Lithium iron phosphate battery charge state estimation method

Abstract

The invention discloses a lithium iron phosphate battery state-of-charge estimation method, which comprises the following steps: establishing a first-order RC equivalent circuit model of a lithium iron phosphate battery, performing pull-type transformation on the first-order RC equivalent circuit model to obtain a complex frequency domain equation, bilinearly changing the complex frequency domain equation to a z plane to obtain a system function expression, and estimating the state-of-charge of the lithium iron phosphate battery according to the system function expression. Performing pull-type inverse transformation on the system function expression to a discrete time domain; identifying the parameters of the lithium iron phosphate battery by using an AFFRLS algorithm to obtain model parameter values; constructing a space state equation set of the first-order RC equivalent circuit model, estimating the state of charge of the lithium iron phosphate battery by using an EKF algorithm, and defining an SOC estimation error function; optimizing the process noise covariance matrix Q and the measurement noise covariance matrix R of the EKF by using a simulated annealing algorithm to obtain a final estimated SOC; the SOC estimation precision and stability of the lithium iron phosphate battery are improved through the simulated annealing algorithm.

Description

Technical Field

[0001] The present invention belongs to the technical field of battery state of charge estimation methods, and relates to a method for estimating the state of charge of a lithium iron phosphate battery. Background Art

[0002] In recent years, energy storage systems have developed rapidly. As the core energy storage unit, the state estimation accuracy of lithium iron phosphate batteries directly affects the safety, lifespan, and energy efficiency management of the battery management system (BMS). Among them, the accurate estimation of the state of charge (SOC) is one of the key technologies of the battery management system. However, due to its unique voltage platform characteristics, the lithium iron phosphate battery poses challenges to the estimation of SOC. In the SOC range of 20% to 90%, the discharge voltage curve of the lithium iron phosphate battery is in the plateau region, and the voltage change amplitude does not exceed 50 mV. This characteristic causes the rate of change of voltage with respect to SOC change in the plateau region to approach zero, and the information content of the observation equation is almost invalid. Traditional voltage feedback-based estimation algorithms (such as the extended Kalman filter) are difficult to correct the SOC error through voltage measurement values, and the risk of model mismatch is exacerbated. Battery aging, temperature fluctuations, and operating conditions changes will significantly change the voltage characteristics in the plateau region. However, the parameters of the traditional equivalent circuit model (ECM) are fixed and it is difficult to dynamically adapt to such nonlinear changes. Coupled with the sharp decline of the Kalman gain of the EKF in the plateau period, the SOC estimation almost completely relies on the ampere-hour integration method. The tiny error of the current sensor may lead to a large SOC estimation deviation after long-term integration.

[0003] The fixed noise covariance parameters are usually set based on experience and cannot adapt to the estimation in the plateau period of lithium iron phosphate batteries, resulting in the filter relying too much on the ampere-hour integration of error accumulation, and the observed values cannot effectively participate in the correction. Moreover, the EKF approximates the nonlinear system through the first-order Taylor expansion, but the extremely low sensitivity of the OCV-SOC curve in the plateau region amplifies the linearization error and distorts the covariance propagation, further reducing the estimation robustness. In addition, the spatial dimensions of the process noise covariance matrix (Q) and the observation noise covariance matrix (R) are high. The manual trial-and-error method is time-consuming and difficult to find the global optimal solution. At the same time, gradient-based algorithms (such as the LM algorithm) rely on the differentiability of the objective function, while the relationship between the SOC estimation error and Q, R is highly nonlinear and there are multiple local optimal solutions, which are prone to falling into suboptimal parameter combinations. Heuristic algorithms (such as particle swarm optimization PSO) are prone to premature convergence, especially when the dynamic change of the objective function surface is caused by battery aging, and the generalization ability of the offline calibrated parameters is poor. The covariance matching method relies on residual statistics, but in the low sensitivity of the plateau period, the voltage measurement noise will be misjudged as SOC change, leading to over-adjustment of parameters and even filter divergence.

[0004] When the traditional Extended Kalman Filter (EKF) method is used to estimate the State of Charge (SOC) of a lithium iron phosphate battery, due to the long plateau period in the open circuit voltage (OCV) State of Charge curve, and the plateau region being approximately linear with a very small slope, the dynamic observation information is severely insufficient, resulting in a low SOC estimation accuracy. The maximum SOC error of existing methods in the lithium iron phosphate plateau region generally exceeds 5%, seriously restricting the implementation of high-precision battery management (such as balancing control and life prediction), and none of them effectively solve the contradiction between the lack of observation information in the plateau period and the global optimization of dynamic parameters, making it difficult to balance the SOC estimation accuracy and robustness. Therefore, there is an urgent need for a SOC estimation method with global optimization ability and engineering feasibility to break through the accuracy bottleneck in the plateau period. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for estimating the State of Charge of a lithium iron phosphate battery, which solves the problem in the prior art that the SOC estimation completely depends on the ampere-hour integration model and cannot accurately estimate the SOC.

