A new energy vehicle lithium battery SOC estimation method based on RGC-PF algorithm

Abstract

A new energy automobile lithium battery SOC estimation method based on RGC-PF algorithm belongs to the technical field of automobile power battery, aiming at the actual situation of limited memory and operation capacity of vehicle chip, solves the problem of large operation amount and long operation time of the lithium battery SOC estimation method of the prior art, the nonlinear characteristics of the lithium battery and the actual situation of limited operation capacity of the vehicle chip are considered, the RGC algorithm is introduced into the lithium battery equivalent circuit identification, the parameter identification of the lithium battery equivalent circuit model is realized, the RGC algorithm is integrated into the PF algorithm, and the estimation of the battery SOC is realized under the city road cycle condition, the environmental adaptability is strong, and the accuracy is higher; compared with the traditional least square method identification, the RGC algorithm identification ECM is simpler and more understandable, and the required identification time is shorter.

Description

Technical Field

[0001] This invention belongs to the field of automotive power battery technology, and relates to a method for estimating the state of charge (SOC) of lithium batteries for new energy vehicles based on the RGC-PF algorithm. Background Technology

[0002] With the increasing popularity of new energy vehicles, the range and lifespan of automotive power batteries have received growing attention and importance. Inaccurate estimation of the state of charge (SOC) of lithium batteries under driving conditions will affect the operation of the battery management system (BMS) and vehicle mileage control. Traditional SOC estimation methods based on capacity discharge have certain lag and inaccuracies, preventing the BMS from accurately analyzing and managing the state of lithium batteries in real time. To address this, existing technologies have proposed an SOC estimation method based on the Particle Filter (PF) algorithm, which is more real-time and accurate. However, accurate estimation of lithium battery SOC by the PF algorithm requires precise parameters of the lithium battery's equivalent circuit model (ECM).

[0003] The acquisition of lithium battery ECM parameters can be broadly categorized into offline and online parameter identification. Currently, the commonly used HPPC experiment, which uses battery external characteristic curve fitting to obtain the equivalent circuit parameters of the battery under different conditions, is an offline identification method. However, it suffers from poor environmental adaptability and low accuracy. Patent application CN201710421602.8 proposes a forgetting factor recursive least squares (FFRLS) method for lithium battery model parameter identification, which is an online identification method. It boasts strong environmental adaptability, can be used under various operating conditions, and offers high accuracy. However, the FFRLS algorithm suffers from high computational complexity, slow processing speed, and difficulty in selecting the forgetting factor. Patent application CN202011162087.4 proposes a method for identifying parameters of marine lithium batteries based on improved particle swarm optimization; patent application CN202110370086.7 proposes a method for identifying parameters of lithium battery equivalent circuit models based on an adaptive beetle-beard optimized neural network; and patent application CN202110044991.3 proposes a method for identifying parameters of lithium battery equivalent circuit models based on an improved ant colony algorithm. All of these are intelligent algorithms. Although they have higher accuracy and are suitable for identifying ECM parameters of lithium batteries under different operating conditions, their algorithms are complex and computationally intensive. In the actual situation where the memory and computing power of automotive chips are limited, they consume a lot of memory and take a long time to operate, which affects the operation results of BMS. Summary of the Invention

[0004] The purpose of this invention is to design a new energy vehicle lithium battery SOC estimation method based on the RGC-PF algorithm, which addresses the limitations of onboard chip memory and computing power, and solves the problems of large computational load and long computation time in existing lithium battery SOC estimation methods.

[0005] The present invention solves the above-mentioned technical problems through the following technical solutions:

[0006] A method for estimating the state of charge (SOC) of lithium batteries for new energy vehicles based on the RGC-PF algorithm includes the following steps:

[0007] S1. Establish the equivalent circuit model of the lithium battery;

[0008] S2. Using the hybrid pulse power characteristic test method (HPPC) for lithium batteries, charge and discharge tests were conducted on lithium batteries with different states of charge to obtain response curves and calculate the open-circuit voltage of the equivalent circuit model of lithium batteries under different states of charge.

