Unscented H infinite filter battery SOC estimation method based on adaptive parameter updating

Abstract

The invention discloses an unscented H infinite filter battery SOC estimation method based on adaptive parameter updating. The method sequentially comprises the steps of model selection, HPPC experiment design, data sample acquisition, SOC unscented H infinite estimation algorithm and the like. According to the method, a model parameter lookup table or a relationship between fitting parameters and the SOC under different SOCs is established through offline experiments, and in the unscented H infinity filtering process, the model parameters are updated online according to the real-time working state of the battery, so that the model precision is improved; meanwhile, the robustness of unscented H infinity is utilized to suppress the influence of noise and model uncertainty, so that the precision and robustness of SOC estimation are improved, and the method is more suitable for practical application, and is suitable for application of electric vehicles, hybrid electric vehicles and the like.

Description

Technical Field

[0001] The present invention relates to a battery SOC estimation method, in particular to a battery SOC estimation method based on an untraceable H-infinity filter based on adaptive parameter updating. Background Art

[0002] The battery state of charge (SOC) is a key parameter in the BMS, which directly affects the endurance, energy management, charge and discharge control, life prediction and safety of electric vehicles; existing SOC estimation methods include the ampere-hour integration method, the open circuit voltage method, the internal resistance method, the data-driven method, and the Kalman filter and its improved algorithms (such as EKF and UKF).

[0003] Chinese patent CN109613432A proposes a method, device and computer-readable storage medium for estimating the state of charge of a battery. The method online identifies the parameter values ​​of the first-order RC equivalent circuit model of the power battery, obtains the state of charge estimated by the model, and uses the ampere-hour integration method to estimate the state of charge of the power battery. Although this method realizes the correction of the accumulated error of the ampere-hour integration and reduces the error accumulated over a long period of operation in the existing ampere-hour integration method, it does not fully deal with the coupling effects of multiple factors such as nonlinearity, aging, and temperature, lacks a dynamic correction mechanism, and is difficult to cope with complex working conditions.

[0004] Chinese patent CN111505505A proposes a method and electronic device for calculating and correcting the state of charge (SOC) of a lithium battery in a hybrid system. The method charges the hybrid system's lithium battery with a first set DC current, obtains a first terminal voltage of the lithium battery after a set holding time, controls the lithium battery to discharge with a second set DC current after the set charging time, and obtains a second terminal voltage of the lithium battery after a set discharge time. The SOC of the lithium battery in the hybrid system is calculated based on the first and second terminal voltages. The method controls the first set current during charging and the second set current during discharging to be equal in magnitude and opposite in direction, and obtains two terminal voltages of the lithium battery at set times during the charging and discharging processes to calculate the SOC of the lithium battery in the hybrid system. While this method implements calculations during the charging and discharging of the lithium battery and enables SOC correction at any time, its correction process relies on terminal voltage measurements under specific charging and discharging conditions and fails to fully consider real-time interference on the terminal voltage from polarization effects, temperature changes, and battery aging under dynamic operating conditions. This results in significant errors in scenarios where the hybrid system frequently switches between charging and discharging modes.

[0005] Chinese patent CN108562858A proposes an SOC measurement device for electric vehicle instruments. This device uses a combination of the ampere-hour integration method and the internal resistance measurement method to predict the remaining battery capacity and service life. The display device can show the remaining available power and the measured voltage, current, and temperature parameters. When the battery power is low, it can automatically issue an alarm signal to remind the driver to charge in time. However, this method is susceptible to cumulative errors, and the internal resistance method is sensitive to temperature and dynamic working conditions, resulting in a decrease in the accuracy of SOC estimation during long-term use or in complex environments.

[0006] In summary, these methods all have the following problems: model parameters are often fixed and difficult to adapt to parameter changes caused by factors such as SOC, temperature and aging during battery use, resulting in estimation errors; robustness to noise and interference is poor; some complex algorithms have large computational complexity and are difficult to apply in real time in BMS. Summary of the Invention

[0007] The purpose of the present invention is to solve the problems in the prior art and propose an unscented H-infinity filtering battery SOC estimation method based on adaptive parameter updating, which can improve model accuracy and improve the accuracy and robustness of SOC estimation.

