Lithium battery soc estimation method based on extended kalman filter

Abstract

The application provides a lithium battery SOC estimation method based on extended Kalman filtering, comprising the following steps: constructing an equivalent circuit model of a lithium battery; establishing a discrete nonlinear system state and an observation equation according to the equivalent circuit model; obtaining a linear system state and an observation equation required by an extended Kalman filtering algorithm according to the discrete nonlinear system state and the observation equation; updating a prediction and correction part of the extended Kalman filtering algorithm; judging whether the updating process of the prediction and correction part of the extended Kalman filtering algorithm is completed; if yes, reconstructing the linear system state and the observation equation required by the extended Kalman filtering algorithm; and estimating the SOC of the lithium battery according to the reconstructed linear system state and the observation equation required by the extended Kalman filtering algorithm. The application can simultaneously consider the convergence speed of the SOC estimation value of the lithium battery in the initial estimation stage and the smoothness of the SOC estimation value waveform of the lithium battery in the stable estimation stage after converging to the allowable error range.

Description

Technical Field

[0001] This invention relates to the field of power lithium battery SOC technology, specifically to a lithium battery SOC estimation method based on extended Kalman filtering. Background Technology

[0002] Currently, the main strategies for estimating the State of Charge (SOC) of lithium batteries include: ampere-hour integration method, open-circuit voltage method, neural network prediction, and extended Kalman filter (EPF) algorithm. Among these, EPF is an improved filtering strategy for nonlinear systems based on the Kalman algorithm and is an effective technique for estimating battery SOC. However, due to problems such as non-Gaussian noise interference, errors in the battery equivalent circuit model, and deviations between the initial system state and the actual state, the SOC estimation results of EPF often do not converge easily, and in severe cases, even diverge.

[0003] To address these issues, some scholars have proposed introducing a Sage-Husa estimator into the Extended Kalman Filter (EKF) algorithm. Based on the maximum a posteriori (MAP) criterion, the Sage-Husa estimator uses a recursive approach to estimate the statistical characteristics of noise in real time, thereby improving the accuracy of SOC estimation in the EKF algorithm. However, this method relies heavily on observation residuals, and when there are deviations between the initial system state and the actual state, it can significantly impact the convergence speed in the initial estimation phase. Furthermore, other scholars have proposed combining the EKF algorithm with a rolling time-domain window, integrating window information over a certain time period to estimate the current SOC. By establishing an arrival function, the prediction is transformed into an optimization problem, and the EKF algorithm's solution approach provides an approximately optimal SOC estimate. However, when the amount of observation data is too large, the estimation results of this strategy are unstable. Therefore, existing strategies for estimating lithium-ion battery SOC are still constrained by non-Gaussian noise interference and are prone to convergence difficulties. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a lithium battery SOC estimation method based on extended Kalman filtering. This method can simultaneously consider the convergence speed of the lithium battery SOC estimate in the initial estimation stage and the smoothness of the lithium battery SOC estimate waveform in the stable estimation stage after convergence within the allowable error range. Furthermore, it can effectively avoid interference from non-Gaussian noise and enhance robustness.

[0005] The technical solution adopted in this invention is as follows:

[0006] A lithium battery SOC estimation method based on extended Kalman filtering includes the following steps: constructing an equivalent circuit model of the lithium battery; establishing discrete nonlinear system state and observation equations based on the equivalent circuit model; obtaining the linear system state and observation equations required by the extended Kalman filter algorithm based on the discrete nonlinear system state and observation equations; updating the prediction and correction part of the extended Kalman filter algorithm by calculating the low-pass filtering result of the prediction and correction part and by piecewise transforming the observation covariance of different orders of magnitude in the prediction and correction part of the extended Kalman filter algorithm; determining whether the update process of the prediction and correction part of the extended Kalman filter algorithm is complete; if so, reconstructing the linear system state and observation equations required by the extended Kalman filter algorithm by introducing the maximum cross-correlation entropy criterion; estimating the SOC of the lithium battery based on the reconstructed linear system state and observation equations required by the extended Kalman filter algorithm.

[0007] According to one embodiment of the present invention, the equivalent circuit model is a Thevenin first-order RC model.