[0009] S3. Measure and record the load current and the actual measured voltage at both ends of the lithium battery.

[0010] S4. Using the battery open-circuit voltage, the actual measured voltage across the battery terminals, and the load current across the battery terminals as the input and output values ​​of the criterion function, and using the battery internal resistance, battery electrochemical polarization resistance, and battery electrochemical polarization capacitance as the parameter estimates of the criterion function, establish a parameter estimation recursive formula and construct the criterion function.

[0011] S5. Use the RGC algorithm to correct the parameter estimates of the criterion function according to the negative gradient direction of the criterion function until the criterion function reaches its minimum value.

[0012] S6. Calculate the terminal voltage of the lithium battery at the next moment based on the parameter estimates, and use the PF algorithm to estimate the SOC of the lithium battery.

[0013] This invention takes into account the nonlinear characteristics of lithium batteries and the limited computing power of on-board chips. The RGC algorithm is introduced into the identification of the equivalent circuit of lithium batteries, realizing online identification of the parameters of the equivalent circuit model of lithium batteries. The RGC algorithm is then integrated into the PF algorithm, realizing the estimation of battery SOC under urban road cycle conditions. It has strong environmental adaptability and higher accuracy. Compared with the traditional least squares identification method, the RGC algorithm is simpler and easier to understand to identify ECM and requires less identification time.

[0014] Furthermore, the process of establishing the equivalent circuit model of the lithium battery in step S1 is as follows:

[0015] The transfer function of the equivalent model of the first-order circuit is as follows:

[0016]

[0017] Using bilinear transformation, we get:

[0018]

[0019] make:

[0020] , ,

[0021] The equivalent circuit model of a lithium battery is:

[0022]

[0023] In actual measurements, the voltage and current signals are discrete quantities. The relationship between the variables in the equivalent circuit model of the lithium battery is transformed into a difference equation:

[0024]

[0025] Substituting a1, a2, and a3 into the difference equation yields the recursive formula:

[0026]

[0027] in, The actual measured voltage of the battery at this moment; The battery open-circuit voltage at this moment was obtained by HPPC testing; The load current across the battery at this moment; This represents the load current across the battery terminals at the previous moment; For the battery's internal resistance in ohms, For the battery's electrochemical polarization resistance, It is the electrochemical polarization capacitor of the battery.

[0028] Further, in step S2, the lithium battery hybrid pulse power characteristic test method is used to conduct charge and discharge tests on lithium batteries with different states of charge to obtain response curves and calculate the open-circuit voltage of lithium batteries under different states of charge. Specifically, constant current charge and discharge tests are conducted on lithium batteries with rated capacity of SOC of 1, 0.9...0.1, respectively, following the process of 1C discharge for 10s → rest for 30s → 1C charge for 10s → rest for 30s. The rated current curve under rated capacity is applied to the lithium battery cell to obtain the battery response curve. Based on the battery response curve, the open-circuit voltage is calculated.

[0029] Furthermore, the measurement and recording of the load current and the actual measured voltage across the lithium battery terminals in step S3 specifically includes: measuring the load current across the lithium battery terminals at this moment. At this moment, the actual measured voltage of the battery Record the load current across the battery terminals at the previous moment. The actual measured voltage of the battery at the previous moment. The battery open-circuit voltage at the previous moment The estimated ohmic internal resistance of the battery at the previous moment. Estimated value of battery electrochemical polarization resistance at the previous moment Estimated value of battery electrochemical polarization resistance at the previous moment .

[0030] Furthermore, the method for constructing the criterion function described in step S4 is as follows: [The method involves] measuring the actual voltage of the battery at this moment... With the current battery open-circuit voltage The difference As the output value of the criterion function, the load current across the battery at the previous moment. The load current across the battery at this moment The actual measured voltage of the battery at the previous moment. The battery open-circuit voltage at the previous moment The estimated ohmic internal resistance of the battery at this moment is used as the input value of the criterion function. Estimated value of battery electrochemical polarization resistance at this moment At this moment, the estimated value of the battery's electrochemical polarization capacitance The parameter estimate at this moment, used as the criterion function, is the battery ohmic internal resistance estimate from the previous moment. Estimated value of battery electrochemical polarization resistance at the previous moment The estimated value of the battery electrochemical polarization capacitance at the previous moment The criterion function is constructed using the parameter estimates from the previous time step.