[0008] To achieve the above object, the present invention proposes an unscented H-infinity filtering battery SOC estimation method based on adaptive parameter updating, which includes the following steps in sequence: a) Fully charge the energy storage system and let it stand until the internal state reaches equilibrium; b) After fully charging at constant current and constant voltage, set several SOC values ​​from 0 to 100%, and then discharge at constant current to the set SOC value; c) Apply a series of constant current pulses to perform HPPC testing, record the voltage and current changes during the pulses, calculate the system parameter values ​​based on the voltage and current, and store them in an offline table; d) determining whether it is the last SOC value, if yes, proceeding to step f), if no, proceeding to step e); e) Enter the next set SOC value and repeat steps b) to c) to obtain the corresponding system parameter values ​​under the SOC value; f) Establish a state-space equation containing the SOC value based on the first-order PNGV model and perform discretization processing; g) Initialize system parameters and set H-infinity filter parameter γ; h) Perform an unscented transformation on the current state to generate 2n+1 sigma points, where n is the dimension of the state vector, and give the weight corresponding to each point; i) Substitute the 2n+1 sigma points into the discretized state space equation and combine them with the weights to obtain the prior state estimate; j) Substitute the prior state estimate into the measurement equation to perform measurement update; k) calculating the filter gain based on the measurement vector covariance matrix and the covariance matrix between the state vector and the measurement vector; 1) Calculate the posterior state estimate by combining the prior state estimate, the measured value, and the gain to obtain the current SOC; m) Determine whether to continue online SOC estimation. If yes, perform a table lookup based on the current SOC value, combine interpolation to update system parameters, and retain the current state to enter step h) to perform SOC estimation at the next sampling moment; if no, end the program.

[0009] Preferably, the constant current pulse in step c) includes a discharge pulse and a charge pulse, and the system parameter values ​​include open circuit voltage (OCV), ohmic internal resistance (R0), polarization internal resistance (Rp) and polarization capacitance (Cp).

[0010] Preferably, the method for calculating the open circuit voltage (OCV), ohmic internal resistance (R0), polarization internal resistance (Rp) and polarization capacitance (Cp) corresponding to each SOC point in step c) is as follows: OCV: take the average voltage value in the static stage as OCV; R0: calculate based on the voltage mutation ΔV and current change ΔI at the moment of the start of the discharge or charging pulse, R0 = ΔV / ΔI; Rp and Cp: perform exponential fitting on the voltage decay curve after the pulse to obtain the time constant τ, then Rp = τ / Cp, where Cp can be estimated by other methods or empirical values ​​or obtained by model fitting.

[0011] Preferably, the state space equation of the first-order PNGV model established in step f) is: Assume that the state vector of the battery is , SOC V S is the state of charge, V S is the polarization voltage; the control variable , I is the charge and discharge current; the output variable is , V is the terminal voltage; it is written as a state space equation and discretized as: ; in, , is the charge and discharge efficiency, is the available capacity of the battery, is the sampling period, 、 are process noise and measurement noise respectively.

[0012] Preferably, the initialization parameters in step g) are given initial state estimation values, initial state estimation error covariance matrix, process noise covariance matrix Q and measurement noise covariance matrix R, and the H-infinity filter parameter γ is set: γ controls the robustness of the filter. The smaller the γ value, the stronger the robustness of the filter, but the accuracy will be sacrificed.

[0013] Preferably, in step g), 2n+1 sigma points are generated: ; in, is the covariance matrix of the posterior state estimation error of the previous time step; n=2 represents 2 state variables; Generate the weights for each point: ; The superscript "m" represents the weight used in calculating the mean, and the superscript "c" represents the weight used in calculating the variance. It satisfies the Gaussian distribution. , Here we take 0.0001, Take 0 or 3-n, here take 0.

[0014] Preferably, in step h), the 2n+1 sigma points are substituted into the discretized state space equation: ; Introducing weights, we get the prior state estimate: ; Taking into account the process noise covariance matrix Q, the covariance matrix of the prior state estimation error is calculated: .

[0015] Preferably, in step i), the prior state estimate is substituted into the measurement equation to perform measurement update: use The unscented transformation generates 2n+1 sigma points. Substituting these 2n+1 points into the measurement equation yields: ; The covariance matrix of the measurement vector is: ; The error covariance matrix between the state vector and the measurement vector is: .

[0016] Preferably, in step j), the filter gain is calculated by taking into account the measurement noise covariance matrix R to obtain the filter gain of the current step size: .

