Power battery remaining capacity online prediction method and system based on electrical fusion model

Abstract

The invention provides a power battery remaining capacity online prediction method and system based on an electrical fusion model. The method comprises the following steps that the electrical fusion model is developed, an equivalent circuit model and a gas-liquid dynamic model are fused, the electrical fusion model is developed, and the battery modeling precision is improved; discretizing an electrical fusion model state equation; an SOC online prediction method is developed, and the reliability, robustness and prediction precision of the power battery online SOC prediction method are improved by applying an electrical fusion model and double extended Kalman filtering; a power battery SOC prediction system is developed, and a signal acquisition module, an SOC prediction module based on an electrical fusion model, a data storage module and a display module are connected and perform data transmission to realize real-time prediction of the remaining capacity of a power battery; and verifying the SOC prediction precision of the power battery through a DST actual measurement working condition. According to the method, the electrical fusion model is innovatively provided by using the gas-liquid dynamic model and the first-order equivalent circuit model, the online prediction method and system for the SOC of the power battery are developed based on the model and the double-extended Kalman filtering, and compared with the same type of technology, the method and system have obvious precision advantages.

Description

Technical Field

[0001] The present invention relates to the field of lithium battery state prediction, specifically to an online prediction method and system for the remaining capacity (State of charge, SOC) of a power battery based on an electrical fusion model, especially involving the fusion of an equivalent circuit model and a gas-liquid dynamics model, and innovatively developing a high-precision electrical fusion model and power battery SOC prediction technology. Background Art

[0002] The power battery is a core component of electric vehicles, and its state of charge (SOC) is a crucial factor in achieving functions such as mileage, output power, brake energy recovery, and charging strategies. Because SOC is an implicit parameter of the power battery and cannot be directly acquired through sensors, research on online prediction methods for power batteries is crucial. Traditional online prediction methods include current integration, estimation methods based on equivalent circuit or electrochemical models, and approaches based on data-driven and machine learning. While each has its own unique characteristics, these methods all have inherent shortcomings. Consequently, current electric vehicles still face problems such as battery charge drop, stalling, and driver range anxiety due to inaccurate battery SOC prediction.

[0003] It is worth noting that in recent years, scholars have continuously explored the advantages of the gas-liquid dynamics model, such as low computational cost and good generalization ability, and have published several battery SOC prediction technologies based on the gas-liquid dynamics model. However, there has been no report on the complementary advantages of the gas-liquid dynamics model and the equivalent circuit model to propose a fusion model with better performance. To this end, the present invention aims to utilize the gas-liquid dynamics model and the first-order equivalent circuit model to innovatively propose an electrical fusion model, and develop a power battery SOC online prediction method and system based on this model and the dual extended Kalman filter, aiming to significantly improve the accuracy of power battery SOC online prediction in engineering applications. Summary of the Invention

[0004] To address these technical challenges, the present invention provides a method and system for online prediction of the remaining charge of a power battery based on an electrical fusion model. By combining an equivalent circuit model with a gas-liquid dynamics model, an innovative electrical fusion model is proposed to improve the simulation accuracy of the battery model. This model is then used to develop a high-precision online SOC prediction method and system.

[0005] The present invention is implemented through the following technical solution: a method and system for online prediction of the remaining power of a power battery based on an electrical fusion model, comprising the following steps:

[0006] Step S1: Developing an electrical fusion model;

[0007] Step S2: Discretize the state equation of the electrical fusion model;

[0008] Step S3: Develop an online SOC prediction method;

[0009] Step S4: Developing a power battery SOC prediction system;

[0010] Step S5: Verify the power battery SOC prediction accuracy;

[0011] In the above scheme, the step S1 of developing the electrical fusion model includes the following steps:

[0012] By complementing the advantages of the equivalent circuit model and the gas-liquid dynamics model, an electrical fusion model is developed.

