A method and device for estimating state of charge of a lithium battery

Abstract

The present invention discloses a method and device for estimating the state of charge of a lithium battery. The method first uses a recursive least square method with a forgetting factor of multiple new information to identify battery parameters, expands the scalar new information into a new information vector, expands the gain vector into a gain matrix, and fully utilizes the parameter information contained in the input and output of the system. The system parameters are identified by an iterative calculation method, which has a higher data utilization rate, a fast convergence speed, and a high accuracy of parameter identification. Then, a weighted multi-new information adaptive extended Kalman filter method is used to estimate the battery SOC, and noise parameters are estimated on the basis of a traditional extended Kalman filter algorithm. Then, the new information vectors at multiple moments are used to replace the correction of the estimated value using the new information at the current moment in the original algorithm, and the new data has a higher weight, which can effectively improve the accuracy.

Description

Technical Field

[0001] The present invention relates to the field of energy storage system operation and control technology, and in particular to a lithium battery state of charge estimation method and device. Background Art

[0002] In recent years, with the rapid development of new energy electric vehicles, lithium-ion batteries have become the main power batteries for new energy electric vehicles due to their advantages such as high power, high energy density, and recyclability. It is particularly important to give full play to the performance of electric vehicle power batteries, which requires the battery management system (BMS) to accurately estimate the battery's state of charge. Accurate estimation of the battery's state of charge (SOC) can ensure the battery's optimal working state, thereby helping to extend battery life, reduce usage costs, and improve the reliability of new energy electric vehicles.

[0003] The battery state of charge cannot be measured directly, but can only be obtained by collecting battery-related parameters, building a model, and then using an estimation algorithm. The recursive least squares (RLS) algorithm is an easy-to-implement algorithm, but data saturation will occur as the data increases; the forgetting factor least squares algorithm (FFRLS) introduces a forgetting factor on its basis to effectively overcome the data saturation phenomenon, but it requires the system input to be ergodic and stationary, and the selection of the initial value has a greater impact on the algorithm's identification results.

[0004] At present, the SOC estimation methods proposed at home and abroad are mainly divided into two categories. One is the SOC estimation method based on the physical characteristics of the battery, such as the ampere-hour integration method and the open circuit voltage method; the other is the intelligent algorithm, such as the neural network method and the Kalman filter method. The ampere-hour integration method has too high requirements on the accuracy of the initial SOC value of the battery and the measured current, and it is easy to produce cumulative errors; the open circuit voltage method estimates the SOC by measuring the open circuit voltage of the battery, but it needs to be stationary for a long time and cannot achieve online detection; the neural network method needs to rely on a large number of samples for data training to obtain higher accuracy. The Kalman filter method is a widely used method in linear systems. The extended Kalman filter (EKF) extends it to nonlinear systems. Although the EKF algorithm takes into account the noise in the actual process, it assumes that the noise is constant, which is obviously inconsistent with the actual situation and will inevitably bring errors to the estimation. In addition, the traditional EKF algorithm only uses the information at the current moment to correct the estimated value, which will bring certain errors to the estimation. Summary of the invention

[0005] The purpose of the present invention is to provide a method and device for estimating the state of charge of a lithium battery, and the technical solution adopted is as follows:

[0006] One aspect of the present invention provides a method for estimating the state of charge of a lithium battery, comprising:

[0007] Establish a discrete differential model of the second-order RC equivalent circuit of a lithium battery;

[0008] Based on the multi-innovation identification theory, the error information in the discrete difference model of the second-order RC equivalent circuit of the lithium battery is expanded into a sampling error multi-innovation matrix;

[0009] Parameter identification of the discrete difference model of the second-order RC equivalent circuit of lithium battery with sampling error multi-innovation matrix;

[0010] A lithium battery state of charge model is established based on the identified parameters and the discrete difference model of the lithium battery second-order RC equivalent circuit;

[0011] Based on the multi-innovation identification theory, the noise error information in the lithium battery state of charge model is expanded into a noise error multi-innovation matrix;

[0012] The lithium battery state of charge model with multiple innovation matrices of noise errors is solved to obtain the lithium battery state of charge estimate.

