A method for combined state of charge and state of health estimation of energy storage batteries

Abstract

A method for jointly estimating the state of charge (SOC) and state of health (SOH) of an energy storage battery includes: establishing an equivalent circuit model of the energy storage battery; identifying parameters of the equivalent circuit model to obtain estimated values ​​of its parameters; when the battery's charge level is greater than 0, using an offline-established new battery OCV-SOC model, and based on the estimated values ​​of the parameters of the equivalent circuit model, the new battery OCV-SOC model, and an extended Kalman filter algorithm, obtaining an estimated value of the actual charged charge; when the battery's charge level is equal to 0, establishing an OCV aging model, and based on the OCV aging model and the extended Kalman filter algorithm, obtaining an estimated value of the battery's usable capacity; and obtaining estimated values ​​of SOC and SOH based on the estimated values ​​of the actual charged charge and the usable capacity. The SOC and SOH estimates obtained using this method are more accurate.

Description

Technical Field

[0001] This invention relates to the field of battery testing technology, and in particular to a method for jointly estimating the state of charge and state of health of an energy storage battery. Background Technology

[0002] The deployment of energy storage batteries maintains the stable operation of new energy power generation systems. Energy management of energy storage batteries is particularly important; a reasonable energy storage battery model, accurate model parameter identification, and battery state estimation are prerequisites for energy management. The battery state of an energy storage battery includes State of Charge (SOC) and State of Health (SOH). SOC reflects the battery's remaining capacity, while SOH reflects the battery's aging status. Battery state estimation is the foundation of battery energy management.

[0003] Traditional energy storage battery models (such as equivalent circuit models) often use the least squares method for parameter identification and the Kalman filter algorithm to estimate the state of charge (SOC). However, offline least squares methods are difficult to characterize the dynamic performance of battery parameters. Furthermore, fixed OCV-SOC curve models are often used in parameter identification and SOC and health estimation. As the number of battery uses increases and the battery ages, the fixed OCV-SOC curve model introduces increasingly larger aging errors, potentially leading to inaccurate SOC and health estimates. Summary of the Invention

[0004] The purpose of this invention is to provide a joint estimation method for the state of charge (SOC) and state of health (SOH) of an energy storage battery. Considering the impact of battery aging on SOC estimation, an OCV aging model is established to accurately correct the aging error of SOC estimation based on the OCV aging model, so that the obtained SOC and SOH estimates are more accurate.

[0005] A joint estimation method for the state of charge and state of health of an energy storage battery, comprising:

[0006] Establish an equivalent circuit model for the energy storage battery;

[0007] The parameters of the equivalent circuit model of the energy storage battery are identified to obtain estimated values ​​of the parameters of the equivalent circuit model of the energy storage battery.

[0008] When the energy storage battery's charge is greater than 0, a new battery OCV-SOC model is established offline. Based on the estimated parameters of the energy storage battery's equivalent circuit model, the new battery OCV-SOC model, and the extended Kalman filter algorithm, an estimated value of the actual charged charge is obtained. When the energy storage battery's charge is equal to 0, an OCV aging model is established, and based on the OCV aging model and the extended Kalman filter algorithm, an estimated value of the energy storage battery's usable capacity is obtained.

[0009] Based on the estimated actual charge and the estimated available capacity of the energy storage battery, the estimated values ​​of SOC and SOH are obtained.

[0010] The joint estimation method for the state of charge and state of health of the aforementioned energy storage battery, wherein parameter identification of the equivalent circuit model of the energy storage battery to obtain estimated values ​​of the parameters of the equivalent circuit model of the energy storage battery specifically includes:

[0011] Based on the equivalent circuit model of the energy storage battery, obtain the parameters of the equivalent circuit model of the energy storage battery.

[0012] While the energy storage battery is in a charging state, the measured values ​​of the terminal voltage and current of the energy storage battery are obtained.

[0013] Based on the measured values ​​of the terminal voltage and current of the energy storage battery, the estimated values ​​of the parameters of the equivalent circuit model of the energy storage battery are obtained through the extended Kalman filter algorithm.

[0014] The method for jointly estimating the state of charge and state of health of the aforementioned energy storage battery, wherein the equivalent circuit model of the energy storage battery is a first-order circuit model, and the parameters of the equivalent circuit model of the energy storage battery are obtained based on the equivalent circuit model, specifically including:

[0015] The KVL equations of the first-order circuit model are obtained based on the first-order circuit model and Kirchhoff's laws.

[0016] Discretize the KVL equations of the first-order circuit model to obtain the terminal voltage expression of the first-order circuit model;

[0017] Based on the terminal voltage expression of the first-order circuit model, the parameters of the first-order circuit model are obtained, and the parameters of the equivalent circuit model of the energy storage battery include the open-circuit voltage of the first-order circuit model.

[0018] The joint estimation method for the state of charge and state of health of the aforementioned energy storage battery, wherein obtaining the estimated parameters of the equivalent circuit model of the energy storage battery by means of an extended Kalman filter algorithm based on the measured values ​​of the terminal voltage and the measured values ​​of the current of the energy storage battery specifically includes:

[0019] Based on the parameter identification model of the equivalent circuit model of the energy storage battery, state prediction equations and measurement equations of the parameters of the equivalent circuit model of the energy storage battery are established.