[0017] As a preference, in step k), the sensor acquires the , then the posterior state estimate for this time step is: ; The covariance matrix of the posterior state estimation error is: ; in, ,in Be careful to set it up to ensure that a solution exists at every step, and each time step The minimum value of satisfies the following conditions: ; In summary, the SOC value at the current moment is: .

[0018] The beneficial effects of the present invention are as follows: the present invention establishes a model parameter lookup table or a relationship between fitting parameters and SOC under different SOCs through offline experiments, and updates the model parameters online according to the real-time battery operating status during the untraceable H-infinity filtering process, thereby improving the model accuracy; at the same time, the robustness of the untraceable H-infinity is utilized to suppress the influence of noise and model uncertainty, thereby improving the accuracy and robustness of SOC estimation, making it more suitable for practical applications, such as electric vehicles and hybrid vehicles.

[0019] The features and advantages of the present invention will be described in detail through embodiments with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 1. It is a schematic diagram of the offline preparation process involved in the method of the present invention; Figure 2 The present invention relates to a schematic diagram of an online implementation process. DETAILED DESCRIPTION

[0021] Model selection: Taking into account both computational complexity and accuracy, the first-order PNGV model is selected to describe the energy storage system, and its state-space equation is established. System parameters include open-circuit voltage, ohmic internal resistance, polarization internal resistance, and polarization capacitance, which are related to the SOC value.

[0022] HPPC experiment design to obtain data samples: fully charge the energy storage system and let it stand until the internal equilibrium is reached; discharge it at a constant current to the set SOC point (SOC is measured using the ampere-hour integration method); apply a series of constant current pulses (including discharge pulses and charge pulses), record the voltage and current changes during the pulses, and calculate the system parameters; repeat the above process to obtain the corresponding system parameters at different SOC values; save the above data in a table.

[0023] SOC traceless H-infinity estimation algorithm: estimate the initial state of the system, including the average value and uncertainty of the estimated initial state; set an important parameter γ, which controls the robustness of the filter, that is, the ability to resist interference; the smaller the γ value, the stronger the robustness, but may sacrifice some accuracy; vice versa; generate a set of Sigma points around the current state estimate, these points are Gaussian distributed around the state estimate, capturing the uncertainty of the state; each Sigma point has a corresponding weight, representing its importance in the estimation; substitute all sigma points into the state space equation to obtain the prior state estimate; combine the current and voltage that the sensor can obtain in real time as measurement values ​​with the prior state estimate, calculate the posterior state estimate at the current moment, and obtain the current SOC value; use the current SOC value to look up the table, update the system parameters, and repeat the above process.

[0024] See Figure 1 、 Figure 2 The unscented H-infinity filter battery SOC estimation method based on adaptive parameter update includes the following steps in sequence: a) Fully charge the energy storage system and let it stand until the internal state reaches equilibrium; b) After fully charging at constant current and constant voltage, set several SOC values ​​from 0 to 100%, and then discharge at constant current to the set SOC value; c) Apply a series of constant current pulses to perform HPPC testing, record the voltage and current changes during the pulses, calculate the system parameter values ​​based on the voltage and current, and store them in an offline table; d) determining whether it is the last SOC value, if yes, proceeding to step f), if no, proceeding to step e); e) Enter the next set SOC value and repeat steps b) to c) to obtain the corresponding system parameter values ​​under the SOC value; f) Establish a state-space equation containing the SOC value based on the first-order PNGV model and perform discretization processing; g) Initialize system parameters and set H-infinity filter parameter γ; h) Perform an unscented transformation on the current state to generate 2n+1 sigma points, where n is the dimension of the state vector, and give the weight corresponding to each point; i) Substitute the 2n+1 sigma points into the discretized state space equation and combine them with the weights to obtain the prior state estimate; j) Substitute the prior state estimate into the measurement equation to perform measurement update; k) calculating the filter gain based on the measurement vector covariance matrix and the covariance matrix between the state vector and the measurement vector; 1) Calculate the posterior state estimate by combining the prior state estimate, the measured value, and the gain to obtain the current SOC; m) Determine whether to continue online SOC estimation. If so, perform a table lookup based on the current SOC value, combine interpolation to update system parameters, and retain the current state to enter step h) to perform SOC estimation at the next sampling moment; if not, terminate the program. Whether to continue online SOC estimation is controlled by manual operation. If manual operation is detected to turn off the continuation of online SOC estimation, terminate this program.