[0031] Furthermore, the formula for the criterion function mentioned in step S4 is as follows:

[0032]

[0033] In the formula:

[0034]

[0035]

[0036]

[0037]

[0038] in, The output value of the criterion function at this moment The output value of the criterion function at this moment; The parameter estimate for this moment; The estimated ohmic internal resistance of the battery at this moment, The estimated value of the battery's electrochemical polarization resistance at this moment. This is the estimated value of the battery's electrochemical polarization capacitance at this moment.

[0039] Furthermore, the parameter estimation recursive formula established in step S4 is as follows:

[0040]

[0041] in, These are the parameter estimates from the previous time step. The parameters are set.

[0042] Furthermore, in step S5, the RGC algorithm is used to correct the parameter estimates of the criterion function according to the negative gradient direction of the criterion function until the criterion function reaches its minimum value, as detailed below:

[0043] The parameter estimation recursive formula satisfies the following for parameter m:

[0044]

[0045] The recursive formula for parameter estimation is expressed as follows:

[0046]

[0047] Repeat the calculation of the parameter estimation recursive formula until the parameter identification is completed.

[0048] Furthermore, the method for calculating the lithium battery terminal voltage at the next moment based on the parameter estimate and estimating the lithium battery SOC using the PF algorithm in step S6 is as follows:

[0049] 1) Initialize parameters: First, set the initial lithium battery SOC and variance R. ω and R ν Then, N initial random particles are generated, denoted as , i=1,2,3…N;

[0050] 2) Use the RGC algorithm to identify ECM parameters, verify the accuracy of particle positions in the particle swarm optimization algorithm, and thus correct the positions of the particles.

[0051] 3) Status update as follows:

[0052]

[0053]

[0054] By substituting the optimized N particle states from the previous moment into the state equation and measurement equation of the lithium battery, the N particle states of the lithium battery and the voltage across the battery terminals at the next moment can be estimated.

[0055] 4) Update measurement data

[0056] The calculation error is:

[0057]

[0058] in, The estimated value of the i-th particle at time k is obtained by calculating the state equation and the measurement equation. The measured voltage value at time k;

[0059] Calculate importance weights:

[0060]

[0061] When the deviation between the measured voltage value and the estimated value obtained from the state equation is small, the importance weight value of the particle is large; conversely, when the deviation is large, the importance weight value of the particle is low.

[0062] Normalized importance weights:

[0063]

[0064] 5) Resampling

[0065] If the following equation holds, then resampling will be performed. After resampling, the weights of particles near the true state will increase. In the following equation, N threshold The threshold for determining whether to resample;

[0066]

[0067] 6) Output

[0068] State estimation:

[0069]

[0070] Particles with larger weights contribute more to the optimal value, while particles with smaller weights contribute less. As the number of iterations increases, the deviation between the optimal estimate and the actual value continuously decreases.

[0071] Variance estimation:

[0072]

[0073] 7) Determine if the process has ended. If not, return to step 3). If it has ended, exit.

[0074] The advantages of this invention are:

[0075] This invention considers the nonlinear characteristics of lithium batteries and the limited computing power of automotive chips. It utilizes a recursive gradient correction (RGC) algorithm to effectively identify the parameters of the lithium battery's equivalent circuit model. Compared to HPPC experiments that use battery external characteristic curve fitting to obtain the battery's equivalent circuit parameters under different conditions, the RGC-based lithium battery ECM parameter identification method is an online parameter identification method. Furthermore, by integrating the RGC algorithm into the PF algorithm, it achieves battery SOC estimation under urban road cyclic conditions, making it applicable to various operating conditions, with strong environmental adaptability and higher accuracy. Compared to the FFRLS algorithm, the RGC algorithm does not require calculating the forgetting factor, resulting in lower computational load, faster processing speed, and higher accuracy. Compared to intelligent algorithms such as improved particle swarm optimization, adaptive beetle-beard optimization neural network, and improved ant colony optimization, the RGC algorithm requires less memory, has lower computational load, and is faster with the same computing power, making it more suitable for the limited memory and computing power of automotive chips. Attached Figure Description