[0008] According to an embodiment of the present invention, the state and observation equations of the discrete nonlinear system are as follows:

[0009]

[0010] Where η is the charging efficiency, C N R is the rated total capacity of the lithium battery, Δt is the sampling interval, τ is the integration time constant, and R p R represents the polarization resistance of the lithium battery, w1, w2, and v represent the uncorrelated zero-mean Gaussian white noise caused by the equivalent circuit model and external interference. i U is the internal ohmic resistance of the lithium battery, i is the operating current of the lithium battery, and U is the internal resistance of the lithium battery. oc U is the open-circuit voltage. p For the cross in C p The polarization voltage at both ends, with the subscript k indicating the time step.

[0011] According to one embodiment of the present invention, the linear system state and observation equations required by the extended Kalman filter algorithm are as follows:

[0012]

[0013] Where, x k z is a state variable. k For the observed variable, w k For system noise, A k Let B be the state transition matrix. k Given the input matrix, C k The observation matrix;

[0014] Furthermore, the state transition matrix A k Input matrix B k Observation matrix C k The specific expression is:

[0015]

[0016] According to an embodiment of the present invention, the step of updating the prediction and correction part of the extended Kalman filter algorithm by calculating the low-pass filtering result of the prediction and correction part and by piecewise transforming the observation covariance of different orders of magnitude in the prediction and correction part of the extended Kalman filter algorithm specifically includes the following steps: calculating the low-pass filtering result of the observation residual in the prediction and correction part of the extended Kalman filter algorithm; determining the allowable error time point between the estimated value and the actual value of the lithium battery SOC based on the low-pass filtering result; piecewise transforming the stage transformation value of the observation covariance in the prediction and correction part of the extended Kalman filter algorithm based on the allowable error time point; and updating the observation covariance in the extended Kalman filter algorithm based on the stage transformation value.

[0017] According to one embodiment of the present invention, the prediction part is:

[0018]

[0019] Where, ε k To observe residuals;

[0020] The correction section is:

[0021]

[0022] Among them, R k To observe the covariance, K is the Kalman gain.

[0023] According to one embodiment of the present invention, the stage transformation value of the observation covariance includes a first order of magnitude value and a second order of magnitude value, and the observation covariance in the extended Kalman filter algorithm is updated according to the stage transformation value using the following formula:

[0024]

[0025] Where t0 is the initial time point, t1 is the time point with tolerance, and R s R is a value of the first order of magnitude. L It is a value of the second order of magnitude.

[0026] According to one embodiment of the present invention, the linear system state and observation equations required for the reconstructed extended Kalman filter algorithm are as follows:

[0027]

[0028] in,

[0029] According to one embodiment of the present invention, estimating the SOC of the lithium battery based on the linear system state and observation equations required by the reconstructed extended Kalman filter algorithm specifically includes the following steps: obtaining an iterative update equation set for the lithium battery based on the linear system state and observation equations required by the reconstructed extended Kalman filter algorithm; and estimating the SOC of the lithium battery based on the iterative update equation set.

[0030] According to one embodiment of the present invention, the iterative update equation system is as follows:

[0031]

[0032] Where M is a diagonal matrix, I is an identity matrix, σ is the bandwidth of the Gaussian kernel function, j is the iteration number of the iterative update equation system (1≤j≤N, N is a positive integer), and when j=N, the SOC estimation result of the lithium battery is output.

[0033] The beneficial effects of this invention are as follows:

[0034] 1) This invention, by transforming the observation covariance of different orders of magnitude in stages, can simultaneously take into account the convergence speed of the lithium battery SOC estimate in the initial estimation stage and the smoothness of the lithium battery SOC estimate waveform in the stationary estimation stage after convergence to within the allowable error range. This effectively avoids the need to solely use a large-order-of-magnitude variation of the numerical value R. L The existing problem is the slow convergence speed of lithium battery SOC estimation, and the use of a single small-order-of-magnitude transformation value R S The existing problem is that the waveform of the estimated SOC value of lithium batteries oscillates significantly.

[0035] 2) This invention uses the low-pass filtering result of the observation residual as the convergence criterion, which can quickly and efficiently determine whether the estimated value of lithium battery SOC has converged to the allowable error range.