[0006] The technical solution adopted by the present invention is a method for estimating the State of Charge of a lithium iron phosphate battery, including the following steps: Step 1: Establish a first-order RC equivalent circuit model of the lithium iron phosphate battery, perform a Laplace transform on the first-order RC equivalent circuit model to obtain a complex frequency domain equation, and bilinearly transform the complex frequency domain equation to the z plane to obtain a system function expression, and perform an inverse Laplace transform on the system function expression to the discrete time domain; Step 2: Use the AFFRLS algorithm to identify the parameters of the lithium iron phosphate battery to obtain the model parameter values; Step 3: Construct a state space equation set of the first-order RC equivalent circuit model, use the EKF algorithm to estimate the State of Charge of the lithium iron phosphate battery and define a SOC estimation error function; Step 4: Use the simulated annealing algorithm to optimize the process noise covariance matrix Q and the measurement noise covariance matrix R of the EKF to obtain the final estimated SOC.

[0007] The characteristics of the present invention also lie in: Step 1 is specifically implemented according to the following steps: Step 1.1: Establish a Thevenin equivalent circuit model of the lithium iron phosphate battery according to Kirchhoff's law: (1) (2) Among them, is the terminal voltage, in volts (V); OCV is the open circuit voltage, in volts (V); is the polarization voltage, in volts (V); i is the current, in amperes (A); is the ohmic internal resistance, with the unit of Ω; is the polarization resistance, with the unit of Ω; is the polarization capacitance, with the unit of F; d is the differential; t is the time, with the unit of s; Step 1.2: Perform the Laplace transform on Formula (1) and Formula (2) to obtain the complex frequency domain equation of the Thevenin equivalent circuit model: (3) where, is the terminal voltage of the complex frequency; OCV ( s ) is the open-circuit voltage of the complex frequency; is the polarization voltage of the complex frequency; i ( s ) is the current of the complex frequency; is the ohmic internal resistance of the complex frequency; s is the complex frequency; Bilinearly transform the complex frequency domain equation to the z plane: (4) z is z plane, Δ t is the discrete time domain step size; Obtain the system function expression: (5) where, G ( z ) is the system function expression, , b , c are meaningless; ; Step 1.3: Perform the inverse Laplace transform on the system function expression to the discrete time domain: (6) where, k represents the k th moment in the discrete time domain, represents the terminal voltage at the k th moment in the discrete time domain, represents the open-circuit voltage at the k th moment in the discrete time domain, , , respectively represent the k th moment in the discrete time domain of , b , c , represents thek The current at -1 moment.

[0008] The battery parameters of lithium iron phosphate in Step 2 are ohmic internal resistance R 0 , polarization resistance R p and polarization capacitance C p .

[0009] Step 2 is specifically implemented according to the following steps: Step 2.1, obtain the cell capacity of the lithium iron phosphate battery and the OCV - SOC correspondence relationship, and obtain the OCV - SOC relationship curve through sixth - order polynomial fitting: (7) Wherein, are polynomial coefficients; Step 2.2, establish the measurement matrix and the parameter matrix . The measurement matrix is: (8) The parameter matrix is: (9) Wherein, represents the open - circuit voltage at the k -1 moment in the discrete time domain, represents the current at the k th moment in the discrete time domain; Obtain: (10) Wherein, is the interference value; Use the AFFRLS algorithm to identify the model parameters: (11) (12) (13) Wherein, is the Kalman gain, is the error matrix, λ is the forgetting factor, I is the identity matrix.

[0010] Step 3 is specifically implemented according to the following steps: Step 3.1, construct the state - space equations of the first - order RC equivalent circuit model: (14) (15) (16) (17) (18) (19) (20) Among them, k -1 represents the k -1th moment, k +1 represents the k +1th moment, x k is the state vector to be estimated at the k th moment, is the system state matrix at the k th moment, is the input matrix at the k th moment, u k is the system input value at the k th moment, y k is the system output value at the k th moment, is the system output matrix at the k th moment, is the system feedforward matrix at the k th moment, w k is the zero-mean process noise of the process noise covariance matrix Q at the k th moment, v k is the zero-mean measurement noise of the measurement noise covariance matrix R at the k th moment, e is the natural constant, τk is the time constant, represents SOC the differential of, C N represents the battery capacity; η represents the coulomb efficiency, with a value of 1; Step 3.2, use EKF to estimate the SOC of the battery cell under Q and R: (21) (22) (23) (24) Among them, is the Kalman gain, For estimating the error covariance, is the estimation error; Step 3.3: Obtain the estimated SOC and the actual SOC of the lithium iron phosphate battery, and define the error function of SOC estimation: (25) where MSE is the mean square error, is the actual value of SOC, is the estimated value of SOC, n is the time window.