[0076] Figure 1 This is a flowchart of the SOC estimation method for lithium batteries in new energy vehicles based on the RGC-PF algorithm in Embodiment 1 of the present invention;

[0077] Figure 2 This is the lithium battery equivalent circuit model of Embodiment 1 of the present invention;

[0078] Figure 3 These are the current curves and battery terminal voltage response curves applied to the lithium battery by the HPPC test in Embodiment 1 of the present invention.

[0079] Figure 4 This is a flowchart of the RGC-PF algorithm according to Embodiment 1 of the present invention;

[0080] Figures 5(a) and 5(b) are respectively the effect diagrams of the intermittent constant current discharge test and the effect diagrams of the charge and discharge test under the urban road cycle condition simulating the real working conditions of a car in Embodiment 1 of the present invention.

[0081] Figures 6(a), 6(b), and 6(c) are respectively the DC internal resistance parameters, polarization resistance parameters, and polarization capacitance identification results of the ECM in Embodiment 1 of the present invention.

[0082] Figure 7 This refers to the lithium battery fitting terminal voltage error in Embodiment 1 of the present invention;

[0083] Figure 8 This refers to the SOC estimation error of the RGC-PF algorithm in Embodiment 1 of the present invention. Detailed Implementation

[0084] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0085] The technical solution of the present invention will be further described below with reference to the accompanying drawings and specific embodiments:

[0086] Example 1

[0087] like Figure 1 As shown, a method for estimating the state of charge (SOC) of lithium batteries for new energy vehicles based on the RGC-PF algorithm includes the following steps:

[0088] S1. Establish the equivalent circuit model of the lithium battery;

[0089] S2. Using the lithium battery hybrid pulse power characteristic test method, charge and discharge tests were conducted on lithium batteries with different states of charge to obtain response curves, and the open-circuit voltage parameters of the equivalent circuit model of lithium batteries under different states of charge were calculated.

[0090] S3. Measure and record the load current and the actual measured voltage at both ends of the lithium battery.

[0091] S4. Using the battery open-circuit voltage, the actual measured voltage across the battery terminals, and the load current across the battery terminals as the input and output values ​​of the criterion function, and using the battery internal resistance, battery electrochemical polarization resistance, and battery electrochemical polarization capacitance as the parameter estimates of the criterion function, establish a parameter estimation recursive formula and construct the criterion function.

[0092] S5. Use the RGC algorithm to correct the parameter estimates of the criterion function according to the negative gradient direction of the criterion function until the criterion function reaches its minimum value.

[0093] S6. Calculate the terminal voltage of the lithium battery at the next moment based on the parameter estimates, and use the PF algorithm to estimate the SOC of the lithium battery.

[0094] like Figure 2 As shown, the equivalent circuit model of a lithium battery includes the battery's ohmic internal resistance R0, the battery's electrochemical polarization capacitance C, and the battery's electrochemical polarization resistance R1, connected by a resistor, capacitor, and voltage source U. ocv The parameter changes simulate the changes in the external characteristic curves of lithium batteries under different states and environments; where U ct I represents the actual measured voltage across the battery terminals. L This represents the load current across the battery terminals.