[0013] The state equation of the equivalent circuit model is:

[0014]

[0015] Among them, SOC d represents the remaining capacity detected by the equivalent circuit model, (t) represents the time variable, η represents the Coulomb efficiency, C N Indicates the rated capacity of the battery, d followed by a letter indicates the differential of the variable, V1 indicates the polarization voltage, R1 indicates the polarization resistance, C1 indicates the polarization capacitance, V indicates the terminal voltage of the equivalent circuit model, OCV indicates the open circuit voltage, R0 indicates the ohmic internal resistance, and I indicates the current, with charging current being positive and discharging current being negative;

[0016] The state equation of the gas-liquid dynamics model is:

[0017]

[0018] Among them, SOC q Indicates the remaining power detected by the gas-liquid dynamics model, U q represents the terminal voltage of the gas-liquid dynamics model, k1, k2, k3, k4, represent the four parameters to be identified in the gas-liquid dynamics model, |I(t)| represents the absolute value of the variable, T1 represents the initial temperature of the battery, T2 represents the current temperature of the battery, OCV I Indicates the initial open circuit voltage of the battery;

[0019] The state equation of the electrical fusion model is:

[0020]

[0021] Wherein, U represents the terminal voltage of the electrical fusion model;

[0022] In the above solution, the step S2 discretizes the state equation of the electrical fusion model;

[0023] State variables of the electrical fusion model: x k =[SOC kV 1,k ] T

[0024] Observed variables of the electrical fusion model: y k =U k

[0025] Parameter vector of the electrical fusion model: θ k =[R0 R1 C1 k3 k4] T

[0026] in,[] T Indicates the transposition of the contents of the brackets;

[0027] x k+1 =f(x k ,I k ,θ k )+μ k Formula 8

[0028] θ k+1 =θ k +ω k Formula 9

[0029]

[0030] U k =g(x k ,I k ,θ k )+ν k Formula 11

[0031]

[0032] The subscript k represents the sampling order, which is 1, 2, 3, ..., [] T represents the transposition of the contents of the brackets, Δt represents the interval of sampling time, μ k Denotes the process noise of Equation 8, ω k represents the process noise of Equation 9, ν k represents the observation noise of Equation 11;

[0033] In the above scheme, the step S3 develops an SOC prediction method;

[0034] SOC prediction based on electrical fusion model and dual extended Kalman filter:

[0035]

[0036]

[0037] [OCV I ,T1]=[OCV(SOCk ),T 2,k Formula 23

[0038] Among them, the "~" symbol above the letter indicates the prior parameter, and the superscript of the letter T represents transpose, P x represents the state covariance matrix, Q x represents the state process noise covariance, F x represents the state transition Jacobian matrix, K x represents the state Kalman gain, R x represents the state observation noise covariance, H x represents the state observation matrix, E(2) represents the second-order identity matrix, V t,k Indicates the collected battery terminal voltage, Adding parameters at the end indicates partial derivative. Obtained by looking up the partial derivative curve table, P θ represents the parameter covariance matrix, Q θ represents the parameter process noise covariance, F θ represents the parameter transfer Jacobian matrix, K θ represents the parameter Kalman gain, R θ represents the parameter observation noise covariance, H θ represents the parameter observation matrix, E(5) represents the fifth-order identity matrix;

[0039] Output x k SOC in k ;

[0040] Go to step S3 (formula 13), and repeat step S3 to achieve online prediction of the remaining power of the power battery.

[0041] The step S4 develops a power battery SOC prediction system, including a signal acquisition module, an SOC prediction module based on an electrical fusion model and a double extended Kalman filter, a data storage module, and a display module;

[0042] The signal acquisition module is used to collect the terminal voltage, surface temperature and current of the battery. The signal acquisition module is connected to the SOC prediction module based on the electrical fusion model and the double extended Kalman filter, and transmits the collected terminal voltage, surface temperature and current signals to the SOC prediction module based on the electrical fusion model and the double extended Kalman filter; the data storage module is connected to the signal acquisition module and the SOC prediction module based on the electrical fusion model, and stores the collected battery terminal voltage, surface temperature, current and predicted SOC; the SOC prediction module based on the electrical fusion model and the double extended Kalman filter is connected to the display module, and the display module displays the predicted SOC in real time as a reference for the driver and the BMS to make decisions.

[0043] The step S5 verifies the power battery SOC prediction accuracy by verifying the battery SOC prediction accuracy through DST actual measurement conditions.

