Lithium ion battery SOC (State of Charge) estimation method and system based on improved volume Kalman filtering, computing equipment and medium

Abstract

The invention discloses a lithium ion battery SOC estimation method and system based on improved volume Kalman filtering, computing equipment and a medium, relates to the technical field of energy storage lithium ion battery management, and solves the problem of SOC estimation errors caused by model mismatching in the prior art. According to the method, a high-fidelity second-order fractional-order equivalent circuit model is established, meanwhile, a dual-time-scale volume Kalman filtering framework is designed, time-varying parameters of the battery can be dynamically tracked, and errors caused by mismatching of the models are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of energy storage lithium-ion battery management, and particularly relates to a method, a system, a computing device and a medium for estimating the state of charge (SOC) of a lithium-ion battery based on improved cubature Kalman filter. Background Art

[0002] The progress of transportation electrification and energy storage technology is a key strategy for addressing climate change and energy shortages. As an advanced energy storage technology, lithium-ion batteries are widely used in high-power scenarios such as electric vehicles and energy storage systems. The state of charge (SOC) of a battery is a core parameter for battery management and is crucial for energy allocation, operation planning, and fault diagnosis. Especially in complex high-power systems, real-time and accurate SOC estimation is particularly important.

[0003] SOC estimation methods are mainly divided into direct measurement methods, data-driven methods, and model-based methods. Direct measurement methods rely on the accurate measurement of parameters such as voltage and current, but are limited by sensor accuracy and cumulative errors. Although data-driven methods (such as deep learning) have made progress, they rely on a large amount of data and lack interpretability. Model-based methods (such as equivalent circuit models) describe battery behavior through a theoretical framework and, combined with technologies such as Kalman filter (KF), can use prior knowledge to improve the estimation accuracy and are widely used in battery management systems (BMS).

[0004] However, equivalent circuit models are mainly modeled based on ideal and static parameters and are difficult to reflect the dynamic changes of the actual circuit, resulting in model mismatch and calculation errors.

[0005] In view of this, there is a need for a method, a system, a computing device and a medium for estimating the SOC of a lithium-ion battery based on improved cubature Kalman filter. Summary of the Invention

[0006] Aiming at the problem of SOC estimation error caused by model mismatch in the prior art, the present invention provides a method, a system, a computing device and a medium for estimating the SOC of a lithium-ion battery based on improved cubature Kalman filter, which can reduce the error caused by model mismatch during SOC estimation. The specific technical solutions are as follows:

[0007] In a first aspect, an embodiment of the present application provides a method for estimating the state of charge (SOC) of a lithium-ion battery based on improved cubature Kalman filtering, characterized by including the following steps: Step 1: Obtain a second-order fractional-order RC equivalent circuit model of the lithium-ion battery, establish a discrete state-space expression of the equivalent circuit model, and obtain an initial value of the model parameters of the equivalent circuit model through an offline parameter identification method; Step 2: Construct a two-time-scale state equation with a noise interference term based on the equivalent circuit model, where the two-time-scale state equation is used to dynamically update the model parameters at two different time scales; Step 3: Iteratively improve the cubature Kalman filtering SOC estimation algorithm based on the two-time-scale state equation; Step 4: Collect the current and voltage values during the operation of the lithium-ion battery, and based on the current and voltage values, estimate the SOC of the lithium-ion battery through the improved cubature Kalman filtering algorithm after iteration.

[0008] In a second aspect, an embodiment of the present application provides a system for estimating the state of charge (SOC) of a lithium-ion battery based on improved cubature Kalman filtering, characterized by including:

[0009] A modeling module, configured to obtain a second-order fractional-order RC equivalent circuit model of the lithium-ion battery, establish a discrete state-space expression of the equivalent circuit model, and obtain an initial value of the model parameters of the equivalent circuit model through an offline parameter identification method;

[0010] A construction module, configured to construct a two-time-scale state equation with a noise interference term based on the equivalent circuit model, where the two-time-scale state equation is used to dynamically update the model parameters of the equivalent circuit model at two different time scales;

[0011] An iteration module, configured to iteratively improve the cubature Kalman filtering SOC estimation algorithm based on the two-time-scale state equation;

[0012] An estimation module, configured to collect the current and voltage values during the operation of the lithium-ion battery, and based on the current and voltage values, estimate the SOC of the lithium-ion battery through the improved cubature Kalman filtering algorithm after iteration.

[0013] In a third aspect, an embodiment of the present application provides a computing device, including: a memory for storing a program; a processor for loading the program to execute the method as described in the first aspect.