[0095] Based on the knowledge of circuit principles, we can conclude that:

[0096] (1)

[0097] Using bilinear transformation, we can obtain:

[0098] (2)

[0099] make:

[0100] , , (3)

[0101] The equivalent circuit model of a lithium battery is:

[0102] (4)

[0103] Since the voltage and current signals obtained in actual measurements are discrete quantities, it is necessary to transform the relationship between the variables in the lithium battery equivalent circuit model into a difference equation. By transforming formula (4) into a difference equation, we can obtain:

[0104] (5)

[0105] Substituting formula (3) into formula (5) yields the recursive formula:

[0106] (6)

[0107] in, The actual measured voltage of the battery at this moment; The battery open-circuit voltage at this moment was obtained by HPPC testing; The load current across the battery at this moment; This represents the load current across the battery terminals at the previous moment; For the battery's internal resistance in ohms, For the battery's electrochemical polarization resistance, It is the electrochemical polarization capacitor of the battery.

[0108] Because the parameters of the equivalent circuit model of a lithium battery are easily affected by environmental and operating conditions, traditional constant-parameter equivalent circuit models cannot reflect the impact of these factors on the battery. Therefore, it is necessary to update the model parameters in real time. Recursive gradient correction (RGC) corrects the parameter estimates along the negative gradient direction of the criterion function until the criterion function reaches its minimum value. Compared with the least squares method, although RGC converges more slowly, it significantly reduces computational cost and running time.

[0109] In formula (7), θ is the parameter to be identified;

[0110]

[0111] (7)

[0112] Assume the system parameter estimates are as follows:

[0113] (8)

[0114] That is, the estimate of θ at time k is the result of a correction based on the estimate of θ at time k-1.

[0115] The choice of parameter m in formula (8) should satisfy the following formula (9):

[0116] (9)

[0117] Substituting equation (9) into equation (8), we get:

[0118] (10)

[0119] Repeat the above steps until parameter identification is complete.

[0120] like Figure 3 As shown, the HPPC method was used to conduct charge-discharge tests on lithium batteries, obtaining response curves and calculating the open-circuit voltage of lithium batteries under different nuclear power states. Charge-discharge tests: Constant current charge-discharge tests were conducted on 15Ah lithium batteries with SOCs of 1, 0.9…0.1%, respectively. The test procedure was: 1C discharge for 10s → rest for 30s → 1C charge for 10s → rest for 30s. Since the battery capacity is 15Ah, the charge-discharge current I at a discharge rate of 1C is 15A. The current curves applied to individual battery cells and the battery response curves are shown in the attached figure. Figure 3 As shown; where, before point U1, the battery is in a static state; U1-U2 represents the instantaneous voltage change during discharge; U2-U3 represents the voltage change during discharge; U3-U4 represents the instantaneous voltage change at the end of discharge; U4-U5 represents the voltage change when the battery is static after discharge; U5-U6 represents the instantaneous voltage change during charging; U6-U7 represents the voltage change during charging; U7-U8 represents the instantaneous voltage change at the end of charging; and U8-U9 represents the voltage change when the battery is static after charging. Based on the battery's response curve, parameters such as open-circuit voltage, battery internal resistance, electrochemical polarization resistance, and electrochemical polarization capacitance are calculated. The average voltage 30 seconds after the battery finishes discharging and 30 seconds after the battery finishes charging is taken as the open-circuit voltage of the corresponding SOC. The calculation formula is as follows:

[0121] (11)

[0122] like Figure 4 As shown, the PF algorithm is a nonlinear filtering algorithm widely used for estimating the state of charge (SOC) of lithium batteries. In the PF algorithm's SOC estimation process, it is necessary to estimate the battery terminal voltage at the next moment based on the parameters of the lithium battery's equivalent circuit model. Applying the RGC algorithm to the PF algorithm allows for real-time identification of the lithium battery's equivalent circuit model parameters, thereby more accurately estimating the terminal voltage at the next moment under different operating conditions. This includes the following steps:

[0123] 1) Initialize parameters

[0124] First, set the initial state of charge (SOC) of the lithium battery, and the variance R. ω and R ν Then, N initial random particles are generated, denoted as , i=1,2,3…N.

[0125] 2) Model parameter identification

[0126] The recursive gradient correction algorithm is used to identify ECM parameters, which facilitates the verification of the accuracy of particle positions in the particle swarm optimization algorithm, thereby correcting the positions of the particles.