[0044] The beneficial effects of the present invention are as follows: the present invention proposes an electrical fusion model by integrating the equivalent circuit model with the gas-liquid dynamics model, thereby improving the accuracy of power battery modeling; using a double extended Kalman filter to identify the parameters of the power battery electrical fusion model in real time, and online predicting the remaining power of the power battery, thereby improving the generalization ability / robustness and dynamic adaptability of the electrical fusion model and the remaining power prediction method. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 : is an implementation process of an embodiment of the present invention;

[0046] Figure 2 : is an OCV-SOC curve of one embodiment of the present invention;

[0047] Figure 3 :It is an embodiment of the present invention curve;

[0048] Figure 4 : is a system composition of an embodiment of the present invention;

[0049] Figure 5 : is the DST operating condition SOC prediction result of one embodiment of the present invention;

[0050] Figure 6 : This is a partial enlarged diagram of the DST working condition SOC prediction result of one embodiment of the present invention. DETAILED DESCRIPTION

[0051] The embodiments described below with reference to the accompanying drawings are exemplary and are intended to explain the present invention, but should not be construed as limiting the present invention.

[0052] Figure 1 The figure shows the technical route of the method and system for online prediction of the remaining power of a power battery based on an electrical fusion model according to the present invention, which includes the following steps:

[0053] Step S1: Develop an electrical fusion model by fusing the equivalent circuit model with the gas-liquid dynamics model to improve the accuracy of battery modeling;

[0054] Step S2: Discretize the state equation of the electrical fusion model;

[0055] Step S3: Develop an online SOC prediction method, using an electrical fusion model and a dual extended Kalman filter to improve the reliability, robustness, and accuracy of the online SOC prediction method for power batteries;

[0056] Step S4: Developing a power battery SOC prediction system, connecting and transmitting data through a signal acquisition module, an SOC prediction module based on an electrical fusion model, a data storage module, and a display module, to achieve real-time prediction of the remaining power of the power battery;

[0057] Step S5: Verify the power battery SOC prediction accuracy by verifying the power battery remaining capacity prediction accuracy through DST actual measurement conditions;

[0058] In the above scheme, the step S1 of developing the electrical fusion model includes the following steps:

[0059] By complementing the advantages of the equivalent circuit model and the gas-liquid dynamics model, an electrical fusion model is developed.

[0060] The state equation of the equivalent circuit model is:

[0061]

[0062] Among them, SOC d represents the remaining capacity detected by the equivalent circuit model, (t) represents the time variable, η represents the Coulomb efficiency, C N Indicates the rated capacity of the battery, d followed by a letter indicates the differential of the variable, V1 indicates the polarization voltage, R1 indicates the polarization resistance, C1 indicates the polarization capacitance, V indicates the terminal voltage of the equivalent circuit model, OCV indicates the open circuit voltage, R0 indicates the ohmic internal resistance, and I indicates the current, with charging current being positive and discharging current being negative;

[0063] The state equation of the gas-liquid dynamics model is:

[0064]

[0065] Among them, SOC q Indicates the remaining power detected by the gas-liquid dynamics model, U q represents the terminal voltage of the gas-liquid dynamics model, k1, k2, k3, k4, represent the four parameters to be identified in the gas-liquid dynamics model, |I(t)| represents the absolute value of the variable, T1 represents the initial temperature of the battery, T2 represents the current temperature of the battery, OCV I Indicates the initial open circuit voltage of the battery;

[0066] The state equation of the electrical fusion model is:

[0067]

[0068] Wherein, U represents the terminal voltage of the electrical fusion model;

[0069] In the above solution, the step S2 discretizes the state equation of the electrical fusion model;

[0070] State variables of the electrical fusion model: x k =[SOC k V 1,k ] T

[0071] Observed variables of the electrical fusion model: y k =U k

[0072] Parameter vector of the electrical fusion model: θ k =[R0 R1 C1 k3 k4] T

[0073] in,[] T Indicates the transposition of the contents of the brackets;

[0074] x k+1 =f(x k ,I k ,θ k )+μ k Formula 8

[0075] θ k+1 =θ k +ω k Formula 9

[0076]

[0077] U k =g(x k ,I k ,θ k )+ν k Formula 11

[0078]

[0079] The subscript k represents the sampling order, which is 1, 2, 3, ..., [] T represents the transposition of the contents of the brackets, Δt represents the interval of sampling time, μ k Denotes the process noise of Equation 8, ω k represents the process noise of Equation 9, ν k represents the observation noise of Equation 11;

[0080] In the above scheme, the step S3 develops an SOC prediction method;