[0025] Example 1: 1. Experimental equipment: Energy storage system battery cell: BYD-LFP40A lithium iron phosphate battery cell, nominal capacity 40Ah, nominal voltage 3.2V.

[0026] Battery testing system: BTS-5V100A battery testing system, which has constant current and constant voltage charging and discharging functions, current accuracy of ±0.1%FS, voltage accuracy of ±0.05%FS, and data acquisition frequency of 10Hz; Constant temperature box: DHS-225A constant temperature box, with a temperature control accuracy of ±0.5℃, is used to control the experimental temperature at 25℃.

[0027] Computer: Desktop computer equipped with Intel i7-13700K processor, 16GB DDR5 memory and Windows 11 Home Chinese version operating system.

[0028] Data processing software: MATLAB, Python.

[0029] 2. HPPC experimental steps: Preparation: Place the battery cells in a thermostat, set the temperature to 25°C, and let it sit for 2 hours to allow the internal temperature to become uniform.

[0030] Full charge: Use the BTS-5V100A battery test system to charge the battery at a constant current of 1C (40A) to 3.6V. Then charge at a constant voltage of 3.6V until the current drops to 0.05C (2A). Let it sit for 1 hour.

[0031] Discharge: Discharge at a constant current of 1C. Every time the SOC drops by 1%, pause the discharge and let the battery stand to allow internal balance to be achieved. Use the ampere-hour integration method to record the discharged capacity and calculate the current SOC.

[0032] HPPC test: HPPC test is performed at each SOC point, including the following steps: discharge at 1C constant current for 10 seconds, rest for 10 seconds; charge at 1C constant current for 10 seconds, rest for 10 seconds; record voltage and current data throughout the process; Repeat: repeat the discharge and HPPC test steps; Data save: save all voltage, current and SOC data.

[0033] 3. Data processing and parameter extraction: Data filtering: The voltage and current data are filtered using a moving average filter to remove noise interference.

[0034] Parameter calculation: Based on the first-order PNGV model, the open circuit voltage (OCV), ohmic internal resistance (R0), polarization internal resistance (Rp) and polarization capacitance (Cp) corresponding to each SOC point are calculated according to the HPPC test data; the specific calculation method is as follows: OCV: the average voltage value in the static stage is taken as OCV; R0: calculated based on the voltage mutation ΔV and current change ΔI at the beginning of the discharge or charging pulse, R0=ΔV / ΔI; Rp and Cp: exponentially fit the voltage decay curve after the pulse to obtain the time constant τ, then Rp=τ / Cp, where Cp can be estimated by other methods or empirical values, or obtained by fitting a more complex model; in this embodiment, the voltage decay curve is fitted by the least squares method, and the values ​​of Rp and Cp are obtained at the same time.

[0035] Create a data table: store the above SOC value-system parameters in the data table for subsequent use.

[0036] 4. Construct an unscented H-infinity filter: Theoretical basis: Unscented H-infinity filtering is a filtering algorithm for nonlinear systems. It processes nonlinear functions through unscented transform (UT), avoiding the complex process of calculating the Jacobian matrix in the extended Kalman filter (EKF). H-infinity filtering does not rely on the statistical characteristics of system noise, but instead ensures the robustness of the filter by minimizing the H-infinity norm of the transfer function of the filtering error system, that is, it has strong resistance to model uncertainty and external interference.

[0037] Establish the state space equation of the first-order PNGV model: Assume that the state vector of the battery is (state of charge and polarization voltage), control variables (charge and discharge current), the output variable is (terminal voltage), which is written as a state-space equation and discretized as: (1); in, , is the charge and discharge efficiency, is the available capacity of the battery, is the sampling period, 、 are process noise and measurement noise respectively.

[0038] Initialization parameters: given the initial state estimate, the initial state estimation error covariance matrix, the process noise covariance matrix Q, and the measurement noise covariance matrix R.

[0039] Set the H-infinity filter parameter γ: γ controls the robustness of the filter. The smaller the γ value, the more robust the filter is, but some accuracy may be sacrificed.

[0040] Generate 2n+1 sigma points (n=2 represents 2 state variables): (2); in, is the covariance matrix of the posterior state estimation error at the previous time step; Generate the weights for each point: (3); The superscript "m" represents the weight used in calculating the mean, and the superscript "c" represents the weight used in calculating the variance. It satisfies the Gaussian distribution. , Here we take 0.0001, Take 0 or 3-n, here take 0.

