Lithium ion battery state estimation method based on advanced mobile vision field

Abstract

The invention provides a lithium ion battery state estimation method based on an advanced mobile vision field. The method comprises the following steps: establishing an equivalent circuit model: establishing a lithium ion battery equivalent circuit model containing a hysteresis effect and temperature related parameters; constructing a nonlinear state space model based on the equivalent circuit model, wherein the state space model comprises a state equation for describing dynamic behaviors of the system and an observation equation for reflecting voltage response; an event trigger linearization mechanism (ETR) is introduced into a mobile vision field estimation (MHE) optimization framework; dynamically adjusting the weight of the cost function by combining a Gaussian-Newton iteration (GN) optimization method; and a high-precision SOC estimation result is output by using real-time measurement data through complete modeling, prediction and optimization processes. On the basis that the advantages of an MHE optimization structure are reserved, battery physical nonlinear factors are further fused, and the SOC estimation method with temperature adaptability solves the problem of estimation deviation caused by the fact that a traditional MHE simplified model ignores physical characteristics.

Description

Technical Field

[0001] This invention relates to the field of battery state assessment, and in particular to a method for estimating the state of lithium-ion batteries based on advanced mobile horizons. Background Technology

[0002] As the core of energy storage systems, lithium-ion batteries rely heavily on their state of charge (SOC), a key parameter that measures the ratio of current usable capacity to rated capacity. Existing SOC estimation methods mainly include the open-circuit voltage method, the ampere-hour integration method, and data-driven methods. Advanced moving horizon estimation (MHE) methods have received widespread attention in recent years due to their ability to handle system nonlinearities and constraints.

[0003] However, existing advanced moving horizon estimation (MHE) methods typically employ simplified models during the modeling phase. These simplified models often neglect the inherent hysteresis characteristics of the battery, leading to an inaccurate description of the battery's dynamic behavior. Furthermore, such models do not adequately consider the significant impact of temperature on the battery's electrochemical properties. Temperature changes alter key parameters such as internal resistance, open-circuit voltage, and chemical reaction rates. Under drastic temperature fluctuations, the battery's charge and discharge performance differs significantly from that at room temperature. When the battery is exposed to environments with drastic temperature changes, such as an electric vehicle moving from a warm indoor environment to a cold outdoor environment in winter, the model cannot accurately reflect the temperature-induced parameter changes, resulting in a sharp drop in SOC estimation accuracy. In scenarios with rapidly changing loads, such as the rapid acceleration and deceleration of an electric vehicle, simplified models cannot track the rapid changes in battery state in a timely and accurate manner, often exhibiting slow convergence and causing the SOC estimate to remain unstable near the true value for an extended period.

[0004] Existing advanced moving horizon estimation (MHE) methods do not consider the inherent hysteresis characteristics of batteries and the influence of temperature on electrochemical characteristics, resulting in poor SOC estimation accuracy in actual operation, especially in scenarios with drastic temperature changes or drastic load dynamics, where slow convergence speed and large estimation errors are likely to occur. Summary of the Invention

[0005] This invention provides a lithium-ion battery state estimation method based on advanced moving horizon (MHE), which solves the problem that existing advanced moving horizon estimation (MHE) methods do not consider the battery's inherent hysteresis characteristics and the influence of temperature on electrochemical characteristics, resulting in poor SOC estimation accuracy in actual operation, especially in scenarios with drastic temperature changes or drastic load dynamics, where slow convergence speed and large estimation errors are prone to occur.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a lithium-ion battery state estimation method based on advanced mobile horizons, comprising the following steps:

[0007] S1. Establish an equivalent circuit model: Establish an equivalent circuit model of a lithium-ion battery that includes hysteresis effect and temperature-related parameters. The equivalent circuit model includes open-circuit voltage, charge transfer resistance, capacitor capacitance, and hysteresis voltage. Hysteresis effect and temperature-related parameters are all functions of temperature and state of charge (SOC).

[0008] S2. Based on the equivalent circuit model, construct a nonlinear state-space model: The state-space model includes state equations describing the dynamic behavior of the system and observation equations reflecting the voltage response;

[0009] S3. The Moving Horizon Estimation (MHE) optimization framework introduces an Event Triggered Linearization (ETR) mechanism. By determining whether the degree of nonlinear change in the current system exceeds a preset threshold, linearization is only performed when the threshold is exceeded, thereby reducing the computational burden.