[0036] 3) This invention estimates the SOC of lithium batteries by introducing a nonlinear regression method based on the maximum cross-correlation entropy criterion to reconstruct the linear system state and observation equations required by the extended Kalman filter algorithm. This effectively avoids interference from non-Gaussian noise and improves the robustness of the filtering algorithm. Attached Figure Description

[0037] Figure 1 This is a flowchart of a lithium battery SOC estimation method based on extended Kalman filtering according to an embodiment of the present invention;

[0038] Figure 2 This is a topology diagram of the equivalent circuit model of a lithium battery according to an embodiment of the present invention.

[0039] Figure 3 This is a voltage curve generated by a single mixed pulse power test according to an embodiment of the present invention;

[0040] Figure 4 This is a flowchart of the prediction and correction part of the updated extended Kalman filter algorithm according to an embodiment of the present invention;

[0041] Figure 5 The staged transformation observation covariance R is an embodiment of the present invention. k Schematic diagram;

[0042] Figure 6 This is a specific implementation process of a lithium battery SOC estimation method based on extended Kalman filtering, according to an embodiment of the present invention. Detailed Implementation

[0043] 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, and 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.

[0044] Figure 1 This is a flowchart of a lithium battery SOC estimation method based on extended Kalman filtering, according to an embodiment of the present invention.

[0045] like Figure 1 As shown, the lithium battery SOC estimation method based on extended Kalman filtering according to an embodiment of the present invention includes the following steps:

[0046] S1, construct the equivalent circuit model of the lithium battery.

[0047] Specifically, such as Figure 2 As shown, the equivalent circuit model can be the Thevenin first-order RC model, where U oc U is the open-circuit voltage. term R is the battery terminal voltage of the lithium battery, i is the operating current of the lithium battery, and R is the operating current of the lithium battery. i R is the internal ohmic resistance of a lithium battery. p and C p For the polarization resistance and polarization capacitance of a lithium battery, τ = R p C p U is the integration time constant. p The polarization voltage is at both ends.

[0048] Furthermore, for Figure 2The equivalent circuit model shown, i.e., the relevant parameters in Thevenin's first-order RC model, such as R... i R p C p Furthermore, the OCV-SOC characteristic curve requires parameter identification. Specifically, the hybrid pulse power test method can be used to identify R. i R p C p And the OCV-SOC characteristic curve parameters, for example, nine SOC test points can be set, respectively at 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, and 90% of the mixed pulse power. The following will use... Figure 3 Taking the voltage curve generated by a single mixed pulse power test as an example, the following explains R... i R p C p And the process of identifying the OCV-SOC characteristic curve.

[0049] Specifically, in combination Figure 3 Due to the internal ohmic resistance R of the lithium battery i The influence of this can cause a sudden drop in voltage from U1 to U2 at time t1. Therefore, the internal ohmic resistance R inside the lithium battery can be reduced. i Defined as the ratio of the voltage difference at time t1 to the discharge current value:

[0050]

[0051] Furthermore, combined Figure 3 During the time interval t3 to t4, the voltage from U4 to U5 rises slowly. Therefore, it can be concluded that U p Entering the zero-input response change cycle, U term The expression is as follows:

[0052] U term =U OC -U p0 e -t / τ (2)

[0053] Among them, U p0 As the initial polarization voltage, the least squares method combined with test data can be used to determine U. term The open-circuit voltage U is obtained by fitting the expression. oc Initial polarization voltage U p0 and the integral time constant τ.

[0054] Furthermore, combined Figure 3 During the time interval t1 to t2, the voltage from U2 to U3 decreases slowly. Therefore, it can be concluded that U p Entering the zero-state response change cycle, U term The expression is as follows:

[0055] U term =U OC -iR i -iR p (1-e -t / τ (3);

[0056] In the above formula, U term By taking the measured value U3 at time t2, the polarization resistance R of the lithium battery can be calculated. p for:

[0057]

[0058] Furthermore, based on the definition of the integration time constant τ, the polarization capacitance C of the lithium battery can be calculated. p =τ / R p .

[0059] Furthermore, combined Figure 3 Let SOC be a variable, U OC (SOC) is an unknown quantity, and U is established. OC The fitting function relationship equation between (SOC) and SOC:

[0060] U OC (SOC)=k1·SOC 5 +k2·SOC 4 +k3·SOC 3 +k4·SOC 2 +k5·SOC+k6 (5)

[0061] Therefore, the characteristic curve of OCV-SOC can be determined.

