Battery charge state estimation method based on two-step method

Abstract

The invention provides a battery state-of-charge estimation method based on a two-step method, and the method employs the two-step method to estimate the SOC of a battery, considers the influence of the environment temperature and the discharge rate on the capacity and characteristics of the battery, carries out the analysis and fitting, explores the relation between an identification parameter and the SOC, determines a relation equation, and corrects the related parameters in a second-order RC equivalent model, thereby achieving the estimation of the SOC of the battery. And the model parameter identification accuracy can be improved. And estimating the SOC of the battery in real time through extended Kalman filtering based on identification and correction results. Besides, under the condition that battery data acquisition is inaccurate or even fails due to reasons such as failure of an acquisition module sensor of the BMS system, parameter identification of the model and evaluation of the SOC can be completed according to the correction result, and the safety and stability of functions of the BMS system are ensured.

Description

Technical Field

[0001] The present invention relates to the field of battery technology, and in particular to a battery state of charge estimation method based on a two-step method. Background Art

[0002] At present, most battery SOC estimation methods are based on experimental data modeling, using improved algorithms or deep learning to estimate battery SOC, without considering the impact of different ambient temperatures and discharge rates on the internal characteristics of the battery. The accuracy of parameter identification determines the accuracy of the battery model, and the experimental data must be comprehensive.

[0003] In the common battery model building process, only HPPC experimental data at an ambient temperature (usually 25°C) is input to complete the identification. In low or high temperature environments, the accuracy of the model parameter identification results will decrease. The present invention completes battery HPPC experiments at different ambient temperatures and improves the model input to improve the accuracy of the identification results.

[0004] For the improved algorithm to estimate SOC, a lot of time is needed to adjust parameters to obtain the measurement noise and process noise under the best estimation state. Otherwise, the filtering accuracy of the algorithm will decrease or even diverge. In addition, external factors such as sensor accuracy will also affect the measurement noise and process noise, resulting in a decrease in estimation accuracy. The deep learning method of estimating SOC is often based on neural networks. Although neural networks have powerful computing capabilities, they have high requirements on the accuracy of training data and high computing costs. Therefore, they are not practical in actual applications. In the present invention, the improved Kalman filter algorithm can effectively eliminate noise and jitter, estimate the state in real time, and has a fast response speed. Therefore, it is used as a method for SOC estimation. Summary of the invention

[0005] The present invention mainly considers that changes in battery capacity and battery temperature under different ambient temperatures and discharge rates may lead to inaccurate battery state of charge estimation, and proposes a battery state of charge estimation method based on a two-step method to improve the estimation accuracy of the battery state of charge.

[0006] The technical solution adopted by the present invention to solve its technical problem is:

[0007] A battery state of charge estimation method based on a two-step method comprises the following steps:

[0008] The first step of the “two-step approach” is to build a model:

[0009] S1. Collect data from HPPC experiments of lithium batteries at different ambient temperatures, including current and voltage, and draw a SOC-OCV relationship curve of lithium batteries;

[0010] S2. Combined with the characteristics of the battery equivalent circuit model, a second-order RC equivalent circuit model is established to simulate the characteristics of the lithium battery, and the parameters in the model are identified, including the battery internal resistance R0, polarization internal resistance R1, R2 and polarization capacitance C1, C2, etc.;

[0011] S3. Based on step S2, the extended Kalman filter algorithm is used to complete the battery SOC estimation of the model and complete the model pre-training;

[0012] The second step of the “two-step method” is to calibrate the model parameters:

[0013] S4. Under the same ambient temperature, the battery is subjected to charge and discharge cycle experiments at different discharge rates, and the data of the experimental battery, including current, voltage, temperature and capacity, are recorded, and a SOC-X (identification parameter) change curve is drawn;

[0014] S5. Perform charge-discharge cycle experiments on the battery at the same discharge rate and different ambient temperatures, record the data of the experimental battery, including current, voltage, temperature and capacity, draw a SOC-X (identification parameter) change curve, and determine the components that are sensitive to temperature changes and have a greater impact on the discharge rate based on the SOC-X (identification parameter) curves at different temperatures and different discharge rates;