[0127] 3) Status Update

[0128] (12)

[0129] (13)

[0130] By substituting the optimized N particle states from the previous moment into the state equation and measurement equation of the lithium battery, the N particle states of the lithium battery and the voltage across the battery terminals at the next moment can be estimated.

[0131] 4) Update measurement data

[0132] The calculation error is:

[0133] (14)

[0134] in, The estimated value of the i-th particle at time k is obtained by calculating the state equation and the measurement equation. The measured voltage value at time k is given.

[0135] Calculate importance weights:

[0136] (15)

[0137] When the deviation between the measured voltage value and the estimated value obtained from the state equation is small, the importance weight value of the particle is large; conversely, when the deviation is large, the importance weight value of the particle is low.

[0138] Normalized importance weights:

[0139] (16)

[0140] 5) Resampling

[0141] If formula (17) holds, then resampling is performed. After resampling, the particle weights near the true state will increase. In formula (17), N threshold The threshold used to determine whether to resample.

[0142] (17)

[0143] 6) Output

[0144] State estimation:

[0145] (18)

[0146] Particles with larger weights contribute more to the optimal value, while particles with smaller weights contribute less. As the number of iterations increases, the deviation between the optimal estimate and the actual value continuously decreases.

[0147] Variance estimation:

[0148] (19)

[0149] 7) Determine if the process has ended. If not, return to step 3). If it has ended, exit.

[0150] Experimental verification

[0151] Lithium iron phosphate (LFP) batteries are among the most widely used power batteries in the electric vehicle industry. To verify the effectiveness of the RGC-PF algorithm, a 15AH LFP battery manufactured by Guoxuan High-Tech Co., Ltd. was used to conduct tests under two different operating conditions. Figure 5(a) shows the intermittent constant current discharge test, and Figure 5(b) shows the charge-discharge test under the urban dynamometer driving schedule (UDDS) simulating real-world automotive conditions.

[0152] Figures 6(a), 6(b), and 6(c) show the equivalent circuit parameters identified by the recursive gradient correction algorithm during intermittent constant current discharge testing. It can be seen that the identified parameters fluctuate within a certain range. This is due to the differences in the external characteristics of the battery under different states, which cause the identified parameters to change continuously. The estimated value of the voltage across the battery terminals can be obtained using the open-circuit voltage curve and the identified equivalent circuit parameters. To analyze the accuracy of the parameters identified by the RGC algorithm, the ECM parameters identified by the FFRLS algorithm and the RGC algorithm were substituted into the formula to obtain U. ct The estimated values ​​at both ends are compared. The FFRLS algorithm has a higher computational complexity than the RGC algorithm, which is intuitively reflected in the fact that the FFRLS algorithm takes more time to identify the equivalent model parameters under the same operating conditions. Since the RGC algorithm mainly performs scalar operations, it has a lower computational cost when identifying parameters. In Table 1, the RGC algorithm takes less time to identify the equivalent model parameters compared to the FFRLS algorithm. Furthermore, the root mean square error (RMSE) of the terminal voltage fitted by FFRLS is close to that fitted by RGC. Similarly, Figure 7 The error between the fitted terminal voltage and the actual terminal voltage is calculated during the intermittent constant current discharge test. It can be seen that although the convergence speed of RGC is slower than that of FFRLS, the estimation effect after convergence is no worse than that of FFRLS.

[0153] Table 1 Comparison of RGC and FFRLS under different operating conditions

[0154]

[0155] Figure 8 The RGC-PF algorithm estimates the SOC of lithium batteries. The equivalent circuit model parameters are estimated using the recursive gradient correction algorithm. The SOC estimation error of the PF algorithm quickly converges to within 2%, which verifies the feasibility of the RGC-PF algorithm for estimating SOC.