[0013] Furthermore, the discrete difference model of the second-order RC equivalent circuit of the lithium battery is established, including:

[0014] According to Kirchhoff's law, the state equation of the second-order RC equivalent circuit of a lithium battery is:

[0015]

[0016] Among them, R0 is the ohmic internal resistance of the lithium battery, R a and C a Represents the electrochemical polarization resistance and polarization capacitance of lithium batteries, R b and C b Represents the concentration polarization resistance and concentration polarization capacitance of the lithium battery, U oc (t) represents the open circuit voltage of the battery at time t, U L (t) represents the terminal voltage of the battery at time t, I L (t) represents the current flowing through the battery at time t, U a (t) is R a The voltage across the terminals, U b (t) is R b The voltage across the terminals;

[0017] Discretization yields:

[0018]

[0019]

[0020] Where k represents the sampling time, SOC(k) represents the state of charge of the lithium battery at the sampling time k, Q n is the capacity of lithium battery, T is the sampling period, η is the coulombic efficiency of lithium battery;

[0021] The transfer function of the z plane obtained by Laplace transform is:

[0022]

[0023]

[0024] Transformed into a discrete difference equation:

[0025]

[0026] Among them, I(k) is the discrete system input at sampling time k, that is, the lithium battery current, and y(k) is the discrete system output at sampling time k, that is, the lithium battery terminal voltage.

[0027] Furthermore, the error information in the discrete difference model of the second-order RC equivalent circuit of the lithium battery is expanded into a sampling error multi-innovation matrix including:

[0028] The sampling error in the discrete difference model of the second-order RC equivalent circuit of a lithium battery is expressed as:

[0029]

[0030]

[0031] Among them, e(k) is the sampling error at sampling time k, and θ is the parameter vector to be identified;

[0032] Expanded to sampling error multi-innovation matrix:

[0033]

[0034]

[0035] Among them, E(p,k) is the sampling error multi-innovation matrix, and p is the number of innovations in the sampling error multi-innovation matrix.

[0036] Furthermore, the discrete difference model of the second-order RC equivalent circuit of the lithium battery with a sampling error multi-innovation matrix is ​​subjected to parameter identification, including:

[0037] Initialize the parameter vector to be identified, the error covariance matrix and the forgetting factor;

[0038] Perform iterative calculations as follows until the termination condition is reached and output the lithium battery parameter identification results:

[0039] Get the measured voltage and current of the lithium battery at sampling time k, and calculate the gain matrix K(k):

[0040] K(k)=P(k-1)*φ(p,k) / [λ*I p +φ T (p,k)*P(k-1)φ(p,k)];

[0041] Among them, λ is the forgetting factor, I p represents the p-order identity matrix;

[0042] Update the identification parameters θ(k):

[0043] θ(k)=θ(k-1)+K(k)*E(p,k);

[0044] Update the error covariance matrix P(k):

[0045]

[0046] Furthermore, the forgetting factor λ is set to 0.995.

[0047] Furthermore, a lithium battery state of charge model is established based on the identified parameters and a discrete difference model of the second-order RC equivalent circuit of the lithium battery, including:

[0048]

[0049] x k =[U a (k),U b (k), SOC(k)];

[0050]

[0051] Among them, k represents the sampling time, x k is the state variable at sampling time k, u k is the discrete system input at sampling time k, i.e., the lithium battery current, y k The output measured at sampling time k is the lithium battery terminal voltage, w k and v k is Gaussian white noise at k sampling time.

[0052] Furthermore, the noise error information in the lithium battery state of charge model is expanded into a noise error multi-innovation matrix, including:

[0053]

[0054] Among them, E p,k is the noise error multi-innovation matrix, is the estimated value of the state variable based on the k-1 sampling time Predict the estimated value at sampling time k, where p is the number of innovations in the noise error multi-innovation matrix.

[0055] Furthermore, it also includes: assigning weights to different innovations to form a weighted multi-innovation matrix of noise parameters:

[0056]

[0057]

[0058]

[0059]

[0060] in, is the weighted multi-innovation matrix of noise parameters, y k-j+1 is the measured output at sampling time kj, It is the estimated value of the output at time k-j+1 predicted based on the estimated value of the state variable at sampling time kj, e k-j+1 is the matrix E p,k The jth new information vector in k-j+1 Yes k-j+1 The weight of is α k-j+1 The normalized value, σ 2 is the variance of the noise parameter.