[0010] S4. Combining the Gauss-Newton (GN) iteration optimization method, a forward-backward recursive process is executed within each sliding window, and temperature factors and hysteresis states are embedded in the optimization process to dynamically adjust the cost function weights.

[0011] S5. Through a complete modeling, prediction and optimization process, and utilizing real-time measurement data, it outputs high-precision SOC estimation results.

[0012] In the preferred embodiment, in S1, the equivalent circuit model of the lithium-ion battery is a first-order equivalent circuit model, which includes a voltage source, series resistor, parallel resistor and capacitor structure.

[0013] The equivalent circuit model is described by the following mathematical expression:

[0014]

[0015] In the formula, U OCV (SOC,T) represents the relationship between the state of charge (SOC), temperature, and open-circuit voltage, derived through fitting experimental data; U1 is the voltage across the R1C1 branch, representing the diffusion voltage; U b R1 represents the battery terminal voltage; R1 and C1 are the parallel branch resistors and capacitors, and R2 represents the battery impedance; the block marked hyst represents the hysteresis effect in the circuit model; Uhyst represents the hysteresis voltage; I is the battery current; and t is the sampling time.

[0016] In the preferred embodiment, the dynamic characteristics of the equivalent circuit model can be described in matrix-vector form, as follows:

[0017]

[0018] In the formula, η k η is the coulombic efficiency of the battery.k Between 0 and 1; γ is a constant that adjusts the hysteresis voltage decay rate; Q bat This represents the total battery capacity.

[0019] In the preferred embodiment, in S2, the state-space model uses state variables to characterize the internal dynamic behavior of the battery. These state variables include SOC, polarization voltage, and hysteresis state, which together constitute the state vector x. k x k The formula is defined as follows:

[0020]

[0021] In the formula, Z k Represents SOC, V 1,k H represents the polarization voltage. k This represents a sluggish state.

[0022] In the preferred embodiment, in S2, the input variables of the state-space model include the battery's charging and discharging current and the current temperature, and the input vector u k The expression is as follows:

[0023]

[0024] In the formula, I k T represents the charging and discharging current of the battery at time k (discharging is positive, charging is negative); k This is the current temperature of the battery.

[0025] In the preferred embodiment, in S2, the state equation describes the dynamic evolution of the state variables over time through a temperature-dependent nonlinear transition relationship, as shown in the formula:

[0026] x k+1 =A k (T k )x k +B k (T k )u k +w k (5);

[0027] In the formula, the state transition matrix A k (T k )for:

[0028]

[0029] Input influence matrix B k (T k The formula representing the influence of input variables on state variables is:

[0030]

[0031] In the formula, Cn is the rated capacity of the battery; τ 1,k =R 1,k C 1,k R is the polarization time constant; 1,k and C 1,k Temperature-corrected polarization resistance and capacitance; Δt is the sampling time interval; f T (T k ) is the temperature correction factor, used to describe the effect of temperature on battery parameters; process noise w k Follows a Gaussian distribution: w k ~N(0,Q) k (T k ));

[0032] After temperature correction, the process noise covariance matrix Q k (T k The formula is:

[0033]

[0034] In the formula, The process noise covariances are SOC, polarization voltage, and hysteresis state, respectively; η T T is the temperature decay coefficient; ref This is a reference temperature, taken as 25℃.

[0035] In the preferred embodiment, in S2, the relationship between the observed equation of the voltage response and the measured voltage is as follows:

[0036] y(k)=U OCV (SOC,T)+U hyst (k)-I(k)·R2-U1(k)+v k (9);

[0037] In the formula, y(k) is the battery voltage, v k To observe noise.