[0014] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium storing a program, and when the program is executed by a processor, the method as described in the first aspect is implemented.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention establishes a high-fidelity fractional-order second-order equivalent circuit model, and at the same time designs a dual-time-scale cubature Kalman filter architecture, which can dynamically track the time-varying parameters of the battery and reduce the error caused by model mismatch. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0017] Figure 1 It is a schematic flow chart of a method for estimating the state of charge (SOC) of a lithium-ion battery based on an improved cubature Kalman filter provided by an embodiment of the present application;

[0018] Figure 2 It is a circuit diagram of an equivalent circuit model provided by an embodiment of the present application;

[0019] Figure 3 It is an iterative flow chart of an improved dual-cubature Kalman filter SOC estimation algorithm provided by an embodiment of the present application;

[0020] Figure 4 It is a flow chart of an intelligent adaptive algorithm based on the Wilcoxon rank sum test provided by an embodiment of the present application;

[0021] Figure 5 It is a schematic structural diagram of a system for estimating the state of charge (SOC) of a lithium-ion battery based on an improved cubature Kalman filter provided by an embodiment of the present application;

[0022] Figure 6 It is a schematic structural diagram of a computing device provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0024] It should be understood that when used in this specification and the appended claims, the terms "comprises" and "comprising" indicate the presence of the described features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or their groups.

[0025] It should also be understood that the terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include the plural forms.

[0026] It should be further understood that the term "and / or" used in the specification of the present invention and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0027] To solve the problem of SOC estimation error caused by model mismatch in the prior art, the present invention provides a lithium-ion battery SOC estimation method, system, computing device and medium based on improved cubature Kalman filter, which can reduce the error caused by model mismatch during SOC estimation.

[0028] Please refer to Figure 1 , Figure 1 which is a lithium-ion battery SOC estimation method based on improved cubature Kalman filter provided by an embodiment of the present application. This method can be applied to a computing device and specifically includes the following steps:

[0029] Step 1: The computing device obtains a second-order fractional-order RC equivalent circuit model of the lithium-ion battery, establishes a discrete state-space expression of the equivalent circuit model, and obtains an initial value of the model parameters of the equivalent circuit model through an offline parameter identification method.

[0030] Among them, the computing device can obtain the parameters and experimental data of these lithium-ion batteries based on common models of lithium-ion batteries and establish a corresponding second-order fractional-order RC equivalent circuit model; then establish a discrete state-space expression of this equivalent circuit model, obtain an initial value of the model parameters of the equivalent circuit model through an offline parameter identification method and assign this initial value to this equivalent circuit model to obtain an equivalent circuit model that can describe the electrical state of this lithium-ion battery.

[0031] Reference can be made to Figure 2 , Figure 2 which is a circuit diagram of an equivalent circuit model provided by an embodiment of the present application. As shown in Figure 2 , this equivalent circuit model may include a power supply, an equivalent series resistance R0, node 1 and node 2; the open-circuit voltage of the power supply can be expressed as U OCV , the load voltage can be expressed as U t , node 1 can be expressed as a parallel circuit unit of a constant phase element CPE1 and a resistor R1, and node 2 can be expressed as a parallel circuit unit of a constant phase element CPE2 and a resistor R1.

[0032] Preferably, the following discrete state - space expression can be established based on this equivalent circuit model:

[0033]

[0034] U k =U oc (SOC k ) - U 1,k -U 2,k -R0I k +v k ,

[0035] where k represents the short - time scale, with the unit of seconds (s), the state variable x at time k k =[U 1,k ,U 2,k ,SOC k T , x k-1 is the state variable at time k - 1, and x kk-1 is the state variable at time k predicted based on the information at time k - 1; U 1,k is the voltage at node 1 in the equivalent circuit model at time k, U 2,k is the voltage at node 2 at time k, SOC k is the SOC of the lithium - ion battery at time k, I k is the current of the lithium - ion battery at time k; the coefficient matrices

[0036] A = diag{-T s α / R1C1, - T s β / R2C2, 1}, B = [T s α / C1, T s β / C2, T s / Q n T , where T s is the sampling period, R1 is the impedance of node 1, R2 is the impedance of node 2, C1 is the capacitance of node 1, C2 is the capacitance of node 2, Q n is the covariance matrix related to noise, α and β are fractional - order capacitance calculation coefficients, j is the number of steps of time delay, L is the preset maximum number of time - delay steps, x k-j is the state variable at time k - j; w k and v k represent process noise and measurement noise respectively, and U oc represents the open - circuit voltage, U​​k Let \(V\) represent the terminal voltage and \(R_0\) represent the equivalent series resistance.

[0037] Step 2: The computing device constructs a two-time-scale state equation with a noise interference term based on the equivalent circuit model.

[0038] Among them, the two-time-scale state equation is used to dynamically update the model parameters at two different time scales. The noise interference term can simulate the influence of environmental factors in the real environment, reflect the measurement error, and improve the robustness of the algorithm.

[0039] Preferably, the two-time-scale state equation with a noise interference term is as follows:

[0040]

[0041] where \(\theta\) l represents the parameter variable in time period \(l\), \(\theta\) l-1 represents the parameter variable in time period \(l - 1\), \(f(*)\) and \(h(*)\) represent the state transition function and the observation function respectively, \(l\) represents the long time scale, with the unit of min; \(w\) x,k is the influence term of the process noise on the state variable at time \(k\), \(w\) θ,l is the influence term of the process noise on the parameter variable in time period \(l\), \(v\) k represents the measurement noise.