[0061] Furthermore, the state of charge model of lithium battery with noise error multi-innovation matrix is ​​solved, including:

[0062] Initialize lithium battery state variables, noise covariance matrix and noise variance matrix;

[0063] Perform iterative calculations as follows until the termination condition is reached and output the lithium battery state of charge:

[0064] Compute the state estimate:

[0065]

[0066] in, is the estimated value of the state variable at the k-1 sampling time;

[0067] Estimate the noise covariance:

[0068]

[0069] Among them, Qk-1 is the noise w k-1 The variance, P k∣k-1 is the noise covariance at sampling time k estimated based on the noise covariance at sampling time k-1, P k-1∣k-1 is the noise covariance estimate at k-1 sampling time;

[0070] Compute the weighted multi-innovation matrix of noise parameters:

[0071]

[0072] Calculate the Kalman gain matrix:

[0073]

[0074] Among them, R k is the noise r k The variance of

[0075] The state variables are modified according to the weighted multi-innovation matrix of the Kalman gain matrix and the noise parameters:

[0076]

[0077] Corrected noise covariance:

[0078] P k∣k =(IK k C k ) k∣k-1 .

[0079] Furthermore, during the calculation process, the noise parameters are updated as follows:

[0080]

[0081] Among them, d k-1 is the intermediate variable, h is the noise forgetting factor, q k w k The mean value, r k v k The mean value, Q k w k The variance, R k v k The variance of is the estimated value of the state variable at the k sampling time, is the estimated value of the state variable based on the k-1 sampling time Predict the estimated value at k sampling time, K k is the Kalman gain at sampling time k, e k is the noise error, P k is the noise covariance at sampling time k.

[0082] Furthermore, the forgetting factor h is set to 0.96.

[0083] The present invention also provides a lithium battery charge state estimation device, comprising:

[0084] The first building block is used to establish a discrete differential model of the second-order RC equivalent circuit of the lithium battery;

[0085] A first expansion module is used to expand the error information in the discrete difference model of the second-order RC equivalent circuit of the lithium battery into a sampling error multi-innovation matrix based on the multi-innovation identification theory;

[0086] Identification module, used to perform parameter identification on the discrete difference model of the second-order RC equivalent circuit of lithium battery with sampling error multi-innovation matrix;

[0087] The second building block is used to establish a lithium battery state of charge model based on the identified parameters and a discrete difference model of the second-order RC equivalent circuit of the lithium battery;

[0088] The second expansion module is used to expand the noise error information in the lithium battery state of charge model into a noise error multi-innovation matrix based on the multi-innovation identification theory;

[0089] as well as,

[0090] The estimation module is used to solve the lithium battery state of charge model with a noise error multi-innovation matrix to obtain the lithium battery state of charge estimate.

[0091] The beneficial effects achieved by the present invention are:

[0092] The present invention proposes a lithium battery state of charge estimation method. On the one hand, it uses the recursive least square method of battery parameters with multiple new information and forgetting factors to make full use of the parameter information contained in the input and output of the system, and identifies the system parameters through iterative calculation. The data utilization rate is higher, the convergence speed is fast, and the accuracy of parameter identification is high. On the other hand, the weighted multi-new information adaptive extended Kalman filter method is used to estimate the battery SOC. In order to reduce the impact of unknown noise on state estimation, the noise parameters are estimated on the basis of the EKF algorithm, and then the new information vectors at multiple moments are used to replace the correction of the estimated value using the new information at the current moment in the original algorithm. In addition, the new data has a higher weight, which can effectively improve the accuracy and has engineering application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0093] Figure 1 This is the second-order RC equivalent circuit model diagram of a lithium battery;

[0094] Figure 2 It is a flow chart of identifying parameters of lithium battery to be identified in the present invention;

[0095] Figure 3 It is a flow chart of the state of charge estimation of the lithium battery in the present invention;

[0096] Figure 4 is an OCV-SOC curve diagram of a lithium battery used in an embodiment of the present invention;

[0097] Figure 5 is a current waveform diagram of a constant power discharge test of a lithium battery used in an embodiment of the present invention;

[0098] Figure 6 This is a current waveform diagram of a lithium battery cycle dynamic pressure test used in an embodiment of the present invention;

[0099] Figure 7 is a voltage error diagram of a parameter identification terminal in a constant power discharge experiment of a lithium battery used in an embodiment of the present invention;

[0100] Figure 8 is a voltage error diagram of a parameter identification terminal in a cyclic dynamic pressure discharge experiment of a lithium battery used in an embodiment of the present invention;

[0101] Fig. 9 Comparison diagram of different SOC algorithm errors in the constant power discharge experiment of the lithium battery used in the embodiment of the present invention;

[0102] Fig.10 It is a comparison diagram of different SOC algorithm errors in the cyclic dynamic pressure discharge experiment of the lithium battery used in the embodiment of the present invention. DETAILED DESCRIPTION

[0103] The present invention is further described below. The following examples are only used to more clearly illustrate the technical solution of the present invention, and are not intended to limit the protection scope of the present invention.