[0020] According to the extended Kalman filter algorithm, the measured values ​​of the terminal voltage and the measured values ​​of the current are input. The state prediction equation of the parameters of the equivalent circuit model of the energy storage battery and the measurement equation of the parameters of the equivalent circuit model of the energy storage battery are combined. Prediction is made through prior estimation, and the Kalman filter gain of the identification model of the parameters of the equivalent circuit model of the energy storage battery is calculated.

[0021] According to the extended Kalman filter algorithm, the parameters of the equivalent circuit model of the energy storage battery are updated by posterior estimation. The Kalman filter gain and posterior estimation covariance of the identification model are updated, and the posterior estimation covariance is used to correct the current parameter estimate, so as to obtain the estimated value of the parameters of the equivalent circuit model of the energy storage battery.

[0022] The joint estimation method for the state of charge and state of health of the aforementioned energy storage battery, wherein an offline-established new battery OCV-SOC model is used, and the estimated value of the actual charged capacity is obtained based on the estimated values ​​of the parameters of the equivalent circuit model of the energy storage battery, the new battery OCV-SOC model, and the extended Kalman filter algorithm, specifically includes:

[0023] Obtain the state prediction equation for the actual charge amount of the energy storage battery when it is in a charging state.

[0024] A measurement model for the terminal voltage is established by using a new battery OCV-SOC model built offline and based on the estimated values ​​of the parameters of the new battery OCV-SOC model and the equivalent circuit model of the energy storage battery.

[0025] According to the extended Kalman filter algorithm, the measured values ​​of the terminal voltage and the current, as well as the estimated values ​​of the parameters of the equivalent circuit model of the energy storage battery, are input. Prediction is performed through prior estimation, and the Kalman filter gain of the estimated charging capacity is calculated. According to the extended Kalman filter algorithm, the Kalman filter gain of the estimated charging capacity and the covariance of the posterior estimation are updated through posterior estimation. The posterior estimation covariance is used to correct the current capacity estimate and obtain the estimated value of the actual charging capacity.

[0026] The above-mentioned method for jointly estimating the state of charge and state of health of energy storage batteries includes, in part, establishing an OCV aging model, which specifically includes:

[0027] In the fully discharged state, the open-circuit voltage of the energy storage battery at the moment of charging is obtained as a data sample, and the least squares method is used to fit the data sample to obtain the OCV aging model.

[0028] The joint estimation method for the state of charge and state of health of the aforementioned energy storage battery, wherein obtaining the estimated value of the usable capacity of the energy storage battery based on the OCV aging model and the extended Kalman filter algorithm specifically includes:

[0029] Establish a predictive equation for the usable capacity of energy storage batteries;

[0030] Based on the OCV aging model, an equation for measuring the terminal voltage of the usable capacity of the energy storage battery is established.

[0031] According to the extended Kalman filter algorithm, the measured values ​​of the terminal voltage and the current are input, and the OCV aging model is introduced using the partial differential equation method. Prediction is performed through prior estimation, and the Kalman filter gain for estimating the available capacity of the energy storage battery is calculated. According to the extended Kalman filter algorithm, the Kalman filter gain and the posterior estimation covariance for estimating the available capacity of the energy storage battery are updated through posterior estimation. The posterior estimation covariance is used to correct the current available capacity, and the estimated value of the available capacity of the energy storage battery is obtained.

[0032] The method for jointly estimating the state of charge (SOC) and state of health (SOH) of the aforementioned energy storage battery, specifically includes obtaining the estimated values ​​of SOC and SOH based on the estimated value of the actual charged amount of electricity and the estimated value of the available capacity of the energy storage battery:

[0033] Obtain the initial value of the remaining charge of the energy storage battery in the charging state, and obtain the estimated value of SOC based on the estimated value of the available capacity of the energy storage battery, the initial value of the remaining charge of the energy storage battery in the charging state, and the estimated value of the actual charge input.

[0034] Obtain the capacity value of the new battery, and obtain the estimated value of the SOH based on the capacity value of the new battery and the estimated value of the available capacity of the energy storage battery.

[0035] The beneficial effects of this invention are as follows:

[0036] This invention employs the extended Kalman filter algorithm for parameter identification, state of charge (SOC) estimation, and state of health (SOH) estimation of the equivalent circuit model of an energy storage battery. The parameter identification method provided by this invention has good online representation capabilities, solving the problem that traditional offline least squares identification cannot accurately reflect the current charge / discharge state, making parameter identification closer to reality and achieving dynamic parameter response under dynamic operating conditions. Furthermore, the SOC and SOH estimation methods provided by this invention consider the impact of battery aging on the OCV-SOC model, thus establishing an OCV aging model to correct aging errors caused by a fixed OCV-SOC curve. Combined with the Kalman filter algorithm, the SOC and SOH estimates obtained by the SOC and SOH estimation methods provided by this invention are more realistic and accurate. Attached Figure Description

[0037] Figure 1 This is a flowchart of a method for jointly estimating the state of charge and state of health of an energy storage battery according to an embodiment.