[0062] S2, establish the state and observation equations of the discrete nonlinear system based on the equivalent circuit model.

[0063] Specifically, it can be targeted at Figure 2 The equivalent circuit model shown, namely the Thevenin first-order RC model, uses the ampere-hour integral method and Kirchhoff's laws to establish the state and observation equations of the discrete nonlinear system, as shown in the following specific expressions:

[0064]

[0065] Where η is the charging efficiency, C N R is the rated total capacity of the lithium battery, Δt is the sampling interval, τ is the integration time constant, and R p R represents the polarization resistance of the lithium battery, w1, w2, and v represent the uncorrelated zero-mean Gaussian white noise caused by the equivalent circuit model and external interference. i U is the internal ohmic resistance of the lithium battery, i is the operating current of the lithium battery, and U is the internal resistance of the lithium battery.oc U is the open-circuit voltage. p For the cross in C p The polarization voltage at both ends, with the subscript k indicating the time step.

[0066] S3. Based on the discrete nonlinear system state and observation equations, the linear system state and observation equations required by the extended Kalman filter algorithm are obtained.

[0067] Specifically, the discrete nonlinear system state and observation equations can be expanded into Taylor series using the Taylor formula. Then, approximate first-order terms can be taken to obtain the linear system state and observation equations required by the extended Kalman filter algorithm.

[0068]

[0069] Where, x k z is a state variable. k For the observed variable, w k For system noise, v k Let A be the state covariance. k Let B be the state transition matrix. k Given the input matrix, C k This is the observation matrix.

[0070] In one embodiment of the present invention, the state variable x k Observed variable z k System noise w k State covariance v k The specific expression is:

[0071] x k =[SOC k U p，k ] T

[0072] z k =U term，k

[0073] w k =[w 1，k w 2，k ] T

[0074] v k =[v k ]

[0075] Among them, the system noise w k ∈(0,Q k State covariance v k ∈(0,R k ),and To observe the covariance.

[0076] In one embodiment of the present invention, the state transition matrix A k Input matrix B k Observation matrix C k The specific expression is:

[0077]

[0078] Furthermore, at the initial time point t0 when estimating the SOC of the lithium battery, the relevant parameters of the extended Kalman filter algorithm can be initialized, mainly including the initialization state. The state estimation covariance P0, and the initial observation covariance R0 set to the first order of magnitude of R s :

[0079]

[0080] Among them, R s It is a value of the first order of magnitude.

[0081] S4 updates the prediction and correction part of the extended Kalman filter algorithm by calculating the low-pass filtering result of the prediction and correction part of the extended Kalman filter algorithm and the observation covariance of different orders of magnitude in the prediction and correction part of the piecewise transformation extended Kalman filter algorithm.

[0082] The prediction part of the extended Kalman filter algorithm is as follows:

[0083]

[0084] The state variables at time k-1 State estimation covariance Based on the above formula, the prior estimate corresponding to time k can be obtained. and Further calculate the estimated values ​​of the observed variables

[0085] The correction part of the extended Kalman filter algorithm is as follows:

[0086]

[0087] In the above formula, To observe the residuals, R k Let K be the observation covariance at time k, and K be the Kalman gain. Combining the prediction and correction formulas above, it can be seen that the Kalman gain K directly affects the convergence speed of the extended Kalman filter algorithm. Therefore, controlling the Kalman gain K can directly and effectively adjust R... k .

[0088] In one embodiment of the present invention, such as Figure 4As shown, step S4 above also specifically includes the following steps:

[0089] S401, Calculate the low-pass filtering result of the observation residuals in the prediction and correction part of the extended Kalman filter algorithm.

[0090] Specifically, a first-order low-pass filter can be used to filter the observed residual ε. k Weighting can be performed, for example, the current observation residual ε can be weighted using the following formula. k Compared with the previous filtering output value E of the first-order low-pass filter k-1 Weighting is performed to obtain the current effective filter output value E. k That is, the observation residual ε k Low-pass filtering results:

[0091]

[0092] Where δ is the filter coefficient. The smaller δ is, the more stable the filter value is, but the lower the sensitivity is. The larger δ is, the higher the sensitivity is, but the more unstable the filter value is.