[0038] In the preferred scheme, in S3, the steps of introducing an event-triggered linearization (ETR) mechanism into the Moving Horizon Estimation (MHE) optimization framework to reduce the computational complexity of nonlinear solutions include:

[0039] S3.1, Define state increment change: In each iteration, calculate the increment between the current state vector and the previous state vector, and use the L2 norm as the criterion for state change. The formula is defined as follows:

[0040] ||Δx k ||2=||x k -x k-1 ||2 (10);

[0041] In the formula, Δx k It is the state increment at the current moment, x k-1 It is the state vector from the previous moment;

[0042] S3.2 Setting the temperature correction trigger threshold: The threshold calculation formula is as follows:

[0043] ε T =ε0+κ T |T k -T ref | (11);

[0044] In the formula, ε0 is the reference trigger threshold, representing the triggering condition at standard temperature; κ T It is the temperature sensitivity coefficient, reflecting the effect of temperature changes on the trigger threshold; T k It is the current temperature, ε T The temperature-adaptive trigger threshold;

[0045] S3.3 Triggering Conditions and Linearization Update: During linearization, the Jacobian between the state transition matrix and the observation matrix is ​​recalculated and corrected based on the current temperature. The system triggers the linearization update operation when the state increment satisfies the following condition (12):

[0046] ||Δx k ||2+γ H |H k -H k-1 |>ε T (12);

[0047] In the formula, γ H H is the weighting coefficient for the hysteresis effect. k The observation matrix;

[0048] S3.4, State Covariance Matrix Update: Combine the hysteresis weight term with the observation matrix Hk to adjust the state covariance matrix P in real time. k (T k To improve the stability and accuracy of state estimation, the updated formula is as follows:

[0049]

[0050] Through the above mechanism, the event-triggered linearization method in S3 can significantly reduce the frequency of unnecessary Hessian calculations while maintaining estimation accuracy, thereby improving the real-time performance and temperature adaptability of the entire estimation algorithm.

[0051] In the preferred embodiment, in step S4, the forward-backward recursive GN iterative optimization method, by combining the influence of temperature factor and hysteresis effect on state estimation, improves estimation accuracy and convergence efficiency. Step S4 includes the following steps:

[0052] S4.1 Temperature-corrected state increment estimation during the forward recursion process. Within the moving horizon estimation framework, the GN iterative algorithm is used for incremental updates to solve the optimization problem based on the nonlinear state-space model. During the forward recursion stage, the current state increment is estimated through the combined effect of the residual, weighting matrix, and state Jacobian matrix. Simultaneously, considering the influence of temperature factors on battery dynamic performance, a temperature-related measurement noise covariance matrix R is constructed. k (T k ), used to correct the state increment, the calculation formula is:

[0053]

[0054] In the formula, Δx k R is the increment at the current moment. k (T k ) is the temperature-dependent measurement noise covariance matrix;

[0055] S4.2, State Increment Correction in the Backward Recursive Process: The state increment calculated in the forward process is further corrected through the backward recursive structure. The correction mechanism considers the historical influence of hysteresis state variables and combines the backward update matrix F constructed with the temperature factor. k (T k This is to improve the battery's responsiveness to different operating conditions. The state correction process uses the recursive relationship of formula (16):

[0056] Δx′ k =F k (T k )Δx k+1 (16);

[0057] In the formula, Δx′ k The corrected increment, Δx k+1 F is the state increment correction term fed back from the next time step during the backward recursion process. k (T k ) is the backward recursive update matrix;

[0058] S4.3 Temperature-Adaptive Convergence Criterion: The convergence threshold is dynamically adjusted according to the current temperature, and the formula is defined as follows:

[0059] ||y k -U b ||2<δ(T k (17);

[0060] δ(T k )=δ0+γ T |T k -T ref| (18);

[0061] In the formula, δ(T) k ) is the convergence threshold; δ0 is the baseline convergence threshold; γ T It is the effect of temperature change on convergence.

[0062] The beneficial effects of this invention are as follows: While retaining the advantages of the MHE optimized structure, this invention further integrates battery physical nonlinear factors and possesses a temperature-adaptive SOC estimation method. It continues the ability of Moving Horizon Estimation (MHE) to handle system nonlinearity and constraints, ensuring global optimization characteristics for battery dynamic behavior. The systematic inclusion of battery hysteresis and the influence of temperature on electrochemical parameters (open-circuit voltage, internal resistance, capacitance, etc.) within the MHE framework solves the estimation bias problem caused by neglecting physical characteristics in traditional simplified MHE models, making the model closer to the actual operating state of the battery.