[0038] Figure 2 This is a schematic diagram of a first-order circuit model according to one embodiment;

[0039] Figure 3 Here is a detailed flowchart for step 102;

[0040] Figure 4 Here is a detailed flowchart for step 301;

[0041] Figure 5 Here is a detailed flowchart for step 303;

[0042] Figure 6 Here is a detailed flowchart for step 104;

[0043] Figure 7 According to one embodiment, a fully discharged open-circuit voltage U OCV0 -C available Fitted curve;

[0044] Figure 8 Here is a detailed flowchart for step 105;

[0045] Figure 9 Here is a detailed flowchart for step 106;

[0046] Figure 10 The waveform of the input excitation current according to one embodiment is shown.

[0047] Figure 11 A waveform diagram of the input voltage according to one embodiment;

[0048] Figure 12This is a schematic diagram illustrating the identification of different aging levels of R0 according to one embodiment;

[0049] Figure 13 According to an embodiment of R p A schematic diagram illustrating the identification of different aging levels;

[0050] Figure 14 According to an embodiment of C p A schematic diagram illustrating the identification of different aging levels;

[0051] Figure 15 According to an embodiment of U ocv A schematic diagram illustrating the identification of different aging levels;

[0052] Figure 16 This is a schematic diagram illustrating the charging capacity estimation according to one embodiment. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0054] Some embodiments of the present invention provide a joint estimation method for the state of charge and state of health of an energy storage battery. The method uses an extended Kalman filter algorithm to identify parameters of the equivalent circuit model of the energy storage battery, estimate the state of charge, and estimate the state of health.

[0055] like Figure 1 As shown, the method may include steps 101 to 106.

[0056] Step 101: Establish the equivalent circuit model of the energy storage battery.

[0057] When the energy storage battery is a lithium-ion battery, the charging and discharging process involves a large amount of chemical reaction information. Dozens of chemical reactions occur internally during charging, and the amount of chemically active components directly affects the battery's capacity. The chemical reaction rate is closely linked to the battery's internal resistance. While establishing a precise chemical model for the energy storage battery is important, the numerous complex chemical reactions are difficult to model, and the inclusion of nearly 50 parameters to be identified significantly increases the difficulty of parameter identification. Therefore, considering the complexity and accuracy of the energy storage battery model, a first-order circuit model is established as the equivalent circuit model for the energy storage circuit. For example, this first-order circuit model can be as follows: Figure 2 As shown.

[0058] Step 102: Perform parameter identification on the equivalent circuit model of the energy storage battery to obtain estimated values ​​of the parameters of the equivalent circuit model of the energy storage battery.

[0059] In some embodiments, such as Figure 3 As shown, the implementation method of step 102 may include steps 301 to 303.

[0060] Step 301: Obtain the parameters of the equivalent circuit model of the energy storage battery based on the equivalent circuit model of the energy storage battery.

[0061] In some embodiments, when the equivalent circuit model of the energy storage battery is a first-order circuit model, such as Figure 4 As shown, the implementation method of step 301 may include steps 401 to 403.

[0062] Step 401: Obtain the KVL equations of the first-order circuit model based on the first-order circuit model and Kirchhoff's laws.

[0063] According to Kirchhoff's laws, the KVL equation for the first-order circuit model is shown in equation (1).

[0064] (1)

[0065] In equation (1), U represents the terminal voltage value, U OCV U represents the open-circuit voltage, i represents the current excitation, R0 represents the battery internal resistance, and U represents the open-circuit voltage. p R is the polarization voltage. p For polarization internal resistance, C p Here, t represents the polarization capacitance, and t represents time.

[0066] Step 402: Discretize the KVL equations of the first-order circuit model and obtain the terminal voltage expression of the first-order circuit model.

[0067] Discretizing the KVL equations of the first-order circuit model yields equation (2), where U k U k-1 Let U be the discrete form of time k and time k-1. For U OCV The discrete form at time k, , U respectively p Discrete form of time k and time k-1; i k i k-1 These are the discrete forms of current i at time k and time k-1, respectively.

[0068] (2)

[0069] in, This represents the sampling interval between time k and time k-1.

[0070] The terminal voltage expression for a first-order circuit model is given by equation (3).

[0071] (3)

[0072] In equation (3), b1, b0, a and U OCV These are the parameters of a first-order circuit model.

[0073] Step 403: Obtain the parameters of the first-order circuit model based on the terminal voltage expression of the first-order circuit model; the parameters of the equivalent circuit model of the energy storage battery include the parameters of the first-order circuit model and the open-circuit voltage.

[0074] In equation (3), the parameters of the first-order circuit model are shown in equation (4).

[0075] (4)

[0076] Step 302: While the energy storage battery is charging, obtain the measured values ​​of the battery's terminal voltage and current.

[0077] In some embodiments, a voltage sensor can be used to obtain the measured value of the terminal voltage of the energy storage battery, and a current sensor can be used to obtain the measured value of the current of the energy storage battery.

[0078] Step 303: Based on the measured values ​​of the terminal voltage and current of the energy storage battery, obtain the estimated values ​​of the parameters of the equivalent circuit model of the energy storage battery through the extended Kalman filter algorithm.