[0081] SOC prediction based on electrical fusion model and dual extended Kalman filter:

[0082] The parameters of the implemented power battery are shown in Table 1;

[0083] Table 1 Power battery parameters

[0084]

[0085] SOC estimation based on equivalent circuit model and double extended Kalman filter:

[0086] Initialization parameters: k = 1, SOC0 = 0.5, OCV I =V 1,0 =3.7,η=0.999,C N =31,Q θ =0.003, Q x =0.05,R θ =0.002, R x =0.02,P θ,0 =E(5),P x,0 =E(2), T1=293.15, the battery current, terminal voltage, and temperature are acquired in real time, and the open circuit voltage is checked by the curve Figure 2 , find the partial derivative of OCV to SOC and check the curve Figure 3 ;

[0087]

[0088]

[0089] in,

[0090]

[0091] [OCV I T1]=[OCV(SOC k )T 2,k ] Formula 23 In which, the “~” symbol above the letter represents the prior parameter, and the superscript of the letter T represents transpose, P x represents the state covariance matrix, Q x represents the state process noise covariance, F x represents the state transition Jacobian matrix, K x represents the state Kalman gain, R x represents the state observation noise covariance, H x represents the state observation matrix, E(2) represents the second-order identity matrix, V t,k Indicates the collected battery terminal voltage, Adding parameters at the end indicates partial derivative. Obtained by looking up the partial derivative curve table, P θ represents the parameter covariance matrix, Q θ represents the parameter process noise covariance, F θ represents the parameter transfer Jacobian matrix, K θrepresents the parameter Kalman gain, R θ represents the parameter observation noise covariance, H θ represents the parameter observation matrix, E(5) represents the fifth-order identity matrix;

[0092] Output x k SOC in k ;

[0093] Go to step S3 (formula 13), and repeat step S3 to achieve online prediction of the remaining power of the power battery.

[0094] Figure 4 The step S4 shows that the power battery SOC prediction system is developed, including a signal acquisition module, an SOC prediction module based on an electrical fusion model and a double extended Kalman filter, a data storage module, and a display module;

[0095] The signal acquisition module is used to collect the terminal voltage, surface temperature and current of the battery. The signal acquisition module is connected to the SOC prediction module based on the electrical fusion model and the double extended Kalman filter, and transmits the collected terminal voltage, surface temperature and current signals to the SOC prediction module based on the electrical fusion model and the double extended Kalman filter; the data storage module is connected to the signal acquisition module and the SOC prediction module based on the electrical fusion model, and stores the collected battery terminal voltage, surface temperature, current and predicted SOC; the SOC prediction module based on the electrical fusion model and the double extended Kalman filter is connected to the display module, and the display module displays the predicted SOC in real time as a reference for the driver and the BMS to make decisions.

[0096] Figure 5 and Figure 6 The step S5 is displayed to verify the power battery SOC prediction accuracy, and the battery SOC prediction accuracy is verified through the DST actual measurement condition; Figure 5 It is shown that under the DST measured working conditions, the maximum prediction error of the online prediction method and system for the remaining power of the power battery based on the electrical fusion model is 1.18%. Figure 6 The partially enlarged view shows that the initial remaining power error SOC=50% is eliminated in only about 4 seconds.

[0097] It should be understood that although this specification is described according to the various steps of the embodiments, not each step contains only an independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each implementation step can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