[0104] An embodiment of the present invention provides a method for estimating the state of charge of a lithium battery, comprising the following steps:

[0105] (1) Establish a second-order RC equivalent circuit model of a lithium battery, and according to Kirchhoff's law, obtain its state equation as follows:

[0106]

[0107] Where R0 is the ohmic internal resistance of the lithium battery, R a and C a Represents the electrochemical polarization resistance and polarization capacitance of lithium batteries, R b and C b Represents the concentration polarization resistance and concentration polarization capacitance of the lithium battery, U oc (t) represents the open circuit voltage of the lithium battery at time t, U L (t) represents the terminal voltage of the lithium battery at time t, I L(t) represents the current flowing through the lithium battery at time t, which is positive when discharging. a (t) is R a The voltage across the terminals, U b (t) is R b The voltage across both ends.

[0108] Discretizing the above formula, we can get the discretized state space model as follows:

[0109]

[0110] Where k represents the sampling time, SOC(k) represents the state of charge at the sampling time k,

[0111]

[0112] In the formula, Q n is the battery capacity, T is the sampling period, and η is the battery coulombic efficiency.

[0113] (2) According to the lithium battery electrical quantity relationship established in step (1), the obtained state equation is Laplace transformed:

[0114]

[0115] s is the Laplace operator;

[0116] Then, by using bilinear transformation to discretize it, we can get the transfer function of the z plane:

[0117]

[0118] In the formula, a1, a2, a3, a4, a5 are corresponding constant coefficients.

[0119]

[0120] Transforming the above equation (5) into a differential equation of a discrete system, we can obtain:

[0121]

[0122] Where I(k) is the system input at sampling time k, that is, the current flowing through the lithium battery, and y(k) is the system output at sampling time k, that is, the terminal voltage of the lithium battery.

[0123] (3) Establish a lithium battery parameter identification model, let:

[0124]

[0125] Assuming the sensor sampling error at time k is e(k), we get:

[0126]

[0127] (4) Introduce the forgetting factor λ and calculate the gain matrix.

[0128] (5) According to the multi-innovation theory, the scalar innovation is expanded into an innovation matrix. The innovation refers to the difference between the observed value and the estimated value.

[0129]

[0130] Where p is the number of new information in the new information matrix;

[0131]

[0132] (6) According to the selected lithium battery terminal voltage and working current data, the established second-order RC lithium battery equivalent circuit model is parameter identified by using the multi-innovation band forgetting factor least squares algorithm, that is, the parameter vector θ to be identified is identified until all data points are traversed and the identification is completed, and the identification result of the lithium battery parameter θ to be identified is obtained;

[0133] (7) The discrete state equation of lithium battery is established as follows:

[0134]

[0135] x k =[U a (k),U b (k),SOC(k)]

[0136]

[0137] Among them, k represents the sampling time, x k is the system state variable at sampling time k, u k is the system input signal at sampling time k, i.e., the lithium battery current, y k is the measured output at sampling time k, i.e., the lithium battery terminal voltage, w k and v k is Gaussian white noise, and its mean is q k and r k , and the variances are Q k and R k .

[0138] (8) The noise parameters are estimated and updated over time as follows:

[0139]

[0140] Among them, d k-1 is an intermediate variable, h is the noise forgetting factor, which usually takes a value between 0.95 and 0.99, and here it takes 0.96. is the estimated value of the state quantity at time k, is the estimated value of the state variable at time k-1 Predict the estimated value at time k, K k is the gain at time k, e k is the error between the observed value and the estimated value, i.e., the new information, P k is the error covariance at time k, P k / k-1 It is the error covariance matrix predicted by the model at time k-1.

[0141] (9) The single new information e k Expanded to a multi-innovation matrix E p,k :

[0142]

[0143] Among them, p is the number of new information in the new information matrix.