[0079] For example, the measured current value can be used as the input to the extended Kalman filter algorithm. The algorithm obtains estimated parameters of the equivalent circuit model of the energy storage battery, and then corrects these estimated parameters based on the measured terminal voltage of the energy storage battery. In some embodiments, such as... Figure 5 As shown, the implementation method of step 303 may include steps 501 to 503.

[0080] Step 501: Based on the parameter identification model of the equivalent circuit model of the energy storage battery, establish the state prediction equation of the parameter of the equivalent circuit model of the energy storage battery and the measurement equation of the parameter of the equivalent circuit model of the energy storage battery.

[0081] The parameter identification ARX model is shown in equation (5).

[0082] (5)

[0083] In equation (5), y k The measured value of the terminal voltage, φk =[i k i k- U k- ] represents a discrete input vector, θ k Let θ be the parameter vector to be identified. k =[b1, b0, a, U] OCV ] T k represents the current sampling time. In the following text, the superscripts k or k-1 indicate the discrete form of the parameter or variable at time k or k-1, which will not be elaborated further.

[0084] The state prediction equations for the parameters of the equivalent circuit model of the energy storage battery are shown in equation (6).

[0085] (6)

[0086] In equation (6), Identify process noise in the state equations for parameters. Let f1 represent the Gaussian normal distribution, Q1 be the process covariance of parameter identification, and f1 represent the mapping relationship between input and output in the parameter identification process model.

[0087] The measurement equations for the parameters of the equivalent circuit model of the energy storage battery are shown in equation (7).

[0088] (7)

[0089] In equation (7), To identify the observation noise in the state equation for the parameters, the subscript 1 indicates the first-level EKF1 operation, which conforms to a Gaussian normal distribution. R1 is the covariance of the parameter identification observations. h1 represents the mapping relationship between the input and output in the parameter identification observation model.

[0090] Step 502: Based on the extended Kalman filter algorithm, the measured values ​​of the input voltage and current are combined with the state prediction equation of the parameters of the equivalent circuit model of the energy storage battery and the measurement equation of the parameters of the equivalent circuit model of the energy storage battery. Prediction is performed through prior estimation, and the Kalman filter gain of the identification model of the parameters of the equivalent circuit model of the energy storage battery is calculated.

[0091] Step 503: According to the extended Kalman filter algorithm, update the parameters of the equivalent circuit model of the energy storage battery by updating the Kalman filter gain and posterior estimation covariance of the identification model. Use the posterior estimation covariance to correct the current parameter estimate and obtain the estimated value of the parameters of the equivalent circuit model of the energy storage battery.

[0092] The following embodiments provide a detailed explanation of the methods in steps 502 and 503.

[0093] Forecast section:

[0094] The prior estimate is shown in equation (8).

[0095] (8)

[0096] The prior estimate of the covariance is shown in equation (9).

[0097] (9)

[0098] Updated section:

[0099] The updated Kalman filter gain is shown in equation (10).

[0100] (10)

[0101] In equations (8)-(10), the superscript k- indicates the left limit value of the parameter or variable at time k. The relevant definitions below are consistent with this, and will not be repeated hereafter. Let be the prior estimate of the parameter at time k. For the prior estimate of covariance in the parameter estimation process, This represents the Kalman filter gain of the EKF algorithm at this layer. This is the transformation vector of the EKF algorithm at this layer. for The inverse vector of , R1 is the parameter identification observation covariance.

[0102] From the Kalman filter gain calculation formula, it can be seen that... and As can be seen from equation (9), It is related to both Q1 and R1. By adjusting... The value reaches the optimal value for estimation and measurement.

[0103] In equation (10), Calculations show that: , , , The measurement noise variance R1 is obtained from the steady-state measurement of the battery voltage. Q1 represents the adjustment parameter based on EKF. Q1 is usually a trade-off between the ability to track parameter changes and noise attenuation.

[0104] The updated posterior estimate of covariance is shown in equation (11).

[0105] (11)

[0106] In equation (11), I is the identity matrix. The updated posterior covariance can be used to test the accuracy of the current filtered estimate and apply it to the prediction at the next time step.

[0107] The input is fed into the EKF model, and time updates, state updates, and corrected estimates are performed, as shown in Equation (12).

[0108] (12)

[0109] Step 103: Determine if the energy storage battery's charge is greater than 0; if yes, proceed to step 104; otherwise, proceed to step 105.

[0110] Step 104: Using the newly established offline battery OCV-SOC model, based on the estimated values ​​of the parameters of the equivalent circuit model of the energy storage battery, the new battery OCV-SOC model, and the extended Kalman filter algorithm, obtain the estimated value of the actual charging capacity.

[0111] In some embodiments, such as Figure 6 As shown, the implementation method of step 104 may include steps 601 to 603.

[0112] Step 601: Obtain the state prediction equation for the actual charge amount of the energy storage battery when it is in the charging state.

[0113] During battery charging, the amount of charge can be expressed by the integral expression of the current, as shown in equation (13). SOC(t) is the time expression of SOC, and SOC(0) is the initial time value of SOC. For lithium battery charging efficiency, C N This refers to the nominal capacity of the lithium battery. Let i be the integral expression for the current i over the time interval [0, t].