[0011] The estimated SOC is obtained by identifying the model parameters and the current value i and the terminal voltage to calculate the OCV, and then obtained by looking up the table through the SOC-OCV curve.

[0012] The actual SOC is the ratio of the cumulative power recorded by the power analyzer to the total power during the process of discharging the lithium iron phosphate battery from full charge to the cut-off voltage.

[0013] Step 4 is specifically implemented according to the following steps: Step 4.1: Calculate the probability of the simulated annealing algorithm accepting the new process noise covariance matrix and the new measurement noise covariance matrix : (26) where, is the acceptance probability, exp is the natural exponential function, is the estimation error within the window, is the minimum error within the loop, is the current temperature, in °C; Step 4.2: Generate a new process noise covariance matrix and a new measurement noise covariance matrix in the neighborhood: (27) where the coefficient β takes 0.1 to obtain the linear temperature decay function with time: (28) where α is the temperature decay coefficient, taking 0.95; Substitute and into the EKF to obtain the new state vector , which is the final estimated value of SOC.

[0014] Before optimizing the process noise covariance matrix Q and the measurement noise covariance matrix R of the EKF using the simulated annealing algorithm in step 4, set the initial temperature and the cooling rate of the simulated annealing algorithm according to the working characteristics of the lithium iron phosphate battery.

[0015] The beneficial effects of the present invention are as follows: The present invention synergistically optimizes the process noise covariance matrix Q and the measurement noise covariance matrix R of the EKF through the simulated annealing algorithm, significantly improving the SOC estimation accuracy and stability of the lithium iron phosphate battery; based on the OCV-SOC characteristic curve and the polarization dynamic characteristics of the lithium iron phosphate battery, an adaptive annealing strategy is designed, and rapid convergence of the parameter space is achieved through global search in the high-temperature stage and local fine-tuning in the low-temperature stage. The perturbation amplitude of Q is increased in the OCV platform area to strengthen model prediction, and the perturbation amplitude of R is increased in the high-slope area to enhance measurement update, thereby solving the problem of update failure caused by slow voltage change in the traditional EKF, ensuring that the true value of the SOC can still be accurately tracked when the battery voltage changes slowly, which is of great significance for improving the performance of the battery management system and extending the battery life. Description of the Drawings

[0016] Figure 1 is the flowchart of the present invention; Figure 2 is the SAEKF estimation result diagram when the SOC initial value and the true value of the present invention have no difference; Figure 3 is in the present invention Figure 2 enlarged view of part A; Figure 4 is the Figure 2 enlarged view of part B of the present invention; Figure 5 is the Figure 2 enlarged view of part C of the present invention; Figure 6 is the SAEKF estimation result error diagram when the SOC initial value and the true value of the present invention have no difference; Figure 7 is the SAEKF estimation result diagram when the difference between the SOC initial value and the true value of the present invention is 20%; Figure 8 is the SAEKF estimation result error diagram when the difference between the SOC initial value and the true value of the present invention is 20%. Detailed Embodiments

[0017] The present invention will be described in detail below in conjunction with the drawings and specific embodiments.

[0018] A method for estimating the state of charge of a lithium iron phosphate battery includes the following steps: Step 1: Establish the Thevenin equivalent circuit model (Davidson equivalent circuit model) of the lithium iron phosphate battery according to Kirchhoff's law: (1) (2) Wherein, is the terminal voltage, unit V; OCV is the open circuit voltage, unit V; is the polarization voltage, unit V; i is the current, unit A; is the ohmic internal resistance, unit Ω; is the polarization resistance, unit Ω; is the polarization capacitance, unit F; d is the differential; t is the time, unit s; Perform Laplace transform on formula (1) and formula (2) to obtain the complex frequency domain equation of the Thevenin equivalent circuit model: (3) Wherein, is the terminal voltage of the complex frequency; OCV ( s ) is the open circuit voltage of the complex frequency; is the polarization voltage of the complex frequency; i ( s ) is the current of the complex frequency; is the ohmic internal resistance of the complex frequency; s is the complex frequency; Bilinearly transform the complex frequency domain equation to the z plane: (4) z is the z plane, Δ t is the discrete time domain step size; Obtain the system function expression: (5) Wherein, G ( z ) is the system function expression, , b , c are meaningless; ; Perform inverse Laplace transform on the system function expression to the discrete time domain: (6) Wherein, k represents the k th moment in the discrete time domain, represents the terminal voltage at the k -th moment in the discrete time domain, represents the open-circuit voltage at the k -th moment in the discrete time domain, , , respectively represent the k -th moment in the discrete time domain of , b , c , represents the current at the k -1-th moment in the discrete time domain.