[0098] The detailed descriptions listed above are merely specific descriptions of the feasibility of implementing the present invention. They are not intended to limit the scope of protection of the present invention. Any equivalent embodiments or changes that do not deviate from the technical spirit of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method and system for online prediction of remaining power of a power battery based on an electrical fusion model, characterized in that: The steps include: Step S1: Developing an electrical fusion model; Step S2: Discretize the state equation of the electrical fusion model; Step S3: Develop an SOC online prediction method; Step S4: Developing a power battery SOC prediction system; Step S5: Verify the power battery SOC prediction accuracy; The step S1 develops an electrical fusion model, characterized in that the electrical fusion model is developed by complementing the advantages of the equivalent circuit model and the gas-liquid dynamics model; The state equation of the equivalent circuit model is: Among them, SOC d represents the remaining capacity detected by the equivalent circuit model, (t) represents the time variable, η represents the Coulomb efficiency, C N Indicates the rated capacity of the battery, d followed by a letter indicates the differential of the variable, V1 indicates the polarization voltage, R1 indicates the polarization resistance, C1 indicates the polarization capacitance, V indicates the terminal voltage of the equivalent circuit model, OCV indicates the open circuit voltage, R0 indicates the ohmic internal resistance, and I indicates the current, with charging current being positive and discharging current being negative; The state equation of the gas-liquid dynamics model is: Among them, SOC q Indicates the remaining power detected by the gas-liquid dynamics model, U q represents the terminal voltage of the gas-liquid dynamics model, k1, k2, k3, k4, represent the four parameters to be identified in the gas-liquid dynamics model, |I(t)| represents the absolute value of the variable, T1 represents the initial temperature of the battery, T2 represents the current temperature of the battery, OCV I Indicates the initial open circuit voltage of the battery; The state equation of the electrical fusion model is: Wherein, U represents the terminal voltage of the electrical fusion model; In the above solution, the step S2 discretizes the state equation of the electrical fusion model; State variables of the electrical fusion model: x k =[SOC k V 1,k ] T Observed variables of the electrical fusion model: y k =U k Parameter vector of the electrical fusion model: θ k =[R0 R1 C1 k3 k4] T in,[] T Indicates the transposition of the contents of the brackets; x k+1 =f(x k ,I k ,θ k )+μ k Formula 8 θ k+1 =θ k +ω k Formula 9 U k = g(x k , I k , θ k ) + ν k Formula XI The subscript k represents the sampling order, which is 1, 2, 3, ..., [] T represents the transposition of the contents of the brackets, Δt represents the interval of sampling time, μ k Denotes the process noise of Equation 8, ω k represents the process noise of Equation 9, ν k represents the observation noise of Equation 11; The step S3 develops an SOC prediction method, characterized in that: SOC prediction based on electrical fusion model and dual extended Kalman filter: [OCV I ,T1]=[OCV(SOC k ),T 2,k Formula 23 The "~" symbol above the letter indicates the prior parameter, and the superscript of the letter T represents transpose, P x represents the state covariance matrix, Q x represents the state process noise covariance, F x represents the state transition Jacobian matrix, K x represents the state Kalman gain, R x represents the state observation noise covariance, H x represents the state observation matrix, E(2) represents the second-order identity matrix, V t,k Indicates the collected battery terminal voltage, Adding parameters at the end indicates partial derivative. Obtained by looking up the partial derivative curve table, P θ represents the parameter covariance matrix, Q θ represents the parameter process noise covariance, F θ represents the parameter transfer Jacobian matrix, K θ represents the parameter Kalman gain, R θ represents the parameter observation noise covariance, H θ represents the parameter observation matrix, E(5) represents the fifth-order identity matrix; Output x k SOC in k ; Go to step S3 (formula 13), and repeat step S3 to achieve online prediction of the remaining power of the power battery.

2. The method and system for online prediction of remaining power of a power battery based on an electrical fusion model according to claim 1, characterized in that: The step S4 develops a power battery SOC prediction system, which includes a signal acquisition module, an SOC prediction module based on an electrical fusion model and a double extended Kalman filter, a data storage module, and a display module.

3. According to the method and system for online prediction of remaining power of a power battery based on an electrical fusion model in claim 2, the step S4 develops a power battery SOC prediction system, characterized in that: The signal acquisition module is used to collect the terminal voltage, surface temperature and current of the battery. The signal acquisition module is connected to the SOC prediction module based on the electrical fusion model and the double extended Kalman filter, and transmits the collected terminal voltage, surface temperature and current signals to the SOC prediction module based on the electrical fusion model and the double extended Kalman filter; the data storage module is connected to the signal acquisition module and the SOC prediction module based on the electrical fusion model, and stores the collected battery terminal voltage, surface temperature, current and predicted SOC; the SOC prediction module based on the electrical fusion model and the double extended Kalman filter is connected to the display module, and the display module displays the predicted SOC in real time as a reference for the driver and the BMS to make decisions.

4. The method and system for online prediction of remaining power of a power battery based on an electrical fusion model according to claim 1, characterized in that: The step S5 verifies the power battery SOC prediction accuracy by verifying the battery SOC prediction accuracy through DST actual measurement conditions.

Images