[0041] Time update: Substitute the 2n+1 sigma points into the discretized state space equation: (4); Introducing weights, we get the prior state estimate: (5); Taking into account the process noise covariance matrix Q, the covariance matrix of the prior state estimation error is calculated: (6); Measurement Update: Yes use The unscented transformation generates 2n+1 sigma points. Substituting these 2n+1 points into the measurement equation yields: (7); The covariance matrix of the measurement vector is: (8); The error covariance matrix between the state vector and the measurement vector is: (9); Filter gain calculation: Considering the measurement noise covariance matrix R, the filter gain for the current step size is obtained: (10); Observation results: Consider the sensor's observations at the current time step. , then the posterior state estimate for this time step is: (11); The covariance matrix of the posterior state estimation error is: (12); in, ,in Be careful to set it up to ensure that a solution exists at every step, and each time step The minimum value of satisfies the following conditions: (13).

[0042] In summary, the SOC value at the current moment is: (14); Update system parameters: According to the current SOC value, obtain the current system parameters through interpolation in the data table and update the parameters.

[0043] Expected results: Through the above embodiment, a high-precision robust estimation method for the SOC of an energy storage system can be established; compared with traditional methods, this method can better cope with the impact of uncertain noise on observation accuracy and improve the performance of the BMS.

[0044] Example 2: 1. Experimental equipment: Energy storage system battery cell: BYD-LFP40A lithium iron phosphate battery cell, nominal capacity 40Ah, nominal voltage 3.2V.

[0045] Battery testing system: BTS-5V100A battery testing system, which has constant current and constant voltage charging and discharging functions, current accuracy of ±0.1%FS, voltage accuracy of ±0.05%FS, and data acquisition frequency of 10Hz; Constant temperature box: DHS-225A constant temperature box, with a temperature control accuracy of ±0.5℃, is used to control the experimental temperature at 25℃.

[0046] Computer: Desktop computer equipped with Intel i7-13700K processor, 16GB DDR5 memory and Windows 11 Home Chinese version operating system.

[0047] Data processing software: MATLAB, Python; 2. HPPC experimental steps: Preparation: Place the battery cells in a thermostat, set the temperature to 25°C, and let it sit for 2 hours to allow the internal temperature to become uniform.

[0048] Full charge: Use the BTS-5V100A battery test system to charge the battery at a constant current of 1C (40A) to 3.6V. Then charge at a constant voltage of 3.6V until the current drops to 0.05C (2A). Let it sit for 1 hour.

[0049] Discharge: Discharge at a constant current of 0.5C. Every time the SOC drops by 0.5%, pause the discharge and let the battery stand to allow internal balance to be achieved. Use the ampere-hour integration method to record the discharged capacity and calculate the current SOC.

[0050] HPPC test: HPPC test is performed at each SOC point, including the following steps: discharge at 1C constant current for 5 seconds, rest for 10 seconds; charge at 1C constant current for 5 seconds, rest for 10 seconds; record voltage and current data throughout the process; repeat: repeat discharge and HPPC test steps; data save: save all voltage, current and SOC data.

[0051] 3. Data processing and parameter extraction: Data filtering: The voltage and current data are filtered using a moving average filter to remove noise interference.

[0052] Parameter calculation: Based on the first-order PNGV model, the open circuit voltage (OCV), ohmic internal resistance (R0), polarization internal resistance (Rp) and polarization capacitance (Cp) corresponding to each SOC point are calculated according to the HPPC test data; the specific calculation method is as follows: OCV: take the average voltage value in the static stage as OCV; R0: calculated according to the voltage mutation ΔV and current change ΔI at the beginning of the discharge or charging pulse, R0=ΔV / ΔI; Rp and Cp: perform exponential fitting on the voltage decay curve after the pulse to obtain the time constant τ, then Rp=τ / Cp, where Cp can be estimated by other methods or empirical values, or obtained by fitting a more complex model; in this embodiment, the voltage decay curve is fitted by the least squares method, and the values ​​of Rp and Cp are obtained at the same time.

[0053] Create a data table: store the above SOC value-system parameters in the data table for subsequent use.