[0015] S6. Use polynomial functions to fit the relationship between the component parameters that are sensitive to temperature changes (mainly ohmic resistance and polarization internal resistance, with an influence range of more than 0.5-2 times) and are greatly affected by the discharge rate (mainly polarization internal resistance, with an influence range of more than 0.5-1 times) and the battery SOC, and obtain the relationship equations and weights between each parameter and SOC at different temperatures and discharge rates; use the change relationship equations to correct the component parameter identification under different discharge rates and different temperature environments in actual driving engineering, and obtain the corrected parameter values;

[0016] S7. Use the extended Kalman filter algorithm to complete the battery SOC estimation of the corrected model.

[0017] As a further technical solution of the present invention, in step S2, combined with Figure 2 The second-order RC equivalent circuit model and Kirchhoff's law give:

[0018]

[0019] Where U L Indicates the terminal voltage of the battery, U OC represents the open circuit voltage of the battery, i represents the current passing through the battery, R0 represents the internal resistance of the battery, R1 and R2 represent the polarization resistance of the battery, C1 and C2 represent the polarization capacitance corresponding to R1 and R2, and U1 and U2 represent the terminal voltage of R1 and R2 respectively;

[0020] Combined with the SOC calculation formula Where: Q is the battery capacity, SOC t0 The initial value of battery SOC can be obtained as the state space equation:

[0021]

[0022] Where, τ1=R1C1, τ2=R2C2, t is the sampling time, w k is the system noise, v k is the observation noise, k is the discrete sampling point, representing the current moment;

[0023] make: u k =I k , D k =[-R0], R is the measurement noise covariance.

[0024] As a further technical solution of the present invention, step S3 includes:

[0025] 1) The state value of the nonlinear discrete system at time k is:

[0026] Among them, f(x k ,u k ) is the state equation, h(x k ,u k ) is the observation equation, x k+1 is the state variable of the system at time k+1, y k is the output variable of the system at time k;

[0027] After linearizing the equation using Taylor expansion, the system is estimated using the Kalman filter algorithm:

[0028] 2) Let function f(x k ,u k ) and h(x k ,u k ) can be derived at each sampling time, and using Taylor expansion, we can get:

[0029]

[0030] 3) Order Substituting into the state space equation of the nonlinear discrete system, we can obtain:

[0031]

[0032] In the formula, is x k The estimated value at time k;

[0033] Next, iterative estimation is performed using the Kalman filter algorithm.

[0034] As a further technical solution of the present invention, the iterative estimation by the Kalman filter algorithm comprises the following steps:

[0035] State initialization: P 0 / 0 =var(x0), where E(·) is the expected value;

[0036] Predict estimated values ​​of state variables:

[0037] Predict the error covariance matrix at time k:

[0038] Calculate the gain matrix K k :

[0039] According to the observed value, the state estimate at time k is updated as:

[0040] Update the error covariance matrix: P k =(ik k C k ) k / k-1 ;

[0041] In summary, the linearized state equation and observation equation are obtained:

[0042] According to the above expression, combined with the principle of extended Kalman algorithm, the battery SOC can be estimated.

[0043] The beneficial effects of the present invention are:

[0044] The "two-step method" of the present invention improves the accuracy of battery SOC estimation. Based on the basic working conditions, the influence of ambient temperature and discharge rate on the internal characteristics of the battery is further considered. In the second-order RC equivalent model, the identification parameters are further corrected. Under the premise of ensuring the stability and safety of the BMS function, the accuracy of equivalent model parameter identification and battery SOC estimation is improved, and the calculation is simpler, faster and more accurate. It can effectively improve the accuracy of battery SOC estimation during vehicle operation, and at the same time improve the estimation accuracy of battery SOH, thereby extending the service life of the vehicle battery. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 A flow chart of a battery state of charge estimation method based on a two-step method provided in an embodiment of the present invention.