[0063] This invention optimizes battery SOC estimation by dynamically determining the incremental changes of the ETR strategy within a framework and incorporating temperature factors. The GN iteration further enhances the model's adaptability to hysteresis effects and temperature variations through the introduction of forward-backward recursive optimization. In each iteration, key parameters such as the covariance matrix and weighting matrix are progressively updated to ensure the accuracy and robustness of battery state estimation under different temperature conditions.

[0064] This invention introduces an event-triggered linearization mechanism into the MHE optimization framework to dynamically determine the degree of nonlinearity in the system. Linearization is only performed when the nonlinearity exceeds a threshold, avoiding the computational redundancy caused by frequent linearization in traditional MHE. At the same time, it ensures the estimation accuracy in scenarios with severe nonlinearity. While maintaining the estimation accuracy, it significantly reduces the frequency of unnecessary Hessian calculations, thereby improving the real-time performance and temperature adaptability of the entire estimation algorithm.

[0065] This invention combines the Gauss-Newton (GN) iteration optimization method with a recursive process of forward prediction and backward correction. It dynamically adjusts key parameters such as the covariance matrix and weighting matrix based on hysteresis and temperature factors, significantly improving the algorithm's adaptability to hysteresis effects and temperature fluctuations, enabling stable convergence even under complex conditions. The GN iteration avoids the complex calculation of higher-order derivatives, making the algorithm easy to implement in embedded systems, highly scalable, and meeting real-time requirements, thus possessing significant potential for widespread application. Attached Figure Description

[0066] The present invention will be further described below with reference to the accompanying drawings and embodiments;

[0067] Figure 1 This is the equivalent circuit model diagram of the present invention;

[0068] Figure 2This is a schematic diagram of the state-space model structure of the present invention;

[0069] Figure 3 This is a flowchart illustrating the specific implementation of the ETR strategy of the present invention in the GN iteration;

[0070] Figure 4 The figures show the simulation results of the present invention under various working conditions at 0℃.

[0071] Figure 5 The figures show the simulation results of the present invention under various working conditions at 25℃.

[0072] Figure 6 The figures show the simulation results of the present invention under various working conditions at 45℃.

[0073] Figure 7 This is a comparison chart of the errors of different methods under BJDST conditions according to the present invention;

[0074] Figure 8 This is a comparison chart of the errors of different methods under DST conditions according to the present invention; Detailed Implementation

[0075] Example 1:

[0076] like Figure 1-8 A lithium-ion battery state estimation method based on advanced mobile horizons includes the following steps:

[0077] S1. Establish an equivalent circuit model: Establish an equivalent circuit model of a lithium-ion battery that includes hysteresis effect and temperature-related parameters. The equivalent circuit model includes open-circuit voltage, charge transfer resistance, capacitor capacitance, and hysteresis voltage. Hysteresis effect and temperature-related parameters are all functions of temperature and state of charge (SOC).

[0078] S2. Based on the equivalent circuit model, construct a nonlinear state-space model: The state-space model includes state equations describing the dynamic behavior of the system and observation equations reflecting the voltage response;

[0079] S3. The Moving Horizon Estimation (MHE) optimization framework introduces an Event Triggered Linearization (ETR) mechanism. By determining whether the degree of nonlinear change in the current system exceeds a preset threshold, linearization is only performed when the threshold is exceeded, thereby reducing the computational burden.

[0080] S4. Combining the Gauss-Newton (GN) iteration optimization method, a forward-backward recursive process is executed within each sliding window, and temperature factors and hysteresis states are embedded in the optimization process to dynamically adjust the cost function weights.

[0081] S5. Through a complete modeling, prediction and optimization process, and utilizing real-time measurement data, it outputs high-precision SOC estimation results.

[0082] The equivalent circuit model of a lithium-ion battery is a first-order equivalent circuit model, including a voltage source, series resistor, parallel resistor, and capacitor structure. The equivalent circuit model can be described by the following mathematical expression:

[0083]

[0084] In the formula, U OCV (SOC,T) represents the relationship between the state of charge (SOC), temperature, and open-circuit voltage, derived through fitting experimental data; U1 is the voltage across the R1C1 branch, representing the diffusion voltage; U b R1 represents the battery terminal voltage; R1 and C1 are the parallel branch resistors and capacitors, and R2 represents the battery impedance; the block marked hyst represents the hysteresis effect in the circuit model; Uhyst represents the hysteresis voltage; I is the battery current; and t is the sampling time.