[0019] Step 2: Obtain the cell capacity and OCV-SOC correspondence of the lithium iron phosphate battery, and obtain the OCV-SOC relationship curve through sixth-order polynomial fitting: (7) where are the polynomial coefficients; Establish the measurement matrix and the parameter matrix . The measurement matrix is: (8) The parameter matrix is: (9) where represents the open-circuit voltage at the k -1-th moment in the discrete time domain, represents the current at the k -th moment in the discrete time domain; Obtain: (10) where is the interference value; Use the AFFRLS algorithm (Adaptive Forgetting Factor Recursive Least Squares) to identify the ohmic resistance R 0 , polarization resistance R p and polarization capacitance C p of the lithium iron phosphate battery: (11) (12) (13) where is the Kalman gain, is the error matrix, λ is the forgetting factor, I is the identity matrix.

[0020] Step 3: Construct the state - space equations of the first - order RC equivalent circuit model: (14) (15) (16) (17) (18) (19) (20) Wherein, k - 1 represents the k - 1th moment, k + 1 represents the k + 1th moment, x k is the state vector to be estimated at the k th moment, is the system state matrix at the k th moment, is the input matrix at the k th moment, u k is the system input value at the k th moment, y k is the system output value at the k th moment, is the system output matrix at the k th moment, is the system feed - forward matrix at the k th moment, w k is the zero - mean process noise of the process noise covariance matrix Q at the k th moment, v k is the zero - mean measurement noise of the measurement noise covariance matrix R at the k th moment, e is the natural constant, τk is the time constant, represents SOC the differential of, C N represents the battery capacity; η represents the coulomb efficiency, with a value of 1; Estimate the SOC of the battery cell using EKF under Q and R: (21) (22) (23) (24) Among them, is the Kalman gain, is the estimated error covariance, is the estimated error; Obtain the estimated SOC and the actual SOC of the lithium iron phosphate battery. The estimated SOC is obtained by identifying the model parameters, the current value i, and the terminal voltage to calculate the OCV, and then obtain it by looking up the table through the SOC-OCV curve; the actual SOC is the ratio of the cumulative power recorded by the power analyzer to the total power during the process of discharging the lithium iron phosphate battery from full charge to the cut-off voltage. Define the error function of SOC estimation: (25) Among them, MSE is the mean square error, is the actual value of SOC, is the estimated value of SOC, n is the time window.

[0021] Step 4: Set the initial temperature and the cooling rate of the simulated annealing algorithm according to the working characteristics of the lithium iron phosphate battery, and calculate the probability of the simulated annealing algorithm receiving the new process noise covariance matrix and the new measurement noise covariance matrix : (26) Among them, is the acceptance probability, exp is the natural exponential function, is the estimated error within the window, is the minimum error within the loop, is the current temperature, in °C; Generate a new process noise covariance matrix and a new measurement noise covariance matrix in the neighborhood: (27) Among them, the coefficient β takes 0.1 to obtain the linear decline function of temperature with time: (28) Among them, α is the temperature decay coefficient, with a value of 0.95; Substitute and into the EKF to obtain the new state vector , which is the final estimated value of SOC.

[0022] The state - space equations of the first - order RC equivalent circuit model serve as the state equation and the observation equation of the EKF, which are used to describe the voltage and current dynamic characteristics of the battery respectively.

[0023] Refer to Figure 2 、 Figure 3 、 Figure 4 、 Figure 5 、 Figure 6 , when the initial value of the estimated SOC is set to the true value, both the traditional EKF and the SAEKF (EKF using the simulated annealing algorithm) can converge, but the SAEKF has higher accuracy.

[0024] Refer to Figure 7 and Figure 8 , when there is a 20% deviation between the initial value of the estimated SOC and the true value, the traditional EKF cannot converge to the true value, but the SAEKF will gradually converge to the true value.