[0046] Figure 2A second-order RC equivalent circuit model diagram provided for an embodiment of the present invention.

[0047] Figure 3 A diagram of the iterative estimation process of the Kalman filter algorithm provided in an embodiment of the present invention.

[0048] Figure 4 A specific flow chart of the second step in a battery state of charge estimation method based on a two-step method provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are 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 creative work are within the scope of protection of the present invention.

[0050] See also Figure 1 , an embodiment of the present invention provides a battery state of charge estimation method based on a two-step method, comprising the following steps;

[0051] The first step of the “two-step approach” is to build a model:

[0052] S1. Collect data from HPPC experiments of lithium batteries at different ambient temperatures, including current and voltage, and draw a SOC-OCV relationship curve of lithium batteries;

[0053] S2. Combined with the characteristics of the battery equivalent circuit model, a second-order RC equivalent circuit model is established to simulate the characteristics of the lithium battery, and the parameters in the model are identified, including the battery internal resistance R0, polarization internal resistance R1, R2 and polarization capacitance C1, C2, etc.;

[0054] S3. Based on step S2, the extended Kalman filter algorithm is used to complete the battery SOC estimation of the model and complete the model pre-training;

[0055] The second step of the “two-step method” is to calibrate the model parameters:

[0056] S4. Under the same ambient temperature, the battery is subjected to charge and discharge cycle experiments at different discharge rates, and the data of the experimental battery, including current, voltage, temperature and capacity, are recorded, and a SOC-X (identification parameter) change curve is drawn;

[0057] S5. Perform charge-discharge cycle experiments on the battery at the same discharge rate and different ambient temperatures, record the data of the experimental battery, including current, voltage, temperature and capacity, draw a SOC-X (identification parameter) change curve, and determine the components that are sensitive to temperature changes and have a greater impact on the discharge rate based on the SOC-X (identification parameter) curves at different temperatures and different discharge rates;

[0058] S6. Use polynomial functions to fit the relationship between the component parameters that are sensitive to temperature changes (mainly ohmic resistance and polarization internal resistance, with an influence range of more than 0.5-2 times) and are greatly affected by the discharge rate (mainly polarization internal resistance, with an influence range of more than 0.5-1 times) and the battery SOC, and obtain the relationship equations and weights between each parameter and SOC at different temperatures and discharge rates; use the change relationship equations to correct the component parameter identification under different discharge rates and different temperature environments in actual driving engineering, and obtain the corrected parameter values;

[0059] S7. Use the extended Kalman filter algorithm to complete the battery SOC estimation of the corrected model.

[0060] In order to facilitate those skilled in the art to better understand the technical solution of the present invention, specific embodiments of the present invention are given as follows:

[0061] The present invention selects 40135 lithium battery as an example to establish an equivalent model and identify parameters. The parameter standards of this type of battery are shown in Table 1:

[0062] Table 1-40135 lithium battery parameter standards

[0063]

[0064] In order to explore the working characteristics of 40135 lithium battery and evaluate the pulse discharge characteristics of the battery, an HPPC discharge experiment was carried out on the battery to obtain the SOC-OCV relationship curve of the battery.

[0065] The HPPC discharge experiment was carried out according to the standard of "Electrical Performance Test Methods for Lithium-ion Power Battery Packs and Systems for Electric Vehicles". In order to obtain the experimental results, the set ambient temperatures were 0℃, 5℃, 10℃, 15℃, 20℃, 25℃, 30℃, 35℃, and 40℃ respectively. The specific process steps are shown in Table 2:

[0066] Table 2-HPPC discharge test steps

[0067]

[0068] Record the battery voltage, current, temperature and other data during the HPPC experiment of 40135 lithium battery at different ambient temperatures. The data collected in this experiment will be used to identify the parameters of each component in the second-order RC equivalent circuit model.

[0069] The present invention uses a second-order RC equivalent model to simulate the operating characteristics of a 40135 lithium battery. Figure 2 The parameters such as battery internal resistance, polarization internal resistance, polarization capacitance, etc. in the model can be identified by combining with HPPC discharge test data.