[0085] In the equivalent circuit model, its dynamic characteristics can be described in matrix-vector form, as follows:

[0086]

[0087] In the formula, η k η is the coulombic efficiency of the battery. k Between 0 and 1; γ is a constant that adjusts the hysteresis voltage decay rate; Q bat This represents the total battery capacity.

[0088] Example 2:

[0089] Further explanation in conjunction with Example 1: For example Figure 2 As shown, in the state-space model, state variables are used to characterize the internal dynamic behavior of the battery. The state variables include SOC, polarization voltage, and hysteresis state, which together constitute the state vector x. k x k The formula is defined as follows:

[0090]

[0091] In the formula, Z k Represents SOC, V 1,k H represents the polarization voltage. k This represents a sluggish state.

[0092] In S2, the input variables of the state-space model include the battery's charging and discharging current I. k and the current temperature T k Input vector u k The expression is as follows:

[0093]

[0094] In the formula, I k T represents the charging and discharging current of the battery at time k (discharging is positive, charging is negative); k This is the current temperature of the battery.

[0095] The state equation describes the dynamic evolution of the state variables over time through a temperature-dependent nonlinear transition relationship, and the formula is as follows:

[0096] x k+1 =A k (T k )x k +B k (T k )u k +w k (5);

[0097] In the formula, the state transition matrix A k (T k )for:

[0098]

[0099] Input influence matrix B k (T k The formula representing the influence of input variables on state variables is:

[0100]

[0101] In the formula, Cn is the rated capacity of the battery (Ah); τ 1,k =R 1,k C 1,k R is the polarization time constant; 1,k and C 1,k Temperature-corrected polarization resistance and capacitance; Δt is the sampling time interval; f T (T k ) is the temperature correction factor, used to describe the effect of temperature on battery parameters; process noise w k Follows a Gaussian distribution: w k ~N(0,Q) k (T k ));

[0102] After temperature correction, the process noise covariance matrix Q k (T k The formula is:

[0103]

[0104] In the formula, The process noise covariances are SOC, polarization voltage, and hysteresis state, respectively; η T T is the temperature decay coefficient;ref This is a reference temperature, usually taken as 25℃.

[0105] In S2, the observation equation is used to map the state variables to the observable voltage output, establishing the relationship between the state and the measured voltage. The formula relating the observation equation of the voltage response to the measured voltage is as follows:

[0106] y(k)=U OCV (SOC,T)+U hyst (k)-I(k)·R2-U1(k)+v k (9);

[0107] In the formula, y(k) is the battery voltage, v k To mitigate observation noise, this observation model effectively links the battery state to actual measurements and provides constraints for SOC estimation.

[0108] Example 3:

[0109] Further explanation based on Examples 1-3: In S3, the steps of introducing an event-triggered linearization (ETR) mechanism into the Moving View Estimation (MHE) optimization framework to reduce the computational complexity of nonlinear solutions include:

[0110] S3.1, Define state increment change: In each iteration, calculate the increment between the current state vector and the previous state vector, and use the L2 norm as the criterion for state change. The formula is defined as follows:

[0111] ||Δx k ||2=||x k -x k-1 ||2 (10);

[0112] In the formula, Δx k It is the state increment at the current moment, x k-1 It is the state vector from the previous moment;

[0113] S3.2 Setting the temperature correction trigger threshold: The threshold calculation formula is as follows:

[0114] ε T =ε0+κ T |T k -T ref | (11);

[0115] In the formula, ε0 is the reference trigger threshold, representing the triggering condition at standard temperature; κ T It is the temperature sensitivity coefficient, reflecting the effect of temperature changes on the trigger threshold; T k It is the current temperature, ε T The temperature-adaptive trigger threshold;

[0116] S3.3 Triggering Conditions and Linearization Update: When the state increment satisfies the following condition (12), the system triggers a linearization update operation:

[0117] ||Δx k ||2+γ H |H k -H k-1 |>ε T (12);

[0118] In the formula, γ H H is the weighting coefficient for the hysteresis effect. k The observation matrix;

