. Based on this, the system automatically switches to dynamic SOC estimation mode and begins to use the ampere-hour integration method to estimate SOC.

[0080] Specifically, the SOC value is updated every 10 seconds. The ampere-hour integral calculation process is as follows:

[0081]

[0082] The starting SOC value is SOC0 = 55%, and the coulombic efficiency η is automatically adjusted based on the current temperature and discharge rate and is set to 0.995. The current I(τ) is collected in real time by the sensor.

[0083] At the 30th second, the system detected a sudden drop of 5mV in voltage, judging it as a large estimation error and triggering the EKF algorithm to perform a correction. During the correction process, the SOC and internal resistance were adjusted based on the difference ΔV between the estimated and measured voltages and the constructed state-space model. The correction results are:

[0084] The estimated SOC value was revised down from 52.3% to 50.8%;

[0085] The internal resistance parameter is increased from 10mΩ to 12mΩ.

[0086] As the system continues to run, after completing 10 complete charge and discharge cycles, the system automatically enters the SOH estimation module.

[0087] The SOH submodule parameters are calculated as follows:

[0088] The actual cumulative discharge capacity is 98.5% of the rated capacity, so:

[0089] =98.5%;

[0090] Current internal resistance R now The initial internal resistance R is 12mΩ. initial The end-of-life internal resistance R is set to 10mΩ. endlife is 20mΩ, then:

[0091]

[0092] Cumulative high temperature exposure time t>45℃=10 hours, α=0.001% / h, calculated as:

[0093]

[0094] Finally, the system uses a weighted fusion method to calculate the comprehensive SOH result:

[0095] .

[0096] Comparative Example 1

[0097] In order to verify the advantages of the present invention, a control experiment was conducted by comparing the method without using EKF correction and dynamic SOH estimation. In the comparative example: the system continuously uses the static ampere-hour integration method to estimate SOC without voltage deviation correction; the current and voltage are not calibrated regularly; the SOH estimation does not introduce SOH R With SOH T , based only on C measured Simple calculation.

[0098] The experiment found that under the same 1C discharge scenario: the voltage drop at 30 seconds was not corrected in time, resulting in a continued expansion of the SOC deviation and an estimation error of 2.5%; the cumulative internal resistance growth was not perceived, and the SOH continued to be overestimated; the comprehensive SOH estimate was 99.3%, which was significantly deviated from the actual health status.

[0099] Compared with Comparative Example 1, the method of the present invention significantly improves the accuracy of raw data by introducing a dynamic current / voltage calibration mechanism and median filtering processing; adopts a multi-state model switching mechanism and an EKF filtering algorithm to significantly reduce the SOC estimation error (to within ±1%); based on multi-dimensional SOH sub-model estimation (capacity / internal resistance / temperature), a more comprehensive and accurate battery health assessment is achieved; supports general MCU platforms, has low hardware resource requirements, and has good portability and practicality.

[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A battery status estimation method based on a general MCU platform, characterized in that: The following steps are involved: S1. Data acquisition and processing: Battery operation data is collected using voltage sensors, current sensors, and temperature sensors. The current sensors and voltage sensors automatically perform zero-point calibration every 5 minutes, and the temperature signal is sampled at three points and processed using a median filter. The collected data are divided into static, charge and discharge, and transient fluctuation data types according to the working status, and used in different estimation models respectively; S2, SOC estimation, including: a) Static state estimation: the SOC is estimated by using the open circuit voltage and SOC relationship curve (OCV-SOC) lookup table, and temperature compensation is introduced to obtain the SOC; b) Dynamic charge and discharge state estimation: SOC estimation is performed using the improved ampere-hour integration method. The integration calculation takes into account the coulomb efficiency η, and η is dynamically adjusted according to the current temperature and charge and discharge rate to obtain the current SOC(t) estimate: Where SOC0 is the initial state of charge, which is derived from the OCV estimate at the last rest state or the result of the last EKF calibration; I(τ) represents the current at time τ, which is collected in real time by the current sensor and is expressed in amperes; Cn represents the rated capacity in ampere-hours. S3, real-time error correction: Construct a voltage monitoring time window of length N to calculate the deviation ΔV between the actual voltage and the estimated voltage in real time. When ΔV exceeds the set threshold, the extended Kalman filter EKF algorithm is triggered to correct the battery state. The state variables of the estimated voltage include the state of charge SOC and the internal resistance R internal ; Dynamically update the initial value SOC0 in the ampere-hour integration method through the filtering correction result.