[0114] (13)

[0115] The dynamic charging capacity during the battery charging process is further obtained as shown in Equation (14), where ΔC is the amount of charge input, which is a dynamically changing value.

[0116] (14)

[0117] Take the actual charge ΔC and the polarization circuit voltage U p The state variables x = [ΔC, U] are composed of... p ] T The state-space expression is formed by the ampere-hour integral equation of the charging state and the voltage equation of the equivalent circuit model, and random process noise and measurement noise are introduced, as shown in equation (15).

[0118] (15)

[0119] The matrix form of equation (15) is shown in equation (16).

[0120] (16)

[0121] In equation (16), , for and Differential form of random perturbation It is a state vector A random Gaussian perturbation process with zero mean conforms to a Gaussian covariance matrix. , where q Q q U They are respectively , The corresponding Gaussian covariance matrix.

[0122] Step 602: Using the offline-established new battery OCV-SOC model, and based on the estimated values ​​of the parameters of the new battery OCV-SOC model and the equivalent circuit model of the energy storage battery, establish a measurement model for the terminal voltage.

[0123] The new battery OCV-SOC model can replace the fixed OCV-SOC curve model used in existing technologies.

[0124] According to the new battery OCV-SOC model, the SOC prediction model is shown in equation (17).

[0125] (17)

[0126] In equation (17), A k B k These are the state-space coefficient matrix and the input coefficient matrix of the discrete system, respectively. Each coefficient matrix can be expressed as equation (17-1). The measurement equation is calculated as shown in equation (18).

[0127] (17-1)

[0128] Calculations show that This step connects the first-layer EKF parameter identification model with the current layer EKF online power estimation model, allowing the identified parameters to be obtained from the parameter identification module and used to estimate the power and state of charge under the identified parameter state.

[0129] The measurement model for the terminal voltage is established as shown in equation (18), with random noise. It is the scalar noise of the measurement matrix, which is also independent and conforms to a Gaussian noise process with variance R².

[0130] (18)

[0131] In this step, it is worth noting that the estimated battery state quantity is actually the battery's current charge change capacity, equivalent to a real-time online battery capacity estimator. Corresponding to the new battery Given the curve relationship and the nominal capacity of the new battery, we can... Transform into Proceed here EKF The algorithm predicts the current battery level when the health status SOH=1.

[0132] Step 603: Based on the extended Kalman filter algorithm, the measured values ​​of the input voltage and current, and the estimated values ​​of the parameters of the equivalent circuit model of the energy storage battery are used to make a prediction through prior estimation, and the Kalman filter gain of the charging capacity estimate is calculated; based on the extended Kalman filter algorithm, the Kalman filter gain of the charging capacity estimate is updated through posterior estimation, and the posterior estimation covariance is updated. The posterior estimation covariance is used to correct the current capacity estimate, and the estimated value of the actual charging capacity is obtained.

[0133] The following embodiments provide a detailed explanation of the method in step 603.

[0134] Forecast section:

[0135] The prior estimate is shown in equation (19).

[0136] (19)

[0137] The prior estimate of the covariance matrix is ​​shown in equation (20).

[0138] (20)

[0139] In equation (20), Q2 is the state perturbation covariance matrix; in equations (18)-(20), This is the prior estimate of the state vector at time k, the left-hand limit instant in the k-state, based on the state vector estimate from the previous time step. Compared with the current i at the previous moment k-1 Calculation yielded A k-1 B k-1 Let the discrete state-space coefficient matrix and input coefficient matrix be the values ​​from the previous time step. For the prior estimate of covariance in the parameter estimation process, Let be the posterior estimated covariance under state k-1.

[0140] Updated section:

[0141] The Kalman filter gain matrix for pre-estimation of state of charge is shown in equation (21).

[0142] (twenty one)

[0143] The vector form is calculated as shown in equation (22).

[0144] (twenty two)

[0145] The posterior covariance matrix of the state estimation error As shown in equation (23).

[0146] (twenty three)

[0147] State update, posterior estimation As shown in equation (24).

[0148] (twenty four)

[0149] This represents the Kalman filter gain of the EKF algorithm at this layer. Let be the transformation vector of the EKF algorithm at this layer, which is a nonlinear output gradient model with respect to the prior state estimate. It can be obtained by taking the partial derivative of the observation equation with respect to the state vector. for The inverse vector of .

[0150] In equation (24), The prior prediction error used for observation correction can be obtained by measuring the steady-state value of the battery terminal voltage. The EKF algorithm adjusts the state estimate by adjusting the noise covariance matrix. The diagonal elements of its covariance matrix are the desired state perturbation variance values, which can be regarded as the ideal value between the tracking ability of this layer estimator and the noise suppression in the estimated state.

[0151] Step 105: Establish an OCV aging model; and based on the OCV aging model and the extended Falman filter algorithm, obtain an estimate of the available capacity of the energy storage battery.

[0152] In some embodiments, the method for establishing an OCV aging model may include: in a fully discharged state, obtaining the open-circuit voltage of the energy storage battery at the moment of charging as a data sample, and fitting the data sample using the least squares method to obtain the OCV aging model.