[0070] Combination Figure 2 The second-order RC equivalent circuit model and Kirchhoff's law can be obtained:

[0071]

[0072] Where U L Indicates the terminal voltage of the battery, U OC represents the open circuit voltage of the battery, i represents the current passing through the battery, R0 represents the internal resistance of the battery, R1 and R2 represent the polarization resistance of the battery respectively, C1 and C2 represent the polarization capacitance corresponding to R1 and R2 respectively, and U1 and U2 represent the terminal voltage of R1 and R2 respectively.

[0073] Combined with the SOC calculation formula Where: Q is the battery capacity, SOC t0 is the initial value of the battery SOC, and the state space equation can be obtained:

[0074]

[0075] Where, τ1=R1C1, τ2=R2C2, t is the sampling time, w k is the system noise, v k is the observation noise, k is the discrete sampling point, representing the current moment.

[0076] make: u k =I k , D k =[-R0], R is the measurement noise covariance;

[0077] At this point, the second-order RC equivalent model of the 40135 lithium battery has been established, and the battery internal resistance, polarization internal resistance, polarization capacitance and other parameters have been identified by combining the HPPC experimental data. The extended Kalman filter algorithm is then used to complete the battery SOC estimation of the model.

[0078] 1) The state value of the nonlinear discrete system at time k is:

[0079] Among them, f(x k ,u k ) is the state equation, h(x k ,u k ) is the observation equation, x k+1 is the state variable of the system at time k+1, y k is the output variable of the system at time k.

[0080] After linearizing the equation using Taylor expansion, the system is estimated using the Kalman filter algorithm:

[0081] 2) Let function f(x k ,u k ) and h(x k ,u k ) can be derived at each sampling time, and using Taylor expansion, we can get:

[0082]

[0083] 3) Order Substituting into the state space equation of the nonlinear discrete system, we can obtain:

[0084]

[0085] In the formula, is x k The estimated value at time k.

[0086] Next, the Kalman filter algorithm is used for iterative estimation, as follows: Figure 3 As shown;

[0087] S1. State initialization: P 0 / 0 =var(x0), where E(·) is the expected value;

[0088] S2, estimated values ​​of predicted state variables:

[0089] S3, predict the error covariance matrix at time k:

[0090] S4. Calculate the gain matrix K k :

[0091] S5. Update the estimated state value at time k based on the observed value:

[0092] S6. Update the error covariance matrix: P k =(ik k C k )k / k-1 ;

[0093] In summary, the linearized state equation and observation equation can be obtained:

[0094] According to the above expression, combined with the principle of extended Kalman algorithm, the battery SOC can be estimated.

[0095] At this point, the first step of the “two-step method” has been completed, and the model training for battery SOC estimation has been completed.

[0096] Then start the second step.

[0097] In the above second-order RC equivalent model, R0 represents the ohmic internal resistance, which describes the loss related to resistance; R1 and C1 describe the loss caused by the reduction of chemical reaction rate and the kinetics of chemical reaction respectively; R2 and C2 describe the loss caused by chemical diffusion inside the lithium battery and the kinetics of chemical diffusion respectively.

[0098] For lithium-ion batteries, the ion concentration gradient and temperature in the battery are the factors that have the greatest impact on the ohmic internal resistance. The ohmic internal resistance R0 increases with the increase of the lithium ion concentration gradient and decreases with the increase of temperature. At the beginning of high-rate discharge, the internal temperature of the battery is low, the chemical reaction is slow, and the battery R0 is high; during the discharge process, the battery temperature rises and the ohmic resistance decreases rapidly; at the end of the discharge, the lithium ion concentration gradient increases, and R0 shows an upward trend. In the case of a large discharge rate, the battery temperature is high and R0 also decreases.