[0054] 4. Construct an unscented H-infinity filter: Theoretical basis: Unscented H-infinity filtering is a filtering algorithm for nonlinear systems. It processes nonlinear functions through unscented transform (UT), avoiding the complex process of calculating the Jacobian matrix in the extended Kalman filter (EKF). H-infinity filtering does not rely on the statistical characteristics of system noise, but instead ensures the robustness of the filter by minimizing the H-infinity norm of the transfer function of the filtering error system, that is, it has strong resistance to model uncertainty and external interference.

[0055] Establish the state space equation of the first-order PNGV model: Assume that the state vector of the battery is (state of charge and polarization voltage), control variables (charge and discharge current), the output variable is (terminal voltage), which is written as a state-space equation and discretized as: (15); in, , is the charge and discharge efficiency, is the available capacity of the battery, is the sampling period, 、 are process noise and measurement noise respectively.

[0056] Initialization parameters: given the initial state estimate, the initial state estimation error covariance matrix, the process noise covariance matrix Q, and the measurement noise covariance matrix R.

[0057] Set the H-infinity filter parameter γ: γ controls the robustness of the filter. The smaller the γ value, the more robust the filter is, but some accuracy may be sacrificed.

[0058] Generate 2n+1 sigma points (n=2 represents 2 state variables): (16); in, is the covariance matrix of the posterior state estimation error at the previous time step; Generate the weights for each point: (17); The superscript "m" represents the weight used in calculating the mean, and the superscript "c" represents the weight used in calculating the variance. It satisfies the Gaussian distribution. , Here we take 0.0001, Take 0 or 3-n, here take 0.

[0059] Time update: Substitute the 2n+1 sigma points into the discretized state space equation: (18); Introducing weights, we get the prior state estimate: (19); Taking into account the process noise covariance matrix Q, the covariance matrix of the prior state estimation error is calculated: (20); Measurement Update: Yes use The unscented transformation generates 2n+1 sigma points. Substituting these 2n+1 points into the measurement equation yields: (twenty one); The covariance matrix of the measurement vector is: (twenty two); The error covariance matrix between the state vector and the measurement vector is: (twenty three); Filter gain calculation: Considering the measurement noise covariance matrix R, the filter gain for the current step size is obtained: (twenty four); Observation results: Consider the sensor's observations at the current time step. , then the posterior state estimate for this time step is: (25); The covariance matrix of the posterior state estimation error is: (26); in, ,in Be careful to set it up to ensure that a solution exists at every step, and each time step The minimum value of satisfies the following conditions: (27); In summary, the SOC value at the current moment is: (28); Update system parameters: According to the current SOC value, obtain the current system parameters through interpolation in the data table and update the parameters.

[0060] Expected results: Compared with Example 1, the amount of parameter table data doubles due to the 50% reduction in the SOC sampling interval; the denser SOC-parameter mapping relationship reduces the maximum interval of linear interpolation from 1% to 0.5% during online table lookup, and the model parameter fitting error is significantly reduced.

[0061] The above embodiments are intended to illustrate the present invention, not to limit the present invention. Any solution that is a simple transformation of the present invention falls within the protection scope of the present invention.

Claims

1. An untraceable H-infinity filter battery SOC estimation method based on adaptive parameter updating, characterized by: The following steps are included in sequence: a) Fully charge the energy storage system and let it stand until the internal state reaches equilibrium; b) After fully charging at constant current and constant voltage, set several SOC values ​​from 0 to 100%, and then discharge at constant current to the set SOC value; c) Apply a series of constant current pulses to perform HPPC testing, record the voltage and current changes during the pulses, calculate the system parameter values ​​based on the voltage and current, and store them in an offline table; d) determining whether it is the last SOC value, if yes, proceeding to step f), if no, proceeding to step e); e) Enter the next set SOC value and repeat steps b) to c) to obtain the corresponding system parameter values ​​under the SOC value; f) Establish a state-space equation containing the SOC value based on the first-order PNGV model and perform discretization processing; g) Initialize system parameters and set H-infinity filter parameter γ; h) Perform an unscented transformation on the current state to generate 2n+1 sigma points, where n is the dimension of the state vector, and give the weight corresponding to each point; i) Substitute the 2n+1 sigma points into the discretized state space equation and combine them with the weights to obtain the prior state estimate; j) Substitute the prior state estimate into the measurement equation to perform measurement update; k) calculating the filter gain based on the measurement vector covariance matrix and the covariance matrix between the state vector and the measurement vector; 1) Calculate the posterior state estimate by combining the prior state estimate, the measured value, and the gain to obtain the current SOC; m) Determine whether to continue online SOC estimation. If yes, perform a table lookup based on the current SOC value, combine interpolation to update system parameters, and retain the current state to enter step h) to perform SOC estimation at the next sampling moment; if no, end the program.