[0025] Example 1: A method for estimating the state of charge of a lithium iron phosphate battery, comprising the following steps: Step 1: Establish a Thevenin equivalent circuit model of the lithium iron phosphate battery according to Kirchhoff's law: (1) (2) Among them, is the terminal voltage, unit V; OCV is the open - circuit voltage, unit V; is the polarization voltage, unit V; i is the current, unit A; is the ohmic internal resistance, unit Ω; is the polarization resistance, unit Ω; is the polarization capacitance, unit F; d is the differential; t is the time, unit s; Perform Laplace transform on formula (1) and formula (2) to obtain the complex - frequency - domain equation of the Thevenin equivalent circuit model: (3) Among them, is the terminal voltage of the complex frequency; OCV ( s ) is the open - circuit voltage of the complex frequency; is the polarization voltage of the complex frequency; i ( s ) is the current of the complex frequency; is the ohmic internal resistance of the complex frequency; s is the complex frequency; Bilinearly transform the complex - frequency - domain equation to the z plane: (4) z is z a plane, Δ t is the discrete time domain step size; Obtain the system function expression: (5) Wherein, G ( z ) is the system function expression, , b , c is meaningless; ; Perform the inverse Laplace transform of the system function expression to the discrete time domain: (6) Wherein, k represents the k -th moment in the discrete time domain, represents the terminal voltage at the k -th moment in the discrete time domain, represents the open - circuit voltage at the k -th moment in the discrete time domain, , , respectively represent the k -th moment in the discrete time domain of , b , c , represents the current at the k - 1 - th moment in the discrete time domain; Step 2: Use the AFFRLS algorithm to identify the parameters of the lithium iron phosphate battery to obtain the model parameter values; Step 3: Construct the state - space equations of the first - order RC equivalent circuit model, and use the EKF algorithm to estimate the state of charge of the lithium iron phosphate battery and define the SOC estimation error function; Step 4: Use the simulated annealing algorithm to optimize the process noise covariance matrix Q and the measurement noise covariance matrix R of the EKF to obtain the final estimated SOC.

[0026] Embodiment 2: A method for estimating the state of charge of a lithium iron phosphate battery, comprising the following steps: Step 1: Establish a first - order RC equivalent circuit model of the lithium iron phosphate battery, perform the Laplace transform on the first - order RC equivalent circuit model to obtain the complex - frequency - domain equation, and bilinearly transform the complex - frequency - domain equation to z the plane to obtain the system function expression, and perform the inverse Laplace transform of the system function expression to the discrete time domain; Step 2: Obtain the cell capacity and the OCV-SOC correspondence of the lithium iron phosphate battery, and obtain the OCV-SOC relationship curve through sixth-order polynomial fitting: (7) wherein, are polynomial coefficients; Establish the measurement matrix and the parameter matrix , and the measurement matrix is: (8) The parameter matrix is: (9) wherein, represents the open-circuit voltage at the k -1th moment in the discrete time domain, represents the current at the k th moment in the discrete time domain; Obtain: (10) wherein, is the interference value; Use the AFFRLS algorithm to identify the ohmic internal resistance R 0 , polarization resistance R p and polarization capacitance C p of the lithium iron phosphate battery: (11) (12) (13) wherein, is the Kalman gain, is the error matrix, λ is the forgetting factor, I is the identity matrix; Step 3: Construct the state-space equations of the first-order RC equivalent circuit model, and use the EKF algorithm to estimate the state of charge of the lithium iron phosphate battery and define the SOC estimation error function; Step 4: Use the simulated annealing algorithm to optimize the process noise covariance matrix Q and the measurement noise covariance matrix R of the EKF to obtain the final estimated SOC.

[0027] Example 3: A method for estimating the state of charge of a lithium iron phosphate battery, comprising the following steps: Step 1: Establish a first-order RC equivalent circuit model of the lithium iron phosphate battery, perform a Laplace transform on the first-order RC equivalent circuit model to obtain a complex frequency domain equation, and bilinearly transform the complex frequency domain equation to zIn the s-plane, the system function expression is obtained, and the system function expression is inverse Laplace-transformed to the discrete time domain; Step 2: Use the AFFRLS algorithm to identify the parameters of the lithium iron phosphate battery to obtain the model parameter values; Step 3: Construct the state-space equations of the first-order RC equivalent circuit model: (14) (15) (16) (17) (18) (19) (20) where, k -1 represents the k -1th moment, k +1 represents the k +1th moment, x k is the state vector to be estimated at the k th moment, is the system state matrix at the k th moment, is the input matrix at the k th moment, u k is the system input value at the k th moment, y k is the system output value at the k th moment, is the system output matrix at the k th moment, is the system feedforward matrix at the k th moment, w k is the zero-mean process noise of the process noise covariance matrix Q at the k th moment, v k is the zero-mean measurement noise of the measurement noise covariance matrix R at the k th moment, e is the natural constant, τk is the time constant, denotes SOC the differential of, C N denotes the battery capacity; η denotes the coulombic efficiency, with a value of 1; Using the EKF to estimate the SOC of the battery cell under Q and R: (21) (22) (23) (24) where is the Kalman gain, is the estimated error covariance, is the estimation error; Obtain the estimated SOC and the actual SOC of the lithium iron phosphate battery. The estimated SOC is obtained by identifying the model parameters, the current value i, and the terminal voltage to calculate the OCV, and then look up the table through the SOC-OCV curve; the actual SOC is the ratio of the cumulative charge recorded by the power analyzer to the total charge during the discharge of the lithium iron phosphate battery from full charge to the cut-off voltage. Define the error function of the SOC estimation: (25) where MSE is the mean square error, is the actual value of SOC, is the estimated value of SOC, n is the time window; Step 4: Use the simulated annealing algorithm to optimize the process noise covariance matrix Q and the measurement noise covariance matrix R of the EKF to obtain the final estimated SOC.