[0042] By using the two-time-scale state equation in the embodiments of the present application, the dynamic changes of the equivalent circuit model at two time scales can be analyzed, making the equivalent circuit model more adaptable to the actual electrical state of the lithium-ion battery and the calculation results more accurate during subsequent SOC estimation.

[0043] Step 3: The computing device iteratively improves the dual-volume Kalman filter SOC estimation algorithm based on the two-time-scale state equation.

[0044] Among them, the iteration of the improved dual-volume Kalman filter SOC estimation algorithm can be divided into short-time-scale iteration and long-time-scale iteration. For details, refer to Figure 3 , Figure 3 which is the iteration flowchart of the improved dual-volume Kalman filter SOC estimation algorithm provided by the embodiments of the present application. As Figure 3 shown, the iteration process specifically includes:

[0045] Step 3.1: The computing device initializes the state and the parameter and the covariance matrices \(P\) x,0 and \(P\) θ,0 , as well as the noise statistical characteristics \(Q\) x,0 , \(Q\) θ,0 , \(R\) x,0 and \(R\)θ,0 ; wherein, is the state vector at the initial moment, is the parameter vector at the initialization moment, P x,0 is the state at the initial moment of the covariance matrix, P θ,0 is the parameter at the initial moment of the covariance matrix, Q x,0 is the covariance matrix of the system process noise at the initial moment, Q θ,0 is the covariance matrix of the process noise related to the parameter θ at the initial moment, R x,0 is the covariance matrix of the measurement noise at the initial moment, R θ,0 is the covariance matrix of the measurement noise when measuring the parameter θ at the initial moment.

[0046] Specifically, the system process noise refers to the uncertain factors existing within the system that affect the evolution of the state variables, such as model errors, external disturbances, etc.; the measurement noise refers to the difference between the measured value and the true value due to factors such as the precision limitation of the measurement device and environmental interference when measuring the state variables; the parameter θ usually refers to some unknown parameters in the system model. For example, in some dynamic systems, it may include system gains, time constants, etc.

[0047] The initialization formula is as follows:

[0048]

[0049] wherein, E(*) represents the mathematical expectation at the initial moment, and the mathematical expectations of different variables are obtained based on the corresponding preset values; x0 is the prior data of the preset state vector, and θ0 is the prior data of the preset parameter vector.

[0050] When initializing the Kalman filter algorithm, E(*) represents the mathematical expectation of the state variables at the initial moment, that is, the mean estimation of the state variables. It is a way to quantify the prior information about the initial state of the system. The prior information may come from previous experimental data, theoretical analysis, or other relevant observational results, etc. Through E(*), we convert this information into a specific numerical value or vector (when the state variables are multi-dimensional) for use as the initial conditions in the Kalman filter algorithm.

[0051] Step 3.2: Calculate the time update of the state of the device on a short time scale, k = 1, 2,..., N; where N is a positive integer;

[0052] (1) Time update:

[0053] k = k + 1,

[0054] (2) Calculate the first cubature point:

[0055] S x,k-1 = SVD(P x,k-1 )

[0056]

[0057] where S x,k-1 represents the singular value of the covariance matrix P x,k-1 related to the state vector x at time k - 1, n represents the vector dimension of the state vector x, represents the state of the i-th cubature point at time k - 1, represents the estimated state vector at time k - 1, ζ n,i is a column vector of dimension n, whose element in the i-th row is and the remaining elements are 0.

[0058] Among them, in the Kalman filter algorithm and its extended algorithms, cubature points are a set of special points used to approximate probability distributions and perform integral calculations. Specifically, cubature points are a set of points selected according to certain mathematical rules, and these points have specific distributions and weights in the state space, and are used to approximate the integral operation of the probability density function in the Kalman filter framework.

[0059] Among them, in the Kalman filter algorithm, two formulas are used to calculate cubature points and start from 1 and n + 1 respectively. Usually, this is to more comprehensively and accurately cover the state space to improve the accuracy of probability distribution approximation and integral calculation. Specifically, the cubature points calculated by the formula starting from 1 are mainly used to capture information in one direction of the state space; the cubature points determined by the formula starting from n + 1 are a supplement to the cubature points starting from 1. In many cases, they will correspond to directions in the state space that are symmetric or complementary to the previously obtained points.

[0060] (3) The state prediction calculation formula for the first cubature point is:

[0061] Or,

[0062]

[0063] where represents the estimated parameter vector in time period l, represents the state of the i-th first cubature point at time k estimated based on the information at time k - 1, I k-1 is the current of the lithium-ion battery at time k - 1; represents the estimated state vector at time k - j.

[0064] It should be noted that the volume point states are a series of deterministic points selected in the state space according to the volume rules. Therefore, the vectors corresponding to the volume point states are determined by the content of the volume point calculation formula. For example, the states of the first and second volume points are the state vectors of the equivalent circuit model, while the states of the third and fourth volume points are the parameter vectors of the equivalent circuit model.

[0065] It should be noted that in this article, for the vector X, represents the estimated value of X at time k estimated based on the information at time k - 1; X kk-1 represents the predicted value of X at time k predicted based on the information at time k - 1. The difference between the two is that the estimated value is the result after filtering calculation or correction, and the predicted value is the prediction result without going through the observation value fusion stage.