[0099] For polarization resistance R1 and R2: at the beginning of discharge, polarization resistance changes with the increase of ion concentration gradient; during discharge, ion concentration gradient changes tend to be stable, and polarization internal resistance also tends to be stable; at the end of discharge, the ion concentration in the positive and negative active materials differs greatly, resulting in an increase in polarization internal resistance. At the same time, as the discharge rate increases, the ion concentration gradient in the battery will gradually increase, resulting in a decrease in polarization internal resistance.

[0100] For polarized capacitors C1 and C2: at different temperatures, the changes in polarized capacitors are relatively chaotic. However, when the SOC is lower than 20%, the polarized capacitors at different temperatures become smaller, and as the temperature decreases, the polarized capacitors decrease to a lesser extent.

[0101] In order to explore the external characteristic data of 40135 lithium battery at different discharge rates and ambient temperatures and the distribution law of surface temperature during battery charging and discharging, a charge and discharge cycle experiment was carried out on the battery. In order to obtain the experimental results, the set ambient temperatures were -10℃, 20℃, 0℃, 5℃, 10℃, 15℃, 20℃, 25℃, 30℃, 35℃, and 40℃.

[0102] Considering that the battery charge and discharge rate has a significant accelerating effect on the attenuation of its life, but at high rates, lithium deposition will inevitably occur at the negative electrode of the battery, and lithium dendrites will pierce the diaphragm, causing the battery to short-circuit, posing a great safety hazard. Therefore, three cycle rates of 1C, 2C, and 3C are selected to conduct charge and discharge cycle experiments on the battery, which can not only study the effect of the rate on the capacity change of the battery, but also avoid the inevitable safety accidents under high rate charge and discharge cycles.

[0103] The set cycle rates are 1C, 2C, and 3C, respectively, and the specific process step settings are shown in Table 3: Among them, the average value of the discharge capacity in the first three cycles is recorded as the initial capacity of the battery.

[0104] Table 3-Charge and discharge cycle test steps

[0105]

[0106] Record the battery current, voltage and temperature data of lithium batteries in experiments at different ambient temperatures and different discharge rates. The experimental data will be used to explore the effects of different ambient temperatures on battery characteristics and the effects of different discharge rates on battery temperature and characteristics.

[0107] The discharged capacity and battery of 3 groups of experimental data with different ambient temperatures and the same discharge rate and 9 groups of experimental data with different discharge rates and the same ambient temperature were recorded, and the SOC-X (identification parameter) curve was drawn.

[0108] Based on the curve graph obtained from the above experiment and the parameter identification results of the first step. Taking 25℃ as the benchmark, at the same discharge rate, the increase and decrease of ambient temperature will affect the discharge capacity and voltage change of the battery. When the ambient temperature is the same, as the discharge rate increases, the battery discharge capacity gradually decays and the battery capacity continues to decrease.

[0109] According to the SOC-X (identification parameter) curves at different temperatures and discharge rates obtained from the above experimental analysis, the components that are sensitive to temperature changes and the components that are greatly affected by the discharge rate are determined. (Temperature range: -10-40℃)

[0110] According to the different changing trends of the above parameters, functions such as interpolation / quadratic polynomial / piecewise polynomial are selected to fit the changing relationship between R0, R1, R2 and SOC.

[0111] The relationship equations between each parameter and SOC at different temperatures and discharge rates are obtained, the target SOC is substituted into the fitting relationship equation, the identification parameter values ​​after temperature correction are obtained, and the correction of the parameters affected by ambient temperature and discharge rate in the model is completed.

[0112] On this basis, the extended Kalman filter algorithm is used to complete the battery SOC estimation of the corrected model. The specific steps are as follows: Figure 4 shown.

[0113] At this point, the second step of the two-step method has been completed, and the accuracy of the established second-order RC equivalent model containing temperature and discharge rate correction has been greatly improved.