[0119] S3.4, State Covariance Matrix Update: The update formula is as follows:

[0120]

[0121] S4 includes the following steps:

[0122] S4.1, Temperature Correction State Increment Estimation in the Forward Recursive Process: The temperature-related measurement noise covariance matrix R was constructed. k (T k ), used to correct the state increment, the calculation formula is:

[0123]

[0124] In the formula, Δx k R is the increment at the current moment. k (T k ) is the temperature-dependent measurement noise covariance matrix;

[0125] S4.2, State increment correction in the backward recursive process: The state correction process adopts the recursive relationship of formula (16):

[0126] Δx k ′=F k (T k )Δx k+1 (16);

[0127] In the formula, Δx′ k The corrected increment, Δx k+1 F is the state increment correction term fed back from the next time step during the backward recursion process. k (T k ) is the backward recursive update matrix;

[0128] S4.3 Temperature-Adaptive Convergence Criterion: The convergence threshold is dynamically adjusted according to the current temperature, and the formula is defined as follows:

[0129] ||y k -U b ||2<δ(T k (17);

[0130] δ(T k )=δ0+γ T |T k -T ref | (18);

[0131] In the formula, δ(T) k ) is the convergence threshold; δ0 is the baseline convergence threshold; γ T It is the effect of temperature change on convergence.

[0132] Example 4:

[0133] Further explanation in conjunction with Examples 1-3: To more clearly illustrate the optimization framework proposed in this invention, such as... Figure 3 As shown, Figure 3 The overall process is shown. This framework optimizes battery SOC estimation by dynamically determining incremental changes in the ETR strategy and incorporating temperature factors, while GN iteration further improves the model's adaptability to hysteresis effects and temperature variations by introducing forward-backward recursive optimization. In each iteration, key parameters such as the covariance matrix and weighting matrix are progressively updated to ensure the accuracy and robustness of battery state estimation under different temperature conditions.

[0134] Example 5:

[0135] Further explanation in conjunction with Examples 1-4: For example Figures 4-6 As shown:

[0136] This invention uses a lithium battery dataset as the basis for simulation. This dataset contains dynamic current load data under different ambient temperatures (0℃, 25℃, 45℃), mainly including two typical operating conditions: Dynamic Stress Test (DST) and Beijing Dynamic Stress Test (BJDST). At each temperature, two scenarios are set with initial SOC values ​​of 0.5 and 0.8, forming four sets of test conditions. This embodiment will describe the simulation process and scheduling results under each test condition at different temperatures, divided by ambient temperature.

[0137] To facilitate comparison with existing methods, this embodiment refers to the advanced moving horizon estimation method for lithium-ion battery state estimation described in this invention as Adv_MHE (Advanced Moving Horizon Estimation), and compares its performance with traditional extended Kalman filter (EKF), adaptive extended Kalman filter (AEKF) methods, as well as their improved versions (Adv_EKF, Adv_AEKF) and the standard MHE algorithm.

[0138] like Figure 4-6 As shown, under low temperature (0℃) conditions, traditional algorithms generally suffer from strong lag and severe error accumulation, while the algorithm of this invention can still stably track the real SOC, demonstrating good anti-interference ability; at the nominal temperature (25℃), Adv_MHE can accurately adapt to the dynamic behavior of the battery and significantly reduce the estimation error; in the high temperature (45℃) scenario, the algorithm still maintains the minimum deviation, especially under the condition of system nonlinear enhancement, it still has excellent time response performance and generalization ability.

[0139] The above results demonstrate that the method of the present invention has stable and reliable applicability under multiple temperatures, operating conditions, and initial states, and can provide a high-precision and robust solution for state estimation of lithium-ion batteries, with good engineering practical value and promotion prospects.

[0140] Example 6:

[0141] Further explanation in conjunction with Examples 1-5: For example Figures 6-8 As shown:

[0142] To further quantify the performance of the proposed method, the root mean square error (RMSE) and mean absolute error (MAE) of the proposed algorithm in SOC estimation are compared and analyzed with existing methods. The comparison of these error metrics verifies the accuracy and robustness advantages of the proposed method under different operating conditions.