[0066] (4) Estimation of the target state of the first volume point:

[0067]

[0068] Among them, represents the estimated target state vector at time k estimated based on the information at time k - 1.

[0069] (5) Prediction covariance:

[0070]

[0071] Among them, P x,kk-1 represents the covariance matrix of the state vector x at time k predicted based on the information at time k - 1,

[0072] Q x,k-1 represents the covariance matrix of the process noise related to the state vector x at time k - 1.

[0073] Step 3.3: Calculate the measurement update of the device state on a short time scale, k = 1, 2,..., N;

[0074] (1) Calculate the second volume point:

[0075] S x,kk-1 = SVD(P x,kk-1 ),

[0076]

[0077] Among them, S x,kk-1 represents the singular value of the covariance matrix P x,kk-1 in Step 3.2(5), represents the state of the i-th second volume point at time k predicted based on the information at time k - 1, Denotes the state variable at time k estimated based on the information at time k-1.

[0078] (2) Measurement estimation of the second volume point:

[0079]

[0080] Among them, Denotes the first measurement vector of the i-th second volume point at time k estimated based on the information at time k-1.

[0081] (3) Measurement estimation of the first target value:

[0082]

[0083] Among them, Denotes the first target value measurement vector at time k estimated based on the information at time k-1.

[0084] (4) Calculate the state measurement covariance:

[0085]

[0086] Among them, P x,zz Denotes the state covariance matrix regarding the first measurement vector and the first target value measurement vector, R x,k-1 Denotes the covariance matrix of the measurement noise related to the state vector x at time k-1.

[0087] (5) Calculate the first cross-covariance:

[0088]

[0089] Among them, P x,xz Denotes the cross-covariance between the state vector and the measurement vector.

[0090] (6) Calculate the first Kalman gain:

[0091]

[0092] Among them, K x,k Denotes the first Kalman gain related to the state vector x at time k.

[0093] (7) Calculate the innovation:

[0094]

[0095] Among them, z k Denotes the actual measurement value at time k, e kDenote the innovation at time k. Innovation is a quantity used in Kalman filtering to measure the difference between the measured value and the predicted value. It reflects the new information contained in the latest measurement data relative to the previous estimation result. Intuitively, it is to determine how much new information can be used to update the state estimation by comparing the actually measured result with the result that should be obtained according to the model prediction.

[0096] (8) State estimation:

[0097]

[0098] Among them, Denote the state vector at estimated time k;

[0099] (9) Covariance estimation:

[0100]

[0101] Among them, P x,k Denote the covariance matrix of the state vector x at time k;

[0102] (10) Determine the sliding window length L at time k by rank-sum test k :

[0103] p = Wilcoxon({e k-i+1 | k = 1, 2,..., L0}, {e k-i+1 | k = L0 + 1, L0 + 2,..., L k-1}),

[0104]

[0105] Among them, the p-value is a statistical index in the rank-sum test algorithm, L0 is the minimum window length, L k-1 Denote the sliding window length at time k - 1, L max is the maximum window length; e k-i+1 Denote the innovation at time k - i + 1; Wilcoxon() is the rank-sum test algorithm.

[0106] (11) Adaptive noise covariance update:

[0107]

[0108] Among them, H x,k Denote the measurement matrix of state x at time k; R x,k Denote the measurement noise covariance matrix of state x at time k; Q x,k Denote the process noise covariance matrix of state x at time k.

[0109] Step 3.4: After completing the state and parameter update on a short time scale, the computing device can perform a time scale judgment. If it is divisible by 60, then proceed to Step 3.5; if it is not divisible by 60, then return to execute Step 3.2;

[0110] Step 3.5: Update the parameters on a long time scale, k = 60l, l = 1, 2,...;

[0111] (1) Calculate the third volume point:

[0112] S θ,l-1 = SVD(P θ,l-1 ),

[0113]

[0114] where S θ,l-1 represents the singular value of the covariance matrix P θ,l-1 related to the parameter θ in the (l - 1)th period, m represents the vector dimension of the parameter θ, represents the state of the i-th third volume point in the (l - 1)th period, represents the estimated parameter vector in the (l - 1)th period, ζ m,i is a column vector with dimension m, and its element in the i-th row is and the remaining elements are 0;

[0115] (2) State estimation of the third volume point:

[0116]

[0117] where, represents the state of the i-th third volume point in the l-th period estimated based on the information in the (l - 1)th period.

[0118] (3) Target state estimation of the third volume point:

[0119]

[0120] where, represents the estimated target parameter vector in the l-th period estimated based on the information in the (l - 1)th period.

[0121] (4) Prediction covariance:

[0122]

[0123] where P θ,ll-1 represents the covariance matrix of the parameter vector θ in the l-th period predicted based on the information in the (l - 1)th period, and Q θ,l-1 represents the covariance matrix of the process noise related to the parameter vector θ in the (l - 1)th period.