[0093] S402, determine the allowable error time point between the estimated value and the actual value of lithium battery SOC based on the low-pass filtering result.

[0094] First, it should be noted that when there is a deviation between the estimated and actual SOC values ​​of the lithium battery, the positive and negative states of the lithium battery SOC will not change during the initial estimation stage until the estimated lithium battery SOC converges to within the allowable error range. At this point, if the initial observation covariance R0 continues to use the first order of magnitude value R... s This can easily cause the estimated SOC of a lithium battery to oscillate around the actual value, thus leading to a positive-to-negative state transition of the lithium battery's SOC.

[0095] Therefore, the present invention can determine the time point at which the estimated SOC value of the lithium battery converges to within the allowable error using the following convergence criterion, that is, the allowable error time point t1 between the estimated SOC value and the actual value of the lithium battery:

[0096] t1=kΔt(E k E k-1 <0, k≥1) (13).

[0097] S403, based on the stage transformation values ​​of the observed covariance in the prediction and correction part of the extended Kalman filter algorithm according to the piecewise transformation of the allowable error time point.

[0098] Specifically, the stage transformation value of the observed covariance may include a value of the first order of magnitude R. s The second order of magnitude of R L Specifically, before the lithium battery SOC estimate converges to within the allowable error at time point t1, the observation covariance R...k Set to the first order of magnitude value R s After the lithium battery SOC estimate converges to within the allowable error at time point t1, the observed covariance R... k Set to the second order of magnitude value R L .

[0099] S404 updates the observation covariance in the extended Kalman filter algorithm based on the stage transformation values.

[0100] First, it should be noted that, as Figure 5 As shown, before the lithium battery SOC estimate converges to within the allowable error at time point t1, the observation covariance R... k Set to the first order of magnitude value R s For example, a small order of magnitude value, which ensures that the estimated SOC of the lithium battery quickly approaches the actual SOC of the lithium battery, and can converge to the allowable error range at the allowable error time point t1.

[0101] Furthermore, referring to Figure 5 After the lithium battery SOC estimate converges to within the allowable error range, i.e., after time point t1 when it converges to within the allowable error range, in order to reduce the dependence on the equivalent circuit model of the lithium battery and reduce the oscillation of the lithium battery SOC estimate waveform, the observation covariance R can be adjusted after time point t1. k Set to the second order of magnitude value R L That is, large-order values, specifically as follows:

[0102]

[0103] S5, determine whether the update process of the prediction and correction part of the extended Kalman filter algorithm is complete.

[0104] Specifically, determine R k Set whether to start from the first order of magnitude value R s Transformed to the second order of magnitude of R L .

[0105] S6, if so, then the maximum cross-correlation entropy criterion information is introduced to reconstruct the linear system state and observation equations required for the extended Kalman filter algorithm.

[0106] Specifically, a nonlinear regression method based on the maximum cross-correlation entropy criterion can be introduced to reconstruct the linear system state and observation equations required by the extended Kalman filter algorithm.

[0107] S7. Estimate the SOC of the lithium battery based on the linear system state and observation equations required by the reconstructed extended Kalman filter algorithm.

[0108] Specifically, the iterative update equations of the lithium battery can be obtained from the linear system state and observation equations required by the reconstructed extended Kalman filter algorithm, and then the SOC of the lithium battery can be estimated based on the iterative update equations.

[0109] The linear system state and observation equations required for the reconstructed extended Kalman filter algorithm are as follows:

[0110]

[0111] in, Its covariance matrix is:

[0112]

[0113] Among them, T k Yes The result of Cholesky decomposition is then multiplied by both sides of the above equation. We can obtain the following formula:

[0114] D k =G(x) k )+e k (17)

[0115] in, And use d i，k D represents k The i-th row, g i,k G(x) represents k If the i-th row of ) is e i，k =d i，k -g i，k .

[0116] Furthermore, it can make The initial value of iteration number j can be set to 1, resulting in the corresponding iterative update equation system:

[0117]

[0118] Where M is a diagonal matrix, I is the identity matrix, σ is the bandwidth of the Gaussian kernel function, and j is the iteration number of the iterative update equation system (1≤j≤N, where N is a positive integer). When j=N, the SOC estimation result of the lithium battery is output, i.e., the result is obtained.