[0028] Example 4: A method for estimating the state of charge of a lithium iron phosphate battery, comprising the following steps: Step 1: Establish a first-order RC equivalent circuit model of the lithium iron phosphate battery, perform a Laplace transform on the first-order RC equivalent circuit model to obtain a complex frequency domain equation, and bilinearly transform the complex frequency domain equation to z the plane to obtain a system function expression, and perform an inverse Laplace transform on the system function expression to the discrete time domain; Step 2: Use the AFFRLS algorithm to identify the parameters of the lithium iron phosphate battery to obtain the model parameter values; Step 3: Construct a state space equation set of the first-order RC equivalent circuit model, use the EKF algorithm to estimate the state of charge of the lithium iron phosphate battery and define the SOC estimation error function; Step 4: Set the initial temperature and the cooling rate of the simulated annealing algorithm according to the working characteristics of the lithium iron phosphate battery, and calculate the probability that the simulated annealing algorithm accepts the new process noise covariance matrix and the new measurement noise covariance matrix : (26) wherein, is the reception probability, exp is the natural exponential function, is the estimated error within the window, is the minimum value of the error within the loop, is the current temperature, in °C; Generate a new process noise covariance matrix and a new measurement noise covariance matrix : (27) wherein, the coefficient β takes 0.1 to obtain a linear temperature decline function over time: (28) wherein, α is the temperature decay coefficient, with a value of 0.95; Substitute and into the EKF to obtain a new state vector , which is the final SOC estimated value.

[0029] Example 5: A method for estimating the state of charge of a lithium iron phosphate battery, comprising the following steps: Step 1: Establish a Thevenin equivalent circuit model of the lithium iron phosphate battery according to Kirchhoff's law: (1) (2) wherein, is the terminal voltage, in V; OCV is the open-circuit voltage, in V; is the polarization voltage, in V; i is the current, in A; is the ohmic internal resistance, in Ω; is the polarization resistance, in Ω; is the polarization capacitance, in F; d is the differential; t is the time, in s; Perform Laplace transform on formula (1) and formula (2) to obtain the complex frequency domain equation of the Thevenin equivalent circuit model: (3) wherein, is the terminal voltage at complex frequency; OCV ( s ) is the open-circuit voltage at complex frequency; is the polarization voltage at complex frequency; i ( s ) is the current at complex frequency; is the ohmic internal resistance at complex frequency; s is the complex frequency; Bilinearly transform the complex frequency domain equation to z plane: (4) z is z plane, Δ t is the discrete time domain step size; Obtain the system function expression: (5) where, G ( z ) is the system function expression, , b , c are meaningless; ; Take the inverse Laplace transform of the system function expression to the discrete time domain: (6) where, k represents the k th moment in the discrete time domain, represents the terminal voltage at the k th moment in the discrete time domain, represents the open circuit voltage at the k th moment in the discrete time domain, , , respectively represent the k th moment in the discrete time domain of , b , c , represents the current at the k -1th moment in the discrete time domain; Step 2: Obtain the cell capacity and OCV-SOC correspondence of the lithium iron phosphate battery, and obtain the OCV-SOC relationship curve by fitting with a sixth-order polynomial: (7) where, are the polynomial coefficients; Establish the measurement matrix and the parameter matrix , the measurement matrix is: (8) The parameter matrix is: (9) where, represents thek The open-circuit voltage at -1 moment indicating the current at the k th moment in the discrete time domain; Obtain: (10) wherein, is the interference value; Using the AFFRLS algorithm to identify the ohmic internal resistance R 0 , polarization resistance R p and polarization capacitance C p of the lithium iron phosphate battery: (11) (12) (13) wherein, is the Kalman gain, is the error matrix, λ is the forgetting factor, I is the identity matrix; Step 3: Construct the state-space equations of the first-order RC equivalent circuit model, and use the EKF algorithm to estimate the state of charge of the lithium iron phosphate battery and define the SOC estimation error function; Step 4: Use the simulated annealing algorithm to optimize the process noise covariance matrix Q and measurement noise covariance matrix R of the EKF to obtain the final estimated SOC.