[0144] At the same time, the gain matrix and state vector also need to be updated as follows:

[0145] K p,k =[K k ,K k-1 ,…K k-p+1 ] (14)

[0146]

[0147] (10) Different weights are assigned to different new information. The weight of new data is greater than that of old data. The functional relationship is as follows:

[0148]

[0149]

[0150] Among them, e k-j+1 is the innovation matrix E p,k The jth new information vector in k-j+1 is its weight, y k-j+1 is the observed quantity at time kj, is the estimated value of the output at time k-j+1 predicted based on the estimated value of the state variable at time kj, σ 2 is the noise variance.

[0151] Normalize the weights:

[0152]

[0153] (11) Based on the selected lithium battery terminal voltage, working current data, and the results of the least squares algorithm parameter identification with multiple innovation band forgetting factors, combined with the OCV-SOC (battery open circuit voltage-battery state of charge) curve, the battery state of charge is estimated until all data points are traversed and the estimation result of the lithium battery state of charge is obtained.

[0154] In step (6) of the embodiment of the present invention, according to the selected lithium battery terminal voltage and working current data, the established second-order RC lithium battery equivalent circuit model is parameter identified by a multi-innovation band forgetting factor least squares algorithm, that is, the parameter vector θ to be identified is identified until all data points are traversed and the identification is completed, and the identification result of the lithium battery parameter θ to be identified is obtained, see Figure 2 , as follows:

[0155] (6.1) Initialization program, assign initial values ​​to the identification parameter vector θ and the covariance matrix P, assign initial values ​​to the forgetting factor λ, set the parameter identification termination conditions, and start the identification program;

[0156] (6.2) Obtain the lithium battery voltage and current measurements at time k, and assign initial values ​​to the data vector φ(k) and the output vector Y(k);

[0157] (6.3) Construct the multi-information matrix E(p,k) and the output matrix Y(p,k);

[0158] (6.4) Calculate the gain matrix K(k) at time k:

[0159] K(k)=P(k-1)*φ(p,k) / [λ*I p +φ T (p,k)*P(k-1)φ(p,k)] (19)

[0160] Among them, I p represents the p-order unit matrix;

[0161] (6.5) Update the identification parameter θ(k):

[0162] θ(k)=θ(k-1)+K(k)*E(p,k) (20)

[0163] (6.6) Update the covariance matrix P(k):

[0164]

[0165] Where I represents the identity matrix.

[0166] (6.7) Determine whether the program termination condition is met. If so, stop the simulation and output the lithium battery parameter identification result; otherwise, k=k+1, return to continue running step (6.4).

[0167] In the above calculation process, the forgetting factor λ takes a fixed value of 0.995.

[0168] Since the true values ​​of the parameters of the second-order RC equivalent circuit model of lithium batteries cannot be directly measured, it is impossible to directly evaluate the accuracy of the parameter identification results of the identification algorithm. By substituting the parameters obtained by the parameter identification algorithm into the state space model, the output value of the second-order RC equivalent circuit of the lithium battery corresponding to each moment of the experiment can be obtained, and the accuracy of the parameter identification algorithm can be evaluated by comparing this model output value with the lithium battery terminal voltage measurement. The error value of the difference between the model output value and the lithium battery terminal voltage measurement is calculated. The smaller the error value, the higher the accuracy of the parameter identification algorithm.

[0169] In step (11) of the embodiment of the present invention, the battery state of charge is estimated based on the selected lithium battery terminal voltage, operating current data, and the result of the least squares algorithm parameter identification with multiple innovation band forgetting factors, combined with the OCV-SOC curve, until all data points are traversed and the estimation result of the lithium battery state of charge is obtained, see Figure 3 , as follows:

[0170] (11.1) Initialization, set the system initial values ​​x0, P0, Q0, R0.

[0171] (11.2) Estimated system status:

[0172]

[0173] (11.3) Estimated error covariance:

[0174]

[0175] (11.4) Calculate the weighted multi-innovation matrix:

[0176]

[0177] (11.5) Calculate the Kalman gain matrix:

[0178]

[0179] (11.6) Correct system status:

[0180]

[0181] (11.7) Corrected error covariance:

[0182] P k∣k =(IK k C k ) k∣k-1 (27)

[0183] (11.8) Determine whether the program termination condition is met. If so, stop the simulation and output the lithium battery charge status result; otherwise, k=k+1, return to continue running step (11.2).