[0156] This invention considers the nonlinear characteristics of lithium batteries and the limited computing power of automotive chips. It utilizes a recursive gradient correction (RCC) algorithm to effectively identify the parameters of the equivalent circuit model (ECM) of lithium batteries. Compared to HPPC experiments, which use external characteristic curve fitting to obtain the ECM parameters of the battery under different conditions, the RGC-based method for identifying lithium battery ECM parameters is an online method, applicable to various operating conditions, with strong environmental adaptability and higher accuracy. Compared to the FFRLS algorithm, the RGC algorithm does not require calculation of the forgetting factor, resulting in lower computational load, faster processing speed, and higher accuracy. Compared to intelligent algorithms such as improved particle swarm optimization, adaptive beetle-beard optimization neural network, and improved ant colony optimization, the RGC algorithm requires less memory, has lower computational load, and is faster with equivalent computing power, making it more suitable for the limited memory and computing power of automotive chips.

[0157] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for estimating the state of charge (SOC) of lithium batteries for new energy vehicles based on the RGC-PF algorithm, characterized in that, Includes the following steps: S1. Establish the equivalent circuit model of the lithium battery; S2. Using the lithium battery hybrid pulse power characteristic test method, charge and discharge tests were conducted on lithium batteries with different states of charge to obtain response curves, and the open-circuit voltage parameters of the equivalent circuit model of lithium batteries under different states of charge were calculated. S3. Measure and record the load current and the actual measured voltage at both ends of the lithium battery. S4. Using the battery open-circuit voltage, the actual measured voltage across the battery terminals, and the load current across the battery terminals as the input and output values ​​of the criterion function, and using the battery internal resistance, battery electrochemical polarization resistance, and battery electrochemical polarization capacitance as the parameter estimates of the criterion function, establish a parameter estimation recursive formula and construct the criterion function. S5. Use the RGC algorithm to correct the parameter estimates of the criterion function according to the negative gradient direction of the criterion function until the criterion function reaches its minimum value. S6. Calculate the terminal voltage of the lithium battery at the next moment based on the parameter estimates, and use the PF algorithm to estimate the SOC of the lithium battery.

2. The method for estimating the SOC of lithium batteries for new energy vehicles based on the RGC-PF algorithm according to claim 1, characterized in that, The process of establishing the equivalent circuit model of the lithium battery in step S1 is as follows: Based on circuit knowledge, the equivalent model transfer function can be obtained: Using bilinear transformation, we get: make: ， ， The equivalent circuit model of a lithium battery is: In actual measurements, the voltage and current signals are discrete quantities. The relationship between the variables in the equivalent circuit model of the lithium battery is transformed into a difference equation, which yields: Substituting a1, a2, and a3 into the difference equation yields the recursive formula: in, The actual measured voltage of the battery at this moment; The battery open-circuit voltage at this moment was obtained by HPPC testing; The load current across the battery at this moment; This represents the load current across the battery terminals at the previous moment; For the battery's internal resistance in ohms, For the battery's electrochemical polarization resistance, It is the electrochemical polarization capacitor of the battery.

3. The method for estimating the SOC of lithium batteries for new energy vehicles based on the RGC-PF algorithm according to claim 2, characterized in that, The lithium battery hybrid pulse power characteristic test method described in step S2 is used to conduct charge and discharge tests on lithium batteries with different states of charge to obtain response curves and calculate the parameters of the equivalent circuit model of lithium batteries under different states of charge. Specifically, constant current charge and discharge tests are conducted on lithium batteries with rated capacity of SOC of 1, 0.9...0.1 respectively, following the process of 1C discharge for 10s → rest for 30s → 1C charge for 10s → rest for 30s. The rated current curve under rated capacity is applied to the lithium battery cell to obtain the battery response curve. Based on the battery response curve, the open circuit voltage is calculated.

4. The method for estimating the SOC of lithium batteries for new energy vehicles based on the RGC-PF algorithm according to claim 3, characterized in that, Step S3, which involves measuring and recording the load current and the actual measured voltage across the lithium battery, specifically includes: measuring the load current across the lithium battery at this moment. At this moment, the actual measured voltage of the battery Record the load current across the battery terminals at the previous moment. The actual measured voltage of the battery at the previous moment. The battery open-circuit voltage at the previous moment The estimated ohmic internal resistance of the battery at the previous moment. Estimated value of battery electrochemical polarization resistance at the previous moment Estimated value of battery electrochemical polarization resistance at the previous moment .