[0030] Example 6: A method for estimating the state of charge of a lithium iron phosphate battery, comprising the following steps: Step 1: Establish a first-order RC equivalent circuit model of the lithium iron phosphate battery, perform a Laplace transform on the first-order RC equivalent circuit model to obtain a complex frequency domain equation, and bilinearly transform the complex frequency domain equation to the z plane to obtain a system function expression, and perform an inverse Laplace transform on the system function expression to the discrete time domain; Step 2: Use the AFFRLS algorithm to identify the parameters of the lithium iron phosphate battery to obtain the model parameter values; Step 3: Construct the state-space equations of the first-order RC equivalent circuit model: (14) (15) (16) (17) (18) (19) (20) Wherein, k -1 represents the k -1th moment, k +1 represents the k +1th moment, x k is the state vector to be estimated at the k th moment, is the system state matrix at the k th moment, is the input matrix at the k th moment, u k is the system input value at the k th moment, y k is the system output value at the k th moment, is the system output matrix at the k th moment, is the system feedforward matrix at the k th moment, w k is the zero-mean process noise of the process noise covariance matrix Q at the k th moment, v k is the zero-mean measurement noise of the measurement noise covariance matrix R at the k th moment, e is the natural constant, τk is the time constant, represents SOC differential, C N represents the battery capacity; η represents the coulombic efficiency, with a value of 1; Estimate the SOC of the battery cell using EKF under Q and R: (21) (22) (23) (24) Wherein, is the Kalman gain, is the estimation error covariance, is the estimation error; Obtain the estimated SOC and actual SOC of the lithium iron phosphate battery. The estimated SOC is obtained through the identified model parameters and the current value i, terminal voltage The OCV is calculated and then obtained by looking up the table in the SOC-OCV curve; the actual SOC is the ratio of the cumulative power recorded by the power analyzer during the process of discharging the lithium iron phosphate battery from full charge to the cut-off voltage to the total power. Define the error function of SOC estimation: (25) where, MSE is the mean square error, is the actual value of SOC, is the estimated value of SOC, n is the time window; Step 4: Set the initial temperature and the cooling rate of the simulated annealing algorithm according to the working characteristics of the lithium iron phosphate battery, and calculate the probability of the simulated annealing algorithm receiving the new process noise covariance matrix and the new measurement noise covariance matrix : (26) where, is the acceptance probability, exp is the natural exponential function, is the estimation error within the window, is the minimum error within the loop, is the current temperature, in °C; Generate a new process noise covariance matrix and a new measurement noise covariance matrix in the neighborhood: (27) where, the coefficient β takes 0.1 to obtain the linear temperature decay function with time: (28) where, α is the temperature decay coefficient, with a value of 0.95; Substitute and into the EKF to obtain the new state vector , which is the final estimated value of SOC.

Claims

1. A method for estimating the state of charge of a lithium iron phosphate battery, characterized in that: The following steps are involved: Step 1: Establish the first-order RC equivalent circuit model of lithium iron phosphate battery, perform Latent transformation on the first-order RC equivalent circuit model to obtain the complex frequency domain equation, and transform the complex frequency domain equation bilinearly to z plane, obtain the system function expression, and perform a Latent inverse transform on the system function expression to the discrete time domain; Step 2: Use the AFFRLS algorithm to identify the parameters of the lithium iron phosphate battery and obtain the model parameter values; Step 3: Construct a spatial state equation group of the first-order RC equivalent circuit model, use the EKF algorithm to estimate the state of charge of the lithium iron phosphate battery and define the SOC estimation error function; Step 4: Use the simulated annealing algorithm to optimize the process noise covariance matrix Q and the measurement noise covariance matrix R of the EKF to obtain the final estimated SOC.

2. The method for estimating the state of charge of a lithium iron phosphate battery according to claim 1, characterized in that: The step 1 is specifically implemented according to the following steps: Step 1.1: Establish the Thevenin equivalent circuit model of lithium iron phosphate battery according to the Kirchhall law: （1） （2） in, is the terminal voltage, unit V; OCV is the open circuit voltage, unit V; is the polarization voltage, unit V; i is the current, unit is A; is the ohmic internal resistance, unit Ω; is the polarization resistance, unit Ω; is the polarization capacitance, unit is F; d is the differential; t is time, unit is s; Step 1.2, perform a Lasker transformation on formula (1) and formula (2) to obtain the complex frequency domain equation of the Thevenin equivalent circuit model: （3） in, is the terminal voltage of the complex frequency; OCV ( s ) is the open circuit voltage at complex frequency; is the polarization voltage of the complex frequency; i ( s ) is the current of complex frequency; is the ohmic internal resistance at complex frequency; s is the complex frequency; Transform the complex frequency domain equation bilinearly to z flat: （4） z for z Plane, Δ t is the discrete time domain step size; Get the system function expression: （5） in, G ( z ) is the system function expression, , b , c meaningless; ; Step 1.3, perform an inverse transformation of the system function expression to the discrete time domain: （6） in, k Represents the discrete time domain k a moment, Represents the discrete time domain k The terminal voltage at a certain moment, Represents the discrete time domain k The open circuit voltage at a given moment, , , They represent the discrete time domain k of the moment , b , c , Represents the discrete time domain k -1 moment of current.