[0153] For example, experimental data from battery B0005 in the NASA battery dataset were selected, and the initial charging voltage at each fully discharged state, i.e., the battery open-circuit voltage at the moment of charging, was taken as the data sample to study the changing trend of the fully discharged charging voltage under different aging conditions on the lithium battery life scale. The data sample was interpolated and fitted to obtain the full life cycle model of the OCV aging scale. The fitting is a 5th order polynomial. The fitting curve was selected using the principle of minimizing the sum of squared deviations, and the polynomial equation method was used for fitting. The fitting formula is shown in equation (25).

[0154] (25)

[0155] In equation (25), the fitted y value is the fully discharged open-circuit voltage U. OCV0 x is the remaining usable battery capacity C under different charge-discharge cycles. available , , , The coefficients represent the fitting coefficients, and n is the order of the fitting polynomial. The fitting curve is shown below. Figure 7 As shown.

[0156] In this case, in some embodiments, such as Figure 8 As shown, the implementation method of step 105 may include steps 801 to 803.

[0157] Step 801: Establish a prediction equation for the available capacity of the energy storage battery.

[0158] The SOH prediction equation is shown in equation (26).

[0159] (26)

[0160] Step 802: Based on the OCV aging model, establish a terminal voltage observation model for the usable capacity of the energy storage battery.

[0161] The observation model for the available capacity of the energy storage battery is shown in Equation (27).

[0162] (27)

[0163] Therefore, the EKF prediction and observation model expression for this layer is shown in equation (28).

[0164] (28)

[0165] The remaining usable capacity of the battery C available Abbreviated as c, , It is the discrete expression for the available capacity of the state variable at time k and time k-1. , These represent the perturbations in the prediction and observation equations, respectively, and are consistent with the process noise and observation noise covariances Q3 and R3 of the EKF layer. k It represents the battery terminal voltage observation function, and is a state vector. Input vector The function expression.

[0166] Step 803: Based on the extended Kalman filter algorithm, the measured values ​​of the input voltage and current are used to introduce the aging model of the OCV using partial differential equations. Prediction is performed through prior estimation, and the Kalman filter gain for the estimated available capacity is calculated. Based on the extended Kalman filter algorithm, the Kalman filter gain for the estimated available capacity and the covariance of the posterior estimation are updated through posterior estimation. The posterior estimation covariance is used to correct the current available capacity, and the estimated value of the available capacity of the energy storage battery is obtained.

[0167] The following embodiments provide a detailed explanation of the method in step 803.

[0168] Forecast section:

[0169] The prior estimate of the state is shown in equation (29).

[0170] (29)

[0171] The prior covariance matrix of the state estimation error is shown in equation (29).

[0172] (30)

[0173] In equation (30), Q3 is the state perturbation covariance matrix. This represents the left-limit instantaneous state estimate at time k in a discrete system, which is equal to the state estimate at the previous time k-1. For the parameter estimation process in state k-1, the prior estimate of the covariance is given. Let be the posterior estimated covariance under state k-1.

[0174] Updated section:

[0175] The updated Kalman filter gain matrix is ​​shown in equation (31).

[0176] (31)

[0177] Let be the Kalman filter gain of the EKF algorithm at this layer. In equation (31), It refers to the estimated state near the prior state estimate. The nonlinear output gradient model, mathematically speaking, estimates the gradient equation near the state, due to the remaining capacity c.k With terminal voltage U k The mapping relationship between them is not explicit, therefore partial differential equations are used to derive the functional relationship between them, reflecting the changes in state estimation. The calculation yields:

[0178] (32)

[0179] Based on the transformation vector Equation (32) is defined to calculate the transformation vector in the available capacity estimation step. The state matrix x in the observation equation k- The state estimation matrix in the state of charge estimation step contains an unclear relationship between the current charge ∆C and the remaining available capacity c. Therefore, partial differential equations are used to derive the functional relationship between the two to reflect the changes in state estimation. Based on the established OCV aging model, the open-circuit voltage is selected as the intermediate parameter of the partial differential equation. Therefore, in the calculation of the gradient coefficient, it is necessary to clarify the relationship between the current charge and the open-circuit voltage and the relationship between the open-circuit voltage and the available charge. The relationship between the current charge and the open-circuit voltage and the relationship between the open-circuit voltage and the available charge selected by the EKF in this layer is shown in equation (33).

[0180] (33)

[0181] The left side of equation (33) expresses the state quantity. With open circuit voltage Partial differential relation, fully discharged open-circuit voltage With available capacity c k The partial differential equation on the right is a mathematical approximation of the partial differential equation. The gradient equation for the offline OCV-SOC aging curve is introduced. Substituting equation (33) into equation (32) calculates the observation matrix. .

[0182] Update the posterior covariance matrix As shown in equation (34).

[0183] (34)

[0184] State update, posterior estimation As shown in equation (35).

[0185] (35)

[0186] Step 106: Based on the estimated actual charge amount and the estimated available capacity of the energy storage battery, obtain the estimated values ​​of SOC and SOH.