2. A battery status estimation method based on a general MCU platform according to claim 1, characterized in that: The temperature compensation correction formula introduced in the open circuit voltage and SOC relationship curve OCV-SOC lookup table method is as follows: Where, f(OCV) is the open circuit voltage lookup function, T is the current temperature, k T is the temperature compensation coefficient.

3. The battery status estimation method based on a general MCU platform according to claim 1, characterized in that: The η varies between 0.95 and 1.0 and is stored in the MCU internal table lookup module in the form of a function.

4. A battery status estimation method based on a general MCU platform according to claim 3, characterized in that: When the temperature T is less than 0°C and the battery is in the charging state, η takes the value of 0.97; When the temperature T is less than 0°C and the device is in the discharge state, η takes the value of 0.98; When the temperature is 0℃≤T≤45℃ and in the charging state, η is 0.99; When the temperature is 0℃≤T≤45℃ and in the discharge state, η takes the value of 0.995; When the temperature T>45℃ and the battery is in the charging state, η is 0.95; When the temperature T>45℃ and it is in the discharge state, η takes the value of 0.

9.

5. The battery status estimation method based on a general MCU platform according to claim 1, characterized in that: In step S3, the initial length of the voltage monitoring time window is N=60s, and is dynamically adjusted according to the temperature change rate and current fluctuation: When the temperature change rate is greater than 1°C / min, shorten the time window length to N=30s; When the current fluctuation is less than 5%, the recovery time window length is N=60s.

6. The battery status estimation method based on a general MCU platform according to claim 1, characterized in that: The extended Kalman filter (EKF) algorithm uses the following state space model to construct state variables and observation equations: State variables: ; Among them, SOC k represents the battery state of charge at the kth moment, R internal,k represents the internal resistance of the battery at the kth moment; Observation Equation: Estimating Voltage ; Among them, OCV (SOC k ) is the functional relationship between open circuit voltage and SOC, I k is the current at the kth moment; Calculate the deviation between the actual voltage and the estimated voltage Among them, V measured,k is the actual terminal voltage collected at the kth moment; When the voltage deviation ΔV k When the preset threshold is exceeded, the EKF algorithm calculates the state variable SOC k With R internal,k Make corrections and update the SOC k Used to calibrate the initial state of charge SOC0 of the ampere-hour integration method.

7. A battery status estimation method based on a general MCU platform according to any one of claims 1 to 6, characterized in that: Also included is an SOH estimation, which includes a capacity fade factor SOH C , internal resistance growth factor SOH R and temperature stress factor SOH T Calculation of , and the comprehensive SOH result based on the weighted fusion model: .

8. The battery status estimation method based on a general MCU platform according to claim 7, characterized in that: The capacity decay factor SOH C Obtained through the following methods: After completing 10 complete charge and discharge cycles, the actual capacity C recorded during the discharge process is counted. measured and with rated capacity C rated The ratio calculation yields:

9. The battery status estimation method based on a general MCU platform according to claim 7, characterized in that: The internal resistance growth factor SOH R During the battery pulse discharge phase, the current internal resistance is calculated based on the voltage drop ΔV1 and the current change ΔI1: And based on this estimate: Among them, R initial Reference internal resistance value for new batteries when calibrated; R endlife The reference internal resistance value for the set end of life.

10. The battery status estimation method based on a general MCU platform according to claim 7, characterized in that: The temperature stress factor is calculated based on the accumulated time at high temperature. When the battery temperature exceeds 45°C, the time starts to count. T The calculation is as follows: Wherein, α is the high temperature aging coefficient, α=0.001% / h.

Images