[0119] In summary, the lithium battery SOC estimation method based on extended Kalman filtering of this invention can be implemented through a time iteration process. To ensure the accuracy of the lithium battery SOC estimation method based on extended Kalman filtering of this invention, the SOC estimation result of the lithium battery is output, i.e., the result is obtained... Then, it can be determined whether the time iteration k satisfies k = Q (Q is a positive integer). If not, k + 1 and repeat steps S3-S7 above until k = Q, thus ending the process of the lithium battery SOC estimation method based on extended Kalman filter of this invention. The following will combine... Figure 6 The specific implementation process of the lithium battery SOC estimation method based on extended Kalman filtering of the present invention is further explained.

[0120] Specifically, such as Figure 6 As shown, it includes the following steps:

[0121] S01, at the start time t0 of SOC estimation, initialize the state. State estimation covariance P0 And set the initial observation covariance R0 to a small order of magnitude R S ;

[0122] S02, Update the observation matrix C according to the above formula (8). k ;

[0123] S03, update the prediction part according to the above formula (10).

[0124] S04, calculate the observation residual ε according to the above formula (11). k ;

[0125] S05, calculate the observation residual ε according to the above formula (12). k The low-pass filter result E k ;

[0126] S06, determine the time point at which the estimated SOC value of the lithium battery converges to within the allowable error, and determine (E k E k-1 Check if <0)&(k≥1) is true. If yes, proceed to step S07; otherwise, proceed to step S08.

[0127] S07, R k Set to the second order of magnitude value R L ;

[0128] S08, determine R k Is it a value of the first order of magnitude R? s If yes, proceed to step S09; otherwise, proceed to step S11.

[0129] S09, update the correction part according to the above formula (11).

[0130] S10, Output the optimal estimation result

[0131] S11, set j to 1, and

[0132] S12, respectively for R k and P k / k-1 Perform Cholesky decomposition;

[0133] S13, iteratively update according to the above formula (18).

[0134] S14, set j = j + 1;

[0135] S15, determine whether the number of iterations j has reached the set total number of iterations N. If yes, proceed to step S16; otherwise, proceed to step S13.

[0136] S16, Settings And proceed to step S10;

[0137] S17 determines whether the time iteration k is complete. If yes, the process ends; otherwise, k = k + 1 is executed and the process returns to step S02.

[0138] The beneficial effects of this invention are as follows:

[0139] 1) This invention, by transforming the observation covariance of different orders of magnitude in stages, can simultaneously take into account the convergence speed of the lithium battery SOC estimate in the initial estimation stage and the smoothness of the lithium battery SOC estimate waveform in the stationary estimation stage after convergence to within the allowable error range. This effectively avoids the need to solely use a large-order-of-magnitude variation of the numerical value R. L The existing problem is the slow convergence speed of lithium battery SOC estimation, and the use of a single small-order-of-magnitude transformation value R S The existing problem is that the waveform of the estimated SOC value of lithium batteries oscillates significantly.

[0140] 2) This invention uses the low-pass filtering result of the observation residual as the convergence criterion, which can quickly and efficiently determine whether the estimated value of lithium battery SOC has converged to the allowable error range.

[0141] 3) This invention estimates the SOC of lithium batteries by introducing a nonlinear regression method based on the maximum cross-correlation entropy criterion to reconstruct the linear system state and observation equations required by the extended Kalman filter algorithm. This effectively avoids interference from non-Gaussian noise and improves the robustness of the filtering algorithm.

[0142] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. "A plurality of" means two or more, unless otherwise explicitly specified.