[0124] Step 3.6: Perform measurement update of parameters on a long time scale:

[0125] (1) Calculate the fourth volume point:

[0126] S θ,ll-1 = SVD(P θ,ll-1 );

[0127]

[0128] where S θ,ll-1 represents the singular value of the covariance matrix P θ,ll-1 in Step 3.5(4), represents the state of the i-th fourth volume point at time l predicted based on the information at time l-1, represents the parameter variable at time l estimated based on the information at time l-1.

[0129] (2) Measurement estimation of the fourth volume point:

[0130]

[0131] where, represents the second measurement vector of the i-th fourth volume point at time l estimated based on the parameter information at time l-1.

[0132] (3) Measurement estimation of the second target value:

[0133]

[0134] where, represents the second target value measurement vector at time l predicted based on the information at time l-1, and the second target value measurement vector is related to the parameter vector.

[0135] (4) Calculate the parameter measurement covariance:

[0136]

[0137] where P θ,zz represents the parameter covariance matrix regarding the second measurement vector of the volume point and the second target value measurement vector, and R θ,l-1 represents the covariance matrix of the measurement noise related to the parameter vector θ at time l-1;

[0138] (5) Calculate the second cross-covariance:

[0139]

[0140] where P θ,xz represents the cross-covariance regarding the parameter vector and the measurement vector;

[0141] (6) Calculate the second Kalman gain:

[0142]

[0143] where K θ,l represents the second Kalman gain related to the parameter vector θ at time period l;

[0144] (7) Parameter estimation:

[0145]

[0146] where represents the estimated parameter vector at time period l, and z θ,l is the actual measurement value at time period l;

[0147] (8) Covariance estimation:

[0148]

[0149] where P θ,l represents the covariance matrix of the parameter vector θ at time period l;

[0150] Step 3.7: Return to execute Step 3.2.

[0151] Preferably, the process of the rank sum test for determining the sliding window length L k is as Figure 4 shown, and specifically includes:

[0152] 1) After initializing the noise covariance in Step 3.1 and performing measurement update on the state at a short time scale in Step 3.3, calculate the innovation based on the measurement value z k ; then divide the historical innovation into two independent sample sets, namely the first sample set Sample1 and the second sample set Sample2:

[0153]

[0154] 2) Combine the first sample set and the second sample set into a data set, sort all the innovations in the data set by size, and assign a rank (ranking) to each innovation. If there are innovations of the same size, assign the average rank to this group of identical innovations;

[0155] 3) Sum the ranks of the innovations in the first sample set and the ranks of the innovations in the second sample set respectively to obtain the rank sums of the two groups of data, denoted as W1 and W2;

[0156] 4) Calculate the U value:

[0157] U = min{n1 * n2 + n1 * (n1 + 1) / 2 - W1, n1 * n2 + n2 * (n2 + 1) / 2 - W1};

[0158] Where n1 is the number of samples in the first sample set, and n2 is the number of samples in the second sample set;

[0159] 5) Significance test:

[0160] Z = (U - n1 * n2 / 2) / sqrt(n1 * n2 * (n1 + n2 + 1)) / 12;

[0161] Where Z is the statistic of the standard normal distribution.

[0162] 6) Determine the p - value and the sliding window length L based on the Z - value k .

[0163] Where, when calculating the noise covariance, the sliding window can update the data and capture the time - varying characteristics well. Specifically, the statistical characteristics of the noise change over time. The sliding window can move on the data sequence at a certain time step, so as to track the change of the noise characteristics and capture the dynamic change of the noise covariance over time in a timely manner. Using a sliding window with an appropriate length to calculate the noise covariance can better adapt to this dynamic change.

[0164] The present invention limits the range of the sliding window length L by setting a minimum window length L0 and a maximum window length L max to limit the sliding window length L k The null hypothesis of the above - mentioned test is that two groups of independent samples before and after the k - i + 1 moment follow the same distribution. When the p - value is obtained, it is judged for significance by comparing with 0.05. If the p - value is greater than 0.05, it means that L k accepts the null hypothesis, that is, there is no change in the noise statistical characteristics at the k - i + 1 moment, then the sliding window length L k can be increased by 1 until the maximum value L max is reached. If the p - value is less than or equal to 0.05, it means that L k rejects the null hypothesis, that is, a change in the noise statistical characteristics before and after is detected at the k - i + 1 moment, then the sliding window length is reset to L0.

[0165] Through the adaptive method based on the Wilcoxon rank - sum test, the change of non - Gaussian noise can be accurately identified, the noise covariance matrix can be adaptively updated, and the performance of the algorithm under complex conditions can be improved.

[0166] Step 4: Collect the current and voltage values during the operation of the lithium - ion battery, and based on the current and voltage values, estimate the SOC of the lithium - ion battery by using the improved double - volume Kalman filtering algorithm after iteration.

[0167] The present invention establishes a high-fidelity fractional-order second-order equivalent circuit model, and at the same time designs a dual-time-scale cubature Kalman filter architecture, which can dynamically track the time-varying parameters of the battery and reduce the error caused by model mismatch. In addition, the present invention also innovatively proposes an adaptive method based on the Wilcoxon rank-sum test, which can accurately identify the changes in non-Gaussian noise, adaptively update the noise covariance matrix, and improve the performance of the algorithm under complex conditions.