[0187] Numerous studies have shown a coupling relationship between the battery's state of charge (SOC) and state of health (SOH). In some embodiments, such as... Figure 9 As shown, the implementation method of step 106 may include steps 901 to 902.

[0188] Step 901: Obtain the initial value of the remaining capacity of the energy storage battery in the charging state, and obtain the estimated value of SOC based on the estimated value of the available capacity of the energy storage battery, the initial value of the remaining capacity of the energy storage battery in the charging state, and the estimated value of the actual charging capacity.

[0189] For example, SOC is defined from the perspective of capacity, as shown in Equation (36).

[0190] (36)

[0191] In equation (36), C0 is the initial value of the remaining battery capacity in the charging state, ∆C is the estimated value of the actual charged capacity, and C available This represents the initial value of the remaining charge of the energy storage battery when it is charging.

[0192] Step 902: Obtain the capacity value of the new battery, and obtain the estimated value of SOH based on the capacity value of the new battery and the estimated value of the available capacity of the energy storage battery.

[0193] For example, SOH is defined from a capacity perspective as shown in equation (37).

[0194] (37)

[0195] In equation (37), C new The capacity of the new battery is a known value.

[0196] The embodiments of this invention provide a joint estimation method for the state of charge (SOC) and state of health (SOH) of an energy storage battery. This method establishes a three-layer extended Kalman filter (EKF) algorithm to achieve first-order equivalent circuit parameter identification, SOC estimation, and SOH estimation for the lithium battery. Specifically, the EKF1 algorithm solves the problem of offline identification using traditional least squares methods and sets the open-circuit voltage (OCV) as a state variable, obtaining real-time tracked dynamic data of the battery circuit's open-circuit voltage. This data is then input into the lower-level EKF model for online state estimation, making the parameter identification closer to reality and achieving dynamic parameter response under dynamic operating conditions. Since this paper considers the impact of aging on OCV-SOC, a U... OCV0 -C available An open-circuit voltage aging model is used to correct aging errors caused by a fixed OCV-SOC curve, U OCV0 -C availableThe open-circuit voltage aging model is used in the EKF3 model for estimating the usable capacity after aging. Therefore, when the battery is in a fully discharged state, once charging begins, the EKF3 model re-estimates the usable capacity after aging, updating the battery's State of Health (SOH). The updated usable battery capacity is then applied to the EKF2 model for estimating the State of Charge (SOC), correcting the errors caused by aging. EKF2 uses the open-circuit voltage curve of the new battery to estimate the real-time SOC value. The SOC under the new state is then superimposed with the aging correction. Experiments verify that the accuracy of SOC estimation is effectively improved. Simultaneously, noise covariance is used to update the state vector and time, avoiding the significant impact of inaccurate initial values.

[0197] The following is an explanation of the experimental verification of the above method.

[0198] Use such during verification Figure 10 and Figure 11 The pulse charge-discharge excitation shown was applied for 27,600 charge-discharge cycles until the end of the battery's lifespan. Experimental results demonstrate that, using the joint estimation method for the state of charge and state of health of the energy storage battery provided in the embodiments of this invention, the parameter identification error affected by aging is effectively corrected. The parameter identification results differ for different degrees of aging, as shown in the figures below. Figures 12 to 15 As shown. Figure 12 The different aging levels of R0 are shown. Figure 13 R was shown p Different aging levels can be identified. Figure 14 Showing C p Different degrees of aging were identified. Figure 15 U was displayed ocv The identification of different aging levels, combined with the correction of the current battery capacity based on battery aging, effectively reduces the error in state of charge estimation. Essentially, it calibrates the current capacity estimate, reducing the battery capacity estimation error to 0.08%. The capacity estimation results are as follows: Figure 16 As shown, the OCV-SOC aging dynamic model proposed in this invention significantly improves the estimation error of the state of charge caused by aging, and the estimation model output results are more realistic and accurate.

[0199] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0200] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for jointly estimating the state of charge and state of health of an energy storage battery, characterized in that, include: Establish an equivalent circuit model for the energy storage battery; The parameters of the equivalent circuit model of the energy storage battery are identified to obtain estimated values ​​of the parameters of the equivalent circuit model of the energy storage battery. When the energy storage battery's charge is greater than 0, a new battery OCV-SOC model is established offline. Based on the estimated parameters of the energy storage battery's equivalent circuit model, the new battery OCV-SOC model, and the extended Kalman filter algorithm, an estimated value of the actual charged charge is obtained. When the energy storage battery's charge is equal to 0, an OCV aging model is established, and based on the OCV aging model and the extended Kalman filter algorithm, an estimated value of the energy storage battery's usable capacity is obtained. Based on the estimated actual charge and the estimated available capacity of the energy storage battery, the estimated values ​​of SOC and SOH are obtained. Specifically, establishing the OCV aging model includes: In the fully discharged state, the open-circuit voltage of the energy storage battery at the moment of charging is obtained as a data sample, and the least squares method is used to fit the data sample to obtain the OCV aging model.