[0143] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0144] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "over," and "on top" of the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0145] In the description of this specification, the 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 present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

Claims

1. A lithium battery SOC estimation method based on extended Kalman filtering, characterized in that, The method comprises the following steps: constructing an equivalent circuit model of a lithium battery; establishing discrete nonlinear system state and observation equations according to the equivalent circuit model; obtaining linear system state and observation equations required by an extended Kalman filter algorithm according to the discrete nonlinear system state and observation equations; updating a prediction part and a correction part of the extended Kalman filter algorithm by calculating low-pass filtering results of the prediction part and the correction part of the extended Kalman filter algorithm and piecewise transforming observation covariances of different orders of magnitude in the prediction part and the correction part of the extended Kalman filter algorithm; judging whether the updating process of the prediction part and the correction part of the extended Kalman filter algorithm is completed or not: if yes, introducing maximum cross-correlation entropy criterion information to reconstruct the linear system state and observation equations required by the extended Kalman filter algorithm; estimating the SOC of the lithium battery according to the reconstructed linear system state and observation equations required by the extended Kalman filter algorithm.

2. The lithium battery SOC estimation method based on extended Kalman filter according to claim 1, characterized in that, The equivalent circuit model is a Thevenin first-order RC model.

3. The lithium battery SOC estimation method based on extended Kalman filter according to claim 2, characterized in that, The discrete nonlinear system state and observation equations are: wherein is the charging efficiency, is the rated total capacity of the lithium battery, is the sampling interval, is the integration time constant, is the polarization resistance of the lithium battery, denotes uncorrelated zero-mean Gaussian white noise caused by the equivalent circuit model and external disturbances, is the ohmic internal resistance inside the lithium battery, is the operating current of the lithium battery, is the open circuit voltage, is the polarization voltage across the terminals, subscript k denotes the time step.

4. The lithium battery SOC estimation method based on extended Kalman filter according to claim 3, characterized in that, The linear system state and observation equations required by the extended Kalman filter algorithm are: wherein, is a state variable, is an observation variable, is a system noise, A k is a state transition matrix, B k is an input matrix, C k is an observation matrix; Furthermore, the state transition matrix A k , the input matrix B k , the observation matrix C k is expressed as 。 5. The lithium battery SOC estimation method based on extended Kalman filter according to claim 4, characterized in that, The updating of the prediction part and the correction part of the extended Kalman filter algorithm by calculating the low-pass filtering results of the prediction part and the correction part of the extended Kalman filter algorithm and piecewise transforming the observation covariances of different orders of magnitude in the prediction part and the correction part of the extended Kalman filter algorithm specifically comprises the following steps: calculating low-pass filtering results of observation residuals in the prediction part and the correction part of the extended Kalman filter algorithm; determining a tolerance error time point between the SOC estimation value and the actual value of the lithium battery according to the low-pass filtering results; piecewise transforming stage transformation values of observation covariances in the prediction part and the correction part of the extended Kalman filter algorithm according to the tolerance error time point; updating the observation covariances in the extended Kalman filter algorithm according to the stage transformation values.

6. The lithium battery SOC estimation method based on extended Kalman filter according to claim 5, characterized in that, The prediction part is: The correction part is: wherein is the observation residual; The stage transformation values of the observation covariances comprise a first order of magnitude value and a second order of magnitude value, and the observation covariances in the extended Kalman filter algorithm are updated according to the stage transformation values by the following formula: wherein, is the observation covariance, K is the Kalman gain.

7. The lithium battery SOC estimation method based on extended Kalman filter according to claim 6, characterized in that, The linear system state and observation equations required by the reconstructed extended Kalman filter algorithm are: wherein is an initial time point, is an allowable error time point, is a first order of magnitude value, is a second order of magnitude value.

8. The lithium battery SOC estimation method based on extended Kalman filter according to claim 7, characterized in that, The estimation of the SOC of the lithium battery according to the linear system state and observation equations required by the reconstructed extended Kalman filter algorithm specifically comprises the following steps: wherein .

9. The lithium battery SOC estimation method based on extended Kalman filter according to claim 8, characterized in that, obtaining an iterative updating equation set of the lithium battery according to the linear system state and observation equations required by the reconstructed extended Kalman filter algorithm; estimating the SOC of the lithium battery according to the iterative updating equation set. The iterative updating equation set is:

10. The lithium battery SOC estimation method based on extended Kalman filter according to claim 9, characterized in that, ​ wherein, M is a diagonal matrix, I is an identity matrix, is a bandwidth of a Gaussian kernel function, j is an iteration number of the iterative update equation set, 1≤ j ≤ N , N is a positive integer, and when j = N an SOC estimation result of the lithium battery is output.

Images