[0168] The method provided by the embodiments of the present application has been described above. The system provided by the embodiments of the present application will be described below.

[0169] Please refer to Figure 5 , Figure 5 , which is a schematic structural diagram of a lithium-ion battery SOC estimation system based on an improved cubature Kalman filter provided by the embodiments of the present application. As Figure 5 shown, the system 500 includes:

[0170] A modeling module 501, configured to obtain a second-order fractional-order RC equivalent circuit model of a lithium-ion battery, establish a discrete state-space expression of the equivalent circuit model, and obtain an initial value of the model parameters of the equivalent circuit model through an offline parameter identification method;

[0171] A construction module 502, configured to construct a dual-time-scale state equation with a noise interference term based on the equivalent circuit model, where the dual-time-scale state equation is used to dynamically update the model parameters at two different time scales;

[0172] An iteration module 503, configured to iteratively improve the dual-cubature Kalman filter SOC estimation algorithm based on the dual-time-scale state equation;

[0173] An estimation module 504, configured to collect current and voltage values during the operation of the lithium-ion battery, and based on the current and voltage values, perform SOC estimation on the lithium-ion battery through the improved dual-cubature Kalman filter algorithm after iteration.

[0174] The system provided by the embodiments of the present application can be understood by referring to the corresponding content in the foregoing method embodiment part, and will not be repeated here.

[0175] As Figure 6 shown, Figure 6A possible schematic diagram of the logical structure of the computing device provided by the embodiments of the present application. The computing device 600 includes: a processor 601, a communication interface 602, a memory 603, and a bus 604. The processor 601, the communication interface 602, and the memory 603 are interconnected via the bus 604. In the embodiments of the present application, the processor 601 is used to control and manage the operations of the computing device 600. For example, the processor 601 is used to execute Figures 1 to 4 the steps in any of the embodiments and / or other processes for the technologies described herein. The communication interface 602 is used to support the computing device 600 to communicate. The memory 603 is used to store the program code and data of the computing device 600.

[0176] Among them, the processor 601 may be a central processing unit, a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a field-programmable gate array, or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute various exemplary logical blocks, modules, and circuits described in connection with the disclosure of the present application. The processor may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a digital signal processor and a microprocessor, and so on. The bus 604 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For the sake of simplicity of representation, Figure 6 only a thick line is shown in the figure, but it does not mean that there is only one bus or one type of bus.

[0177] In another embodiment of the present application, a computer-readable storage medium is further provided. The computer-readable storage medium includes instructions that, when run on a computer, cause the computer to execute the above Figures 1 to 4 method described in any of the embodiments.

[0178] Those of ordinary skill in the art can realize that the units of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the components of each example have been generally described according to their functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0179] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0180] In the several embodiments provided in the embodiments of the present application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections between each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0181] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0182] In addition, in each embodiment of the present invention, the functional units can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

[0183] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The foregoing storage medium includes: USB flash drives, read-only memories (ROMs), random access memories (RAMs), mobile hard disks, magnetic disks, or optical disks, and other media that can store program codes.

[0184] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the various embodiments of the present invention, and they should all be covered within the scope of the claims and the specification of the present invention.

Claims

1. A lithium-ion battery SOC estimation method based on improved volumetric Kalman filtering, characterized in that: The steps include: Step 1: obtaining a second-order fractional-order RC equivalent circuit model of a lithium-ion battery, establishing a discrete state space expression of the equivalent circuit model, and obtaining initial values ​​of model parameters of the equivalent circuit model by an offline parameter identification method; Step 2: constructing a dual-time-scale state equation containing a noise interference term based on the equivalent circuit model, wherein the dual-time-scale state equation is used to dynamically update model parameters of the equivalent circuit model at two different time scales; Step 3: Iteratively improving the dual-volume Kalman filter SOC estimation algorithm based on the dual-time-scale state equation; Step 4: collecting current and voltage values ​​of the lithium-ion battery during operation, and estimating the SOC of the lithium-ion battery by using the iterative improved dual-volume Kalman filter algorithm based on the current and voltage values.

2. The method according to claim 1, characterized in that The model parameters include the state variable x and parameter variable θ of the equivalent circuit model; the discrete state space expression of the equivalent circuit model is as follows: The k =U oc (SOC k )-U 1,k -U 2,k -R0I k +v k , Among them, k represents the short time scale, the unit is seconds, and the state variable x at time k is k =[U 1,k ,U 2,k ,SOC k ] T , x k-1 is the state variable at time k-1, x k|k-1 is the state variable at time k predicted based on the information at time k-1; U 1,k is the voltage of node 1 at time k in the equivalent circuit model, U 2,k is the voltage of node 2 at time k, SOC k is the SOC of the lithium-ion battery at time k, I k is the current of the lithium-ion battery at time k; the coefficient matrix A = diag{-T s α / R1C1,-T s β / R2C2,1}, B=[T s α / C1,T s β / C2,T s / Q n ] T , Where T s is the sampling period, R1 is the impedance of node 1, R2 is the impedance of node 2, C1 is the capacitance of node 1, C1 is the capacitance of node 2, Q n is the covariance matrix associated with the noise, α and β are the coefficients for fractional capacitance calculation, j is the number of time delay steps, L is the preset maximum time delay step, x k-j is the state variable at time kj; w k and v k represent process noise and measurement noise respectively, U oc Indicates the open circuit voltage, U k represents the terminal voltage, and R0 represents the equivalent series resistance.