3. The method for estimating the state of charge of a lithium iron phosphate battery according to claim 2, characterized in that: The parameter of the lithium iron phosphate battery in step 2 is the ohmic internal resistance R 0. Polarization resistance R p and polarization capacitance C p .

4. The method for estimating the state of charge of a lithium iron phosphate battery according to claim 3, characterized in that: The step 2 is specifically implemented according to the following steps: Step 2.1, obtain the corresponding relationship between the cell capacity and OCV-SOC of the lithium iron phosphate battery, and obtain the OCV-SOC relationship curve through sixth-order polynomial fitting: （7） in, are the polynomial coefficients; Step 2.2: Establish the measurement matrix and the parameter matrix , the measurement matrix is: （8） The parameter matrix is: （9） in, Represents the discrete time domain k -1 moment open circuit voltage, Represents the discrete time domain k The current at a moment; get: （10） in, is the interference value; Using AFFRLS algorithm to identify model parameters: （11） （12） （13） in, is the Kalman gain, is the error matrix, λ For the forgetting factor, I is the identity matrix.

5. The method for estimating the state of charge of a lithium iron phosphate battery according to claim 4, characterized in that: The step 3 is specifically implemented according to the following steps: Step 3.1, construct the spatial state equation group of the first-order RC equivalent circuit model: （14） （15） （16） （17） （18） （19） （20） in, k -1 means k -1 moment, k +1 means k +1 moment, x k For the k The state vector to be estimated at the moment, For the k The system state matrix at each moment is: For the k The input matrix at each moment is u k For the k The system input value at a certain moment, y k For the k The system output value at a moment, For the k The system output matrix at each moment is: For the k The system feedforward matrix at the moment is: w k For the k The zero-mean process noise is the process noise covariance matrix Q at the moment, v k For the k The zero-mean measurement noise of the measurement noise covariance matrix R at time instants, e is a natural constant, τk is the time constant, express SOC The differential of C N Indicates battery capacity; η Represents Coulomb efficiency, with a value of 1; Step 3.2, use EKF to estimate the SOC of the battery cell under Q and R: （21） （22） （23） （24） in, is the Kalman gain, is the estimated error covariance, is the estimation error; Step 3.3, obtain the estimated SOC and actual SOC of the lithium iron phosphate battery, and define the error function of SOC estimation: （25） Among them, MSE is the mean square error, is the actual value of SOC, is the estimated SOC value, n is the time window.

6. The method for estimating the state of charge of a lithium iron phosphate battery according to claim 5, characterized in that: The estimated SOC is obtained by identifying the model parameters and current value i , terminal voltage Calculate the OCV and then obtain it by looking up the SOC-OCV curve.

7. The method for estimating the state of charge of a lithium iron phosphate battery according to claim 5, characterized in that: The actual SOC is the ratio of the cumulative power recorded by the power analyzer during the process of the lithium iron phosphate battery being discharged from full charge to the cut-off voltage to the total power.

8. The method for estimating the state of charge of a lithium iron phosphate battery according to any one of claims 5 to 7, characterized in that: The step 4 is specifically implemented according to the following steps: Step 4.1, calculate the new process noise covariance matrix received by the simulated annealing algorithm and the new measurement noise covariance matrix Probability: （26） in, is the acceptance probability, exp is the natural exponential function, is the estimation error within the window, is the minimum error within the cycle, is the current temperature, in °C; Step 4.2: Generate a new process noise covariance matrix in the neighborhood and the new measurement noise covariance matrix : （27） Among them, the coefficient β Taking 0.1, we get the linear decrease function of temperature over time: （28） Among them, α is the temperature attenuation coefficient, which is 0.95; Will and Substitute into EKF and get the new state vector , which is the final SOC estimation value.

9. The method for estimating the state of charge of a lithium iron phosphate battery according to any one of claims 1 to 7, characterized in that: In step 4, before optimizing the process noise covariance matrix Q and the measurement noise covariance matrix R of the EKF using the simulated annealing algorithm, the initial temperature and the cooling rate of the simulated annealing algorithm are set according to the working characteristics of the lithium iron phosphate battery.

Images