2. The method for estimating battery SOC using an untraceable H-infinity filter based on adaptive parameter updating according to claim 1, wherein: The constant current pulse in step c) includes a discharge pulse and a charge pulse, and the system parameter values ​​include open circuit voltage (OCV), ohmic internal resistance (R0), polarization internal resistance (Rp) and polarization capacitance (Cp).

3. The method for estimating battery SOC using an untraceable H-infinity filter based on adaptive parameter updating according to claim 2, wherein: The method for calculating the open circuit voltage (OCV), ohmic internal resistance (R0), polarization internal resistance (Rp) and polarization capacitance (Cp) corresponding to each SOC point in step c) is as follows: OCV: the average voltage value in the static stage is taken as OCV; R0: calculated based on the voltage mutation ΔV and current change ΔI at the beginning of the discharge or charge pulse, R0 = ΔV / ΔI; Rp and Cp: exponentially fit the voltage decay curve after the pulse to obtain the time constant τ, then Rp = τ / Cp, where Cp can be estimated by other methods or empirical values ​​or obtained by model fitting.

4. The method for estimating battery SOC using an untraceable H-infinity filter based on adaptive parameter updating according to claim 2, wherein: The state space equation of the first-order PNGV model established in step f) is: Assume that the state vector of the battery is , SOC V S is the state of charge, V S is the polarization voltage; the control variable , I is the charge and discharge current; The output variable is , V is the terminal voltage; it is written as a state space equation and discretized as: （1）； in, , is the charge and discharge efficiency, is the available capacity of the battery, is the sampling period, 、 are process noise and measurement noise respectively.

5. The method for estimating battery SOC using an untraceable H-infinity filter based on adaptive parameter updating according to claim 4, wherein: In step g), the initialization parameters are given initial state estimation value, initial state estimation error covariance matrix, process noise covariance matrix Q and measurement noise covariance matrix R, and the H-infinity filter parameter γ is set: γ controls the robustness of the filter. The smaller the γ value, the stronger the robustness of the filter, but the accuracy will be sacrificed.

6. The method for estimating battery SOC using an untraceable H-infinity filter based on adaptive parameter updating according to claim 5, wherein: In step g), 2n+1 sigma points are generated: （2）； in, is the covariance matrix of the posterior state estimation error of the previous time step; n=2 represents 2 state variables; Generate the weights for each point: （3）； The superscript "m" represents the weight used in calculating the mean, and the superscript "c" represents the weight used in calculating the variance. It satisfies the Gaussian distribution. , Here we take 0.0001, Take 0 or 3-n, here take 0.

7. The method for estimating battery SOC using an untraceable H-infinity filter based on adaptive parameter updating according to claim 6, wherein: In step h), the 2n+1 sigma points are substituted into the discretized state space equation: （4）； Introducing weights, we get the prior state estimate: （5）； Taking into account the process noise covariance matrix Q, the covariance matrix of the prior state estimation error is calculated: （6）。 8. The method for estimating battery SOC using an untraceable H-infinity filter based on adaptive parameter updating according to claim 7, wherein: In step i), the prior state estimate is substituted into the measurement equation to perform measurement update: right use The unscented transformation generates 2n+1 sigma points. Substituting these 2n+1 points into the measurement equation yields: （7）； The covariance matrix of the measurement vector is: （8）； The error covariance matrix between the state vector and the measurement vector is: （9）。 9. The method for estimating battery SOC using an untraceable H-infinity filter based on adaptive parameter updating according to claim 8, wherein: In step j), the filter gain is calculated by taking into account the measurement noise covariance matrix R and obtaining the filter gain of the current step size: （10）。 10. The method for estimating battery SOC using untraceable H-infinity filtering based on adaptive parameter updating according to claim 9, wherein: In step k), the sensor acquires the , then the posterior state estimate for this time step is: （11）； The covariance matrix of the posterior state estimation error is: （12）； in, ,in Be careful to set it up to ensure that a solution exists at every step, and each time step The minimum value of satisfies the following conditions: （13）； In summary, the SOC value at the current moment is: （14）。

Images