[0114] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0115] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A battery state of charge estimation method based on a two-step method, characterized in that: The steps include: The first step of the "two-step method" is to build a model: S1. Collect data from HPPC experiments of lithium batteries at different ambient temperatures, including current and voltage, and draw a SOC-OCV relationship curve of the lithium battery; S2. Combined with the characteristics of the battery equivalent circuit model, a second-order RC equivalent circuit model is established to simulate the characteristics of the lithium battery, and the parameters in the model are identified, including the battery internal resistance R0, polarization internal resistance R1, R2 and polarization capacitance C1, C2; S3. Based on step S2, the extended Kalman filter algorithm is used to complete the battery SOC estimation of the model and complete the model pre-training; The second step of the "two-step method" - calibration of model parameters: S4. Under the same ambient temperature, the battery is subjected to charge and discharge cycle experiments at different discharge rates, and the data of the experimental battery, including current, voltage, temperature and capacity, is recorded, and a SOC-X change curve is drawn; S5. Carry out charge-discharge cycle experiments on the battery at the same discharge rate and different ambient temperatures, record the data of the experimental battery, including current, voltage, temperature and capacity, draw a SOC-X change curve, and determine the components that are sensitive to temperature changes and have a greater impact on the discharge rate based on the SOC-X curves at different temperatures and different discharge rates; S6. Use a polynomial function to fit the relationship between the component parameters that are sensitive to temperature changes and have a greater impact on the discharge rate and the battery SOC, and obtain the relationship equation and weight between each parameter and the SOC at different temperatures and discharge rates; By using the variation relationship equation, the component parameter identification under different discharge rates and temperature environments in actual driving engineering is corrected to obtain the corrected parameter value; S7. Use the extended Kalman filter algorithm to complete the battery SOC estimation of the corrected model.

2. A battery state of charge estimation method based on a two-step method according to claim 1, characterized in that: In step S2, the second-order RC equivalent circuit model and Kirchhoff's law are combined to obtain: Where U L Indicates the terminal voltage of the battery, U OC represents the open circuit voltage of the battery, i represents the current passing through the battery, R0 represents the internal resistance of the battery, R1 and R2 represent the polarization resistance of the battery, C1 and C2 represent the polarization capacitance corresponding to R1 and R2, and U1 and U2 represent the terminal voltage of R1 and R2 respectively; Combined with the SOC calculation formula Where: Q is the battery capacity, SOC t0 The initial value of battery SOC can be obtained as the state space equation: Where, τ1=R1C1, τ2=R2C2, t is the sampling time, w k is the system noise, v k is the observation noise, k is the discrete sampling point, representing the current moment; D k =[-R0], R is the measurement noise covariance.

3. A battery state of charge estimation method based on a two-step method according to claim 1, characterized in that: Step S3 includes: 1) The state value of the nonlinear discrete system at time k is: Among them, f(x k ,u k ) is the state equation, h(x k ,u k ) is the observation equation, x k+1 is the state variable of the system at time k+1, y k is the output variable of the system at time k; After linearizing the equation using Taylor expansion, the system is estimated using the Kalman filter algorithm: 2) Let function f(x k ,u k ) and h(x k ,u k ) can be derived at each sampling time, and using Taylor expansion, we can get: 3) Order Substituting into the state space equation of the nonlinear discrete system, we can obtain: In the formula, is x k The estimated value at time k; Next, iterative estimation is performed using the Kalman filter algorithm.

4. A battery state of charge estimation method based on a two-step method according to claim 3, characterized in that: The iterative estimation by the Kalman filter algorithm comprises the following steps: State initialization: P 0 / 0 =var(x0), where E(·) is the expected value; Predict estimated values ​​of state variables: Predict the error covariance matrix at time k: Calculate the gain matrix K k : According to the observed value, the state estimate at time k is updated as: Update the error covariance matrix: P k =(ik k C k ) k / k-1 ; In summary, the linearized state equation and observation equation are obtained: The battery SOC can be estimated by combining the extended Kalman algorithm principle.

5. A battery state of charge estimation method based on a two-step method according to claim 1, characterized in that: In step S6, being sensitive to temperature changes means that the influence range of ohmic resistance and polarization internal resistance is more than 0.5-2 times, and having a greater influence on the discharge rate means that the influence range of polarization internal resistance is more than 0.5-1 times.

Images