[0184] The OCV-SOC curve acquisition process is as follows:

[0185] The fully charged lithium battery was left to stand for 1 hour, and then the lithium battery was pulse-discharged with a constant amplitude, and the SOC difference was kept at 10% during the pulse discharge interval until the lithium iron phosphate battery reached the cut-off voltage; after standing for 1 hour, the lithium iron phosphate battery was pulse-charged with a constant amplitude, and the SOC difference was kept at 10% during the pulse charging interval until the lithium iron phosphate battery was fully charged. The open circuit voltage value sequence of the lithium iron phosphate battery during the entire charging and discharging process was obtained through the experimental platform, and the mean value of each charge and discharge was obtained to obtain the OCV value corresponding to SOC = 10%*i, i = 0, 1, 2, ... 10, and finally the OCV-SOC curve was fitted by an octal polynomial.

[0186] Example

[0187] The embodiment of the present invention adopts a second-order RC equivalent circuit model of a lithium battery, and the topology is as follows: Figure 1 As shown. The experimental object of this embodiment is a lithium iron phosphate battery with a rated voltage of 3.5V and a rated capacity of 20Ah. In order to ensure the safety of the charge and discharge experiment, the experimental boundary conditions are as follows: the charging overvoltage protection is 3.75V, the discharge cut-off voltage is 2.5V, the overcharge and over-discharge capacity protection is 20Ah, and the charging current is not more than 10A. The fully charged lithium battery is subjected to a constant power discharge test and a cyclic dynamic stress test (dynamic stress test, DST) until the discharge cut-off voltage is 2.5V and the experiment is stopped. The constant power test current waveform is shown as follows Figure 5 The current waveform of the cyclic dynamic pressure test is shown in Figure 6 The voltage and current data obtained from the two working condition experiments are used to verify the parameter identification effect and feasibility of the multi-information with forgetting factor least squares method (MI-FFRLS) and the accuracy of the weighted multi-information adaptive extended Kalman filter algorithm (WMI-AEKF) in estimating SOC.

[0188] The OCV-SOC curve of the lithium battery used in this embodiment is as follows: Figure 4 As shown. Since the actual values ​​of the electrical parameters of the lithium battery cannot be directly obtained, this embodiment evaluates the accuracy of the parameter identification result by comparing the output value of the parameter model identified at each moment with the experimental measured value at the corresponding moment. Figure 7 and Figure 8As shown in the figure, under the constant power discharge test condition, the average error of the terminal voltage can be controlled within 0.015V, and under the cyclic dynamic pressure discharge test condition, the average error of the terminal voltage can be controlled within 0.02V. This shows that the MI-FFRLS algorithm has a high-precision online parameter identification result and can be applied to subsequent SOC estimation.

[0189] The MI-FFRLS algorithm is used for parameter identification, and the ideal SOC value is obtained by the ampere-hour integration method. The SOC estimation is performed in the pulse discharge experiment and the cyclic dynamic pressure test experiment. The EKF algorithm, AEKF algorithm and WMI-AEKF algorithm are used to obtain the battery SOC, and the error is compared with the true value to verify the accuracy of the algorithm. The initial SOC value is set to 0.8. Fig. 9 and Fig.10 As shown in the figure, the SOC value based on the WMI-AEKF algorithm is closer to the true value and its error is also smaller. Under the constant power discharge experimental condition, the average SOC estimation error of the WMI-AEKF algorithm is 0.78%, while the average SOC values ​​of the EKF algorithm and the AEKF algorithm are 1.45% and 1.02% respectively; under the cyclic dynamic pressure discharge experimental condition, the average SOC estimation error of the WMI-AEKF algorithm is 0.69%, while the average SOC values ​​of the EKF algorithm and the AEKF algorithm are 2.47% and 2.02% respectively.

[0190] Based on the results of the above embodiments, it can be seen that the lithium battery state of charge estimation method based on multi-innovation theory of the present invention can adapt to the working conditions of lithium battery nonlinear systems, has high parameter identification accuracy, and small state of charge estimation error.