3. The method according to claim 2, characterized in that: The dual-time-scale state equation containing noise interference terms is as follows: Among them, θ l represents the parameter variable in time period l, θ l-1 represents the parameter variable in the time period l-1, f(*) and h(*) represent the state transfer function and observation function respectively, l represents the long time scale, the unit is min; w x,k is the influence of process noise at time k on the state variable, w θ,l is the influence of process noise on the parameter variable in period l, v k represents the measurement noise.

4. The method according to claim 3, characterized in that: The iteratively improved dual-volume Kalman filter SOC estimation algorithm based on the dual-time-scale state equation includes: Step 3.1: Initialize state and parameters The covariance matrix P of the two x,0 and P θ,0 , and the noise statistics Q x,0 , Q θ,0 , R x,0 and R θ,0 ;in, is the state vector at the initial moment, is the parameter vector at the initialization time, P x,0 The initial state The covariance matrix, P θ,0 is the initial time parameter The covariance matrix, Q x,0 is the covariance matrix of the system process noise at the initial moment, Q θ,0 is the covariance matrix of the process noise associated with the parameter θ at the initial moment, R x,0 is the covariance matrix of the measurement noise at the initial moment, R θ,0 is the measurement noise covariance matrix when the parameter θ is measured at the initial moment; the initialization formula is as follows: Wherein, E(*) represents the mathematical expectation at the initial moment, and the mathematical expectations of different variables are obtained based on the corresponding preset values; x0 is the prior data of the preset state vector, and θ0 is the prior data of the preset parameter vector; Step 3.2: Perform time update of the state on a short time scale, k = 1, 2, ..., N; where N is a positive integer; (1) Time update: k=k+1, (2) Calculate the first volume point: S x,k-1 =SVD(P x,k-1 ), Among them, S x,k-1 Represents the covariance matrix P associated with the state vector x at time k-1 x,k-1 The singular values ​​of , n represents the vector dimension of the state vector x, represents the state of the i-th volume point at time k-1, represents the estimated k-1 time state vector, ζ n,i is a column vector of dimension n, whose elements in the i-th row are The remaining elements are 0; (3) The state prediction calculation formula of the first volume point is: or, in, represents the estimated parameter vector in time period l, represents the state of the i-th first volume point at time k estimated based on the information at time k-1, I k-1 is the current of the lithium-ion battery at time k-1; represents the estimated state vector at time kj; (4) Target state estimation at the first volume point: in, represents the estimated target state vector at time k based on the information at time k-1; (5) Prediction covariance: Among them, P x,kk-1 represents the covariance matrix of the state vector x at time k predicted based on the information at time k-1, Q x,k-1 represents the covariance matrix of the process noise associated with the state vector x at time k-1; Step 3.3: Update the state measurement on a short time scale, k = 1, 2, ..., N; (1) Calculate the second volume point: S x,kk-1 =SVD(P x,kk-1 ), Among them, S x,kk-1 represents the covariance matrix P in step 3.2(5) x,kk-1 The singular values ​​of represents the state of the i-th second volume point at time k predicted based on the information at time k-1, represents the state variable at time k estimated based on the information at time k-1; (2) Measurement estimation of the second volume point: in, represents a first measurement vector of the i-th second volume point at time k estimated based on information at time k-1; (3) First target value measurement estimation: in, represents the first target value measurement vector at time k estimated based on the information at time k-1; (4) Calculate the state measurement covariance: Among them, P x,zz represents the state covariance matrix about the first measurement vector and the first target value measurement vector, R x,k-1 represents the covariance matrix of the measurement noise associated with the state vector x at time k-1; (5) Calculate the first cross-covariance: Among them, P x,xz represents the cross-covariance between the state vector and the measurement vector; (6) Calculate the first Kalman gain: Among them, K x,k represents the first Kalman gain associated with the state vector x at time k; (7) Calculate new interest: Among them, z k represents the actual measured value at time k, e k Indicates the new information at time k; (8) State estimation: in, represents the estimated state vector at time k; (9) Covariance estimation: Among them, P x,k Represents the covariance matrix of the state vector x at time k; (10) Rank sum test determines the sliding window length L at time k k : p=Wilcoxon({e k-i+1 |k=1,2,...,L0},{e k-i+1 |k=L0+1,L0+2,...,L k-1 }), Among them, the p value is the statistical indicator in the rank sum test algorithm, L0 is the minimum window length, and L k-1 Indicates the length of the sliding window at time k-1, L max is the maximum window length; e k-i+1 represents the new information at time k-i+1; Wilcoxon() is the rank sum test algorithm; (11) Adaptive noise covariance update: Among them, H x,k R represents the measurement matrix of state x at time k; x,k represents the measurement noise covariance matrix of state x at time k; Q x,k represents the process noise covariance matrix about state x at time k; Step 3.4: Determine the time scale. If k is divisible by 60, then l=l+1 and execute step 3.