2. The method for jointly estimating the state of charge and state of health of an energy storage battery according to claim 1, characterized in that, The parameter identification of the equivalent circuit model of the energy storage battery to obtain estimated values ​​of the parameters of the equivalent circuit model specifically includes: Based on the equivalent circuit model of the energy storage battery, obtain the parameters of the equivalent circuit model of the energy storage battery; When the energy storage battery is in charging mode, the measured values ​​of the terminal voltage and current of the energy storage battery are obtained; Based on the measured values ​​of the terminal voltage and current of the energy storage battery, the estimated values ​​of the parameters of the equivalent circuit model of the energy storage battery are obtained through the extended Kalman filter algorithm.

3. The method for jointly estimating the state of charge and state of health of an energy storage battery according to claim 2, characterized in that, The equivalent circuit model of the energy storage battery is a first-order circuit model. The parameters of the equivalent circuit model of the energy storage battery are obtained based on the following: The KVL equations of the first-order circuit model are obtained based on the first-order circuit model and Kirchhoff's laws. Discretize the KVL equations of the first-order circuit model to obtain the terminal voltage expression of the first-order circuit model; Based on the terminal voltage expression of the first-order circuit model, the parameters of the first-order circuit model are obtained, and the parameters of the equivalent circuit model of the energy storage battery include the open-circuit voltage of the first-order circuit model.

4. The method for jointly estimating the state of charge and state of health of an energy storage battery according to claim 2, characterized in that, Based on the measured values ​​of the terminal voltage and current of the energy storage battery, the estimated values ​​of the parameters of the equivalent circuit model of the energy storage battery are obtained through the extended Kalman filter algorithm, specifically including: Based on the parameter identification model of the equivalent circuit model of the energy storage battery, state prediction equations and measurement equations of the parameters of the equivalent circuit model of the energy storage battery are established. According to the extended Kalman filter algorithm, the measured values ​​of the terminal voltage and the measured values ​​of the current are input. The state prediction equation of the parameters of the equivalent circuit model of the energy storage battery and the measurement equation of the parameters of the equivalent circuit model of the energy storage battery are combined. Prediction is made through prior estimation, and the Kalman filter gain of the identification model of the parameters of the equivalent circuit model of the energy storage battery is calculated. According to the extended Kalman filter algorithm, the parameters of the equivalent circuit model of the energy storage battery are updated by posterior estimation. The Kalman filter gain and posterior estimation covariance of the identification model are updated, and the posterior estimation covariance is used to correct the current parameter estimate, so as to obtain the estimated value of the parameters of the equivalent circuit model of the energy storage battery.

5. The method for jointly estimating the state of charge and state of health of an energy storage battery according to claim 2, characterized in that, Using a newly established offline battery OCV-SOC model, and based on the estimated parameters of the equivalent circuit model of the energy storage battery, the new battery OCV-SOC model, and the extended Kalman filter algorithm, the estimated value of the actual charging capacity is obtained, specifically including: Obtain the state prediction equation for the actual charge amount of the energy storage battery when it is in a charging state. A measurement model for the terminal voltage is established by using a new battery OCV-SOC model built offline and based on the estimated values ​​of the parameters of the new battery OCV-SOC model and the equivalent circuit model of the energy storage battery. According to the extended Kalman filter algorithm, the measured values ​​of the terminal voltage and the current, as well as the estimated values ​​of the parameters of the equivalent circuit model of the energy storage battery, are input. Prediction is performed through prior estimation, and the Kalman filter gain of the estimated charging capacity is calculated. According to the extended Kalman filter algorithm, the Kalman filter gain of the estimated charging capacity and the covariance of the posterior estimation are updated through posterior estimation. The posterior estimation covariance is used to correct the current capacity estimate and obtain the estimated value of the actual charging capacity.

6. The method for jointly estimating the state of charge and state of health of an energy storage battery according to claim 2, characterized in that, Based on the OCV aging model and the extended Kalman filter algorithm, the estimated value of the usable capacity of the energy storage battery is obtained specifically including: Establish a predictive equation for the usable capacity of energy storage batteries; Based on the OCV aging model, an equation for measuring the terminal voltage of the usable capacity of the energy storage battery is established. According to the extended Kalman filter algorithm, the measured values ​​of the terminal voltage and the current are input, and the OCV aging model is introduced using the partial differential equation method. Prediction is performed through prior estimation, and the Kalman filter gain for estimating the available capacity of the energy storage battery is calculated. According to the extended Kalman filter algorithm, the Kalman filter gain and the posterior estimation covariance for estimating the available capacity of the energy storage battery are updated through posterior estimation. The posterior estimation covariance is used to correct the current available capacity, and the estimated value of the available capacity of the energy storage battery is obtained.

7. The method for jointly estimating the state of charge and state of health of an energy storage battery according to claim 1, characterized in that, Based on the estimated actual charge amount and the estimated available capacity of the energy storage battery, obtaining the estimated values ​​of SOC and SOH specifically includes: Obtain the initial value of the remaining charge of the energy storage battery in the charging state, and obtain the estimated value of SOC based on the estimated value of the available capacity of the energy storage battery, the initial value of the remaining charge of the energy storage battery in the charging state, and the estimated value of the actual charge input. Obtain the capacity value of the new battery, and obtain the estimated value of the SOH based on the capacity value of the new battery and the estimated value of the available capacity of the energy storage battery.

Images