[0191] Another aspect of the present invention further provides a lithium battery state of charge estimation device, comprising:

[0192] The first building block is used to establish a discrete differential model of the second-order RC equivalent circuit of the lithium battery;

[0193] A first expansion module is used to expand the error information in the discrete difference model of the second-order RC equivalent circuit of the lithium battery into a sampling error multi-innovation matrix based on the multi-innovation identification theory;

[0194] Identification module, used to perform parameter identification on the discrete difference model of the second-order RC equivalent circuit of lithium battery with sampling error multi-innovation matrix;

[0195] The second building block is used to establish a lithium battery state of charge model based on the identified parameters and a discrete difference model of the second-order RC equivalent circuit of the lithium battery;

[0196] The second expansion module is used to expand the noise error information in the lithium battery state of charge model into a noise error multi-innovation matrix based on the multi-innovation identification theory;

[0197] as well as,

[0198] The estimation module is used to solve the lithium battery state of charge model with a noise error multi-innovation matrix to obtain the lithium battery state of charge estimate.

[0199] It is worth pointing out that the device embodiment corresponds to the above-mentioned method embodiment, and the implementation methods of the above-mentioned method embodiments are all applicable to the device embodiment and can achieve the same or similar technical effects, so they will not be repeated here.

[0200] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0201] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0202] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0203] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1The steps for the functions specified in one or more boxes.

[0204] 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 it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for estimating the state of charge of a lithium battery, characterized in that: include: Establish a discrete differential model of the second-order RC equivalent circuit of a lithium battery; Based on the multi-innovation identification theory, the error innovation in the discrete difference model of the second-order RC equivalent circuit of the lithium battery is expanded into a sampling error multi-innovation matrix; Parameter identification of the discrete difference model of the second-order RC equivalent circuit of lithium battery with sampling error multi-innovation matrix; A lithium battery state of charge model is established based on the identified parameters and the discrete difference model of the lithium battery second-order RC equivalent circuit; Based on the multi-innovation identification theory, the noise error innovation in the lithium battery state of charge model is expanded into a noise error multi-innovation matrix; The lithium battery state of charge model with multiple innovation matrices of noise errors is solved to obtain the lithium battery state of charge estimate.

2. A lithium battery state of charge estimation method according to claim 1, characterized in that: The discrete difference model of establishing the second-order RC equivalent circuit of the lithium battery includes: According to Kirchhoff's law, the state equation of the second-order RC equivalent circuit of a lithium battery is: Among them, R0 is the ohmic internal resistance of the lithium battery, R a and C a Represents the electrochemical polarization resistance and polarization capacitance of lithium batteries, R b and C b Represents the concentration polarization resistance and concentration polarization capacitance of the lithium battery, U oc (t) represents the open circuit voltage of the battery at time t, U L (t) represents the terminal voltage of the battery at time t, I L (t) represents the current flowing through the battery at time t, U a (t) is R a The voltage across the terminals, U b (t) is R b The voltage across the terminals; Discretization yields: Where k represents the sampling time, SOC(k) represents the state of charge of the lithium battery at the sampling time k, Q n is the capacity of lithium battery, T is the sampling period, η is the coulombic efficiency of lithium battery; The transfer function of the z plane obtained by Laplace transform is: Transformed into a discrete difference equation: Among them, I(k) is the discrete system input at sampling time k, that is, the lithium battery current, and y(k) is the discrete system output at sampling time k, that is, the lithium battery terminal voltage.

3. A lithium battery state of charge estimation method according to claim 2, characterized in that: Expanding the error innovation in the discrete difference model of the second-order RC equivalent circuit of the lithium battery into a sampling error multi-innovation matrix includes: The sampling error in the discrete difference model of the second-order RC equivalent circuit of a lithium battery is expressed as: Among them, e(k) is the sampling error at sampling time k, and θ is the parameter vector to be identified; Expanded to sampling error multi-innovation matrix: Among them, E(p,k) is the sampling error multi-innovation matrix, and p is the number of innovations in the sampling error multi-innovation matrix.

4. A lithium battery state of charge estimation method according to claim 3, characterized in that: Parameter identification of the discrete difference model of the second-order RC equivalent circuit of a lithium battery with a sampling error multi-innovation matrix is ​​performed, including: Initialize the parameter vector to be identified, the error covariance matrix and the forgetting factor; Perform iterative calculations as follows until the termination condition is reached and output the lithium battery parameter identification results: Get the measured voltage and current of the lithium battery at sampling time k, and calculate the gain matrix K(k): K(k)=P(k-1)*φ(p,k) / [λ*I p +φ T (p,k)*P(k-1)φ(p,k)]; Among them, λ is the forgetting factor, I p represents the p-order identity matrix; Update the identification parameters θ(k): θ(k)=θ(k-1)+K(k)*E(p,k); Update the error covariance matrix P(k):

5. A lithium battery state of charge estimation method according to claim 4, characterized in that: The forgetting factor λ is set to 0.