5. If k is not divisible by 60, then return to execute step 3.

2. Step 3.5: Update the parameters on a long time scale, k = 60l, l = 1, 2, ...; (1) Calculate the third volume point: S θ,l-1 =SVD(P θ,l-1 ), Among them, S θ,l-1 Represents the covariance matrix P related to the parameter θ in the l-1 period θ,l-1 The singular values ​​of , m represents the vector dimension of the parameter θ, represents the state of the i-th third volume point in the l-1 period, represents the estimated parameter vector in the l-1 period, ζ m,i is a column vector of dimension m, whose elements in the i-th row are The remaining elements are 0; (2) State estimation of the third volume point: in, represents the state of the i-th third volume point in period l estimated based on the information in period l-1; (3) Target state estimation at the third volume point: in, represents the estimated target parameter vector for period l estimated based on the information of period l-1; (4) Prediction covariance: Among them, P θ,ll-1 represents the covariance matrix of the parameter vector θ in period l predicted based on the information in period l-1, Q θ,l-1 represents the covariance matrix of the process noise associated with the parameter vector θ in the l-1 period; Step 3.6: Update the parameters over a long time scale: (1) Calculate the fourth volume point: S θ,ll-1 =SVD(P θ,ll-1 ); Among them, S θ,ll-1 represents the covariance matrix P in step 3.5(4) θ,ll-1 The singular values ​​of represents the state of the fourth volume point in the i-th period of time predicted based on the information of the l-1 period, represents the l-period parameter variable estimated based on the information of the l-1 period; (2) Measurement estimation of the fourth volume point: in, A second measurement vector representing the i-th fourth volume point of the l-th period estimated based on the parameter information of the l-1 period; (3) Second target value measurement estimation: in, represents a second target value measurement vector for period l predicted based on information of period l-1, wherein the second target value measurement vector is related to the parameter vector; (4) Calculate the parameter measurement covariance: Among them, P θ,zz represents the parameter covariance matrix of the second measurement vector of the volume point and the second target value measurement vector, R θ,l-1 represents the covariance matrix of the measurement noise associated with the parameter vector θ in the l-1 period; (5) Calculate the second cross-covariance: Among them, P θ,xz represents the cross-covariance between the parameter vector and the measurement vector; (6) Calculate the second Kalman gain: Among them, K θ,l represents the second Kalman gain associated with the parameter vector θ during period l; (7) Parameter estimation: in, represents the estimated parameter vector for period l, z θ,l is the actual measured value during period l; (8) Covariance estimation: Among them, P θ,l represents the covariance matrix of the parameter vector θ for period l; Step 3.7: Return to step 3.

2.

5. The method according to claim 4, characterized in that: The rank sum test determines the sliding window length L at time k k , specifically including: 1) Divide the historical information into two independent sample sets, namely the first sample set Sample1 and the second sample set Sample2: 2) Combining the first sample set and the second sample set into a data set, and sorting all the new information in the data set by size, and assigning a rank to each of the new information; if there are new information of the same size, assigning an average rank to the same group of new information; 3) summing the ranks of the novelty of the first sample set and the novelty of the second sample set respectively to obtain the sum of the ranks of the two sets of data, denoted as W1 and W2; 4) Calculate U value: U=min{n1*n2+n1(n1+1) / 2-W1,n1*n2+n2(n2+1) / 2-W1}; Wherein, n1 is the number of samples in the first sample set, and n2 is the number of samples in the second sample set; 5) Significance test: Z=(U-n1*n2 / 2) / sqrt(n1*n2*(n1+n2+1)) / 12; Among them, Z is the statistic of the standard normal distribution; 6) Determine the p-value based on the Z-value and the normal distribution table, and determine the sliding window length L based on the p-value k .

6. A lithium-ion battery SOC estimation system based on improved volumetric Kalman filtering, characterized in that: include: A modeling module, used to obtain a second-order fractional-order RC equivalent circuit model of a lithium-ion battery, establish a discrete state space expression of the equivalent circuit model, and obtain initial values ​​of model parameters of the equivalent circuit model through an offline parameter identification method; A construction module, used for constructing a dual-time-scale state equation containing a noise interference term based on the equivalent circuit model, wherein the dual-time-scale state equation is used for dynamically updating model parameters of the equivalent circuit model at two different time scales; An iteration module, used for iteratively improving a dual-volume Kalman filter SOC estimation algorithm based on the dual-time-scale state equation; The estimation module is used to collect the current and voltage values ​​of the lithium-ion battery during operation, and estimate the SOC of the lithium-ion battery based on the current and voltage values ​​by using the iterative improved dual-volume Kalman filter algorithm.

7. A computing device, characterized in that include: Memory, used to store programs; A processor, configured to load the program to execute the method according to any one of claims 1 to 5.

8. A computer-readable storage medium storing a program, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images