5. The method for estimating the SOC of lithium batteries for new energy vehicles based on the RGC-PF algorithm according to claim 4, characterized in that, The method for constructing the criterion function described in step S4 is as follows: The actual measured voltage of the battery at this moment... With the current battery open-circuit voltage The difference As the output value of the criterion function, the load current across the battery at the previous moment. The load current across the battery at this moment The actual measured voltage of the battery at the previous moment. The battery open-circuit voltage at the previous moment The estimated ohmic internal resistance of the battery at this moment is used as the input value of the criterion function. Estimated value of battery electrochemical polarization resistance at this moment At this moment, the estimated value of the battery's electrochemical polarization capacitance The parameter estimate at this moment, used as the criterion function, is the battery ohmic internal resistance estimate from the previous moment. Estimated value of battery electrochemical polarization resistance at the previous moment The estimated value of the battery electrochemical polarization capacitance at the previous moment The criterion function is constructed using the parameter estimates from the previous time step.

6. The method for estimating the SOC of lithium batteries for new energy vehicles based on the RGC-PF algorithm according to claim 5, characterized in that, The formula for the criterion function mentioned in step S4 is as follows: In the formula: in, The output value of the criterion function at this moment The output value of the criterion function at this moment; The parameter estimate for this moment; The estimated ohmic internal resistance of the battery at this moment, The estimated value of the battery's electrochemical polarization resistance at this moment. This is the estimated value of the battery's electrochemical polarization capacitance at this moment.

7. The method for estimating the SOC of lithium batteries for new energy vehicles based on the RGC-PF algorithm according to claim 6, characterized in that, The parameter estimation recursive formula mentioned in step S4 is as follows: in, These are the parameter estimates from the previous time step. The parameters are set.

8. The method for estimating the SOC of lithium batteries for new energy vehicles based on the RGC-PF algorithm according to claim 7, characterized in that, Step S5 describes using the RGC algorithm to correct the parameter estimates of the criterion function according to the negative gradient direction until the criterion function reaches its minimum value, as detailed below: The parameter estimation recursive formula satisfies the following for parameter m: The recursive formula for parameter estimation is expressed as follows: Repeat the calculation of the parameter estimation recursive formula until the parameter identification is completed.

9. A method for estimating the SOC of lithium batteries for new energy vehicles based on the RGC-PF algorithm according to claim 8, characterized in that, The method for calculating the lithium battery's terminal voltage at the next moment based on the parameter estimates, and estimating the lithium battery's SOC using the PF algorithm, described in step S6, is as follows: 1) Initialize parameters: First, set the initial lithium battery SOC and variance R. ω and R ν Then, N initial random particles are generated, denoted as , i=1,2,3…N; 2) Use the RGC algorithm to identify ECM parameters, verify the accuracy of particle positions in the particle swarm optimization algorithm, and thus correct the positions of the particles. 3) Status update as follows: By substituting the optimized N particle states from the previous moment into the state equation and measurement equation of the lithium battery, the N particle states of the lithium battery and the voltage across the battery terminals at the next moment can be estimated. 4) Update measurement data The calculation error is: in, The estimated value of the i-th particle at time k is obtained by calculating the state equation and the measurement equation. The measured voltage value at time k; Calculate importance weights: When the deviation between the measured voltage value and the estimated value obtained from the state equation is small, the importance weight value of the particle is large; conversely, when the deviation is large, the importance weight value of the particle is low. Normalized importance weights: 5) Resampling If the following equation holds, then resampling will be performed. After resampling, the weights of particles near the true state will increase. In the following equation, N threshold The threshold for determining whether to resample; 6) Output State estimation: Particles with larger weights contribute more to the optimal value, while particles with smaller weights contribute less. As the number of iterations increases, the deviation between the optimal estimate and the actual value decreases continuously. Variance estimation: 7) Determine if the process has ended. If not, return to step 3). If it has ended, exit.

Images