995.

6. A lithium battery state of charge estimation method according to claim 2, characterized in that: The lithium battery state of charge model is established based on the identified parameters and the discrete difference model of the lithium battery second-order RC equivalent circuit, including: x k =[U a (k),U b (k),SOC(k)]; Among them, k represents the sampling time, x k is the state variable at sampling time k, u k is the discrete system input at sampling time k, i.e., the lithium battery current, y k The output measured at sampling time k is the lithium battery terminal voltage, w k and v k is Gaussian white noise at k sampling time.

7. A lithium battery state of charge estimation method according to claim 6, characterized in that: The noise error innovation in the lithium battery state of charge model is expanded into a noise error multi-innovation matrix, including: Among them, E p,k is the noise error multi-innovation matrix, is the estimated value of the state variable based on the k-1 sampling time Predict the estimated value at sampling time k, where p is the number of innovations in the noise error multi-innovation matrix.

8. A lithium battery state of charge estimation method according to claim 7, characterized in that: Also includes: Assign weights to different innovations to form a weighted multi-innovation matrix of noise parameters: in, is the weighted multi-innovation matrix of noise parameters, y k-j+1 is the measured output at the kj sampling time, e k-j+1 is the matrix E p,k The jth new information vector in k-j+1 Yes k-j+1 The weight of is α k-j+1 The normalized value, y k-j+1 is the observed quantity at the kj sampling time, is the estimated value of the output at time k-j+1 predicted based on the estimated value of the state variable at sampling time kj, σ 2 is the variance of the noise parameter.

9. A lithium battery state of charge estimation method according to claim 8, characterized in that: Solve the lithium battery state of charge model with multiple innovation matrices with noise errors, including: Initialize lithium battery state variables, noise covariance matrix and noise variance matrix; Perform iterative calculations as follows until the termination condition is reached and output the lithium battery state of charge: Compute the state estimate: in, is the estimated value of the state variable at the k-1 sampling time; Estimate the noise covariance: Among them, Q k-1 is the noise w k-1 The variance, P k∣k-1 is the noise covariance at sampling time k estimated based on the noise covariance at sampling time k-1, P k-1∣k-1 is the noise covariance estimate at k-1 sampling time; Compute the weighted multi-innovation matrix of noise parameters: Calculate the Kalman gain matrix: Among them, R k is the noise r k The variance of The state variables are modified according to the weighted multi-innovation matrix of the Kalman gain matrix and the noise parameters: Corrected noise covariance: P k∣k =(I-K k C k )P k∣k-1 。 10. A lithium battery state of charge estimation method according to claim 9, characterized in that: During the calculation, the noise parameters are updated as follows: Among them, d k-1 is the intermediate variable, h is the noise forgetting factor, q k w k The mean value, r k v k The mean value, Q k w k The variance, R k v k The variance of is the estimated value of the state variable at the k sampling time, is the estimated value of the state variable based on the k-1 sampling time Predict the estimated value at k sampling moments, K k is the Kalman gain at sampling time k, e k is the noise error, P k is the noise covariance at sampling time k.

11. A lithium battery state of charge estimation method according to claim 10, characterized in that: The forgetting factor h is set to a value of 0.

96.

12. A lithium battery state of charge estimation device, characterized in that: include: The first building block is used to establish a discrete differential model of the second-order RC equivalent circuit of the lithium battery; A first expansion module is used to expand the error innovation in the discrete difference model of the second-order RC equivalent circuit of the lithium battery into a sampling error multi-innovation matrix based on the multi-innovation identification theory; Identification module, used to perform parameter identification on the discrete difference model of the second-order RC equivalent circuit of lithium battery with sampling error multi-innovation matrix; The second building block is used to establish a lithium battery state of charge model based on the identified parameters and a discrete difference model of the second-order RC equivalent circuit of the lithium battery; The second expansion module is used to expand the noise error innovation in the lithium battery state of charge model into a noise error multi-innovation matrix based on the multi-innovation identification theory; as well as, The estimation module is used to solve the lithium battery state of charge model with a noise error multi-innovation matrix to obtain the lithium battery state of charge estimate.

Images