[0143] Figure 7 and Figure 8 The RMSE and MAE error values ​​of each algorithm under BJDST and DST conditions are shown respectively. The comparison results show that the algorithm of the present invention effectively reduces the error under all conditions and demonstrates superior estimation performance.

[0144] The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A lithium-ion battery state estimation method based on advanced mobile horizons, characterized by: Includes the following steps: S1. Establish an equivalent circuit model: Establish an equivalent circuit model of a lithium-ion battery that includes hysteresis effect and temperature-related parameters. The equivalent circuit model includes open-circuit voltage, charge transfer resistance, capacitor capacitance, and hysteresis voltage. Hysteresis effect and temperature-related parameters are functions of temperature and state of charge. S2. Based on the equivalent circuit model, construct a nonlinear state-space model: The state-space model includes state equations describing the dynamic behavior of the system and observation equations reflecting the voltage response; S3. An event-triggered linearization mechanism is introduced into the moving horizon estimation optimization framework. By determining whether the degree of nonlinear change in the current system exceeds a preset threshold, linearization is only performed when the threshold is exceeded, so as to reduce the computational burden. S4. Combining the Gauss-Newton iterative optimization method, a forward-backward recursive process is executed within each sliding window, and temperature factors and hysteresis states are embedded in the optimization process to dynamically adjust the weight of the cost function. S5. Through a complete modeling, prediction and optimization process, and utilizing real-time measurement data, it outputs high-precision SOC estimation results.

2. The lithium-ion battery state estimation method based on advanced mobile horizons according to claim 1, characterized in that: in In S1, the equivalent circuit model of the lithium-ion battery is a first-order equivalent circuit model, which includes a voltage source, series resistor, parallel resistor and capacitor structure. The equivalent circuit model is described by the following mathematical expression: In the formula, U OCV (SOC,T) represents the relationship between the state of charge, temperature, and open-circuit voltage, derived through fitting experimental data; U1 is the voltage across the R1C1 branch, representing the diffusion voltage; U b R1 represents the battery terminal voltage; R1 and C1 are the parallel branch resistors and capacitors, and R2 represents the battery impedance; the block marked hyst represents the hysteresis effect in the circuit model; Uhyst represents the hysteresis voltage; I is the battery current; and t is the sampling time.

3. The lithium-ion battery state estimation method based on advanced mobile horizons according to claim 2, characterized in that: In the equivalent circuit model, its dynamic characteristics can be described in matrix-vector form, as follows: In the formula, η k η is the coulombic efficiency of the battery. k Between 0 and 1; γ is a constant that adjusts the hysteresis voltage decay rate; Q bat This represents the total battery capacity.

4. The lithium-ion battery state estimation method based on advanced mobile horizons according to claim 1, characterized in that: In S2, the state-space model uses state variables to characterize the internal dynamic behavior of the battery. These state variables include SOC, polarization voltage, and hysteresis state, which together form the state vector x. k x k The formula is defined as follows: In the formula, Z k Represents SOC, V 1,k H represents the polarization voltage. k This represents a sluggish state.

5. The lithium-ion battery state estimation method based on advanced mobile horizons according to claim 1, characterized in that: S2 In the state-space model, the input variables include the battery's charging and discharging current and the current temperature, and the input vector u k The expression is as follows: In the formula, I k T represents the charging and discharging current of the battery at time k; k This is the current temperature of the battery.

6. The lithium-ion battery state estimation method based on advanced mobile horizons according to claim 1, characterized in that: In S2, the state equation describes the dynamic evolution of the state variables over time through a temperature-dependent nonlinear transition relationship, as shown in the following formula: x k+1 =A k (T k )x k +B k (T k )u k +w k (5); In the formula, the state transition matrix A k (T k )for: Input influence matrix B k (T k The formula representing the influence of input variables on state variables is: In the formula, Cn is the rated capacity of the battery; τ 1,k =R 1,k C 1,k R is the polarization time constant; 1,k and C 1,k Temperature-corrected polarization resistance and capacitance; Δt is the sampling time interval; f T (T k ) is the temperature correction factor, used to describe the effect of temperature on battery parameters; process noise w k Follows a Gaussian distribution: w k ~N(0,Q) k (T k )); After temperature correction, the process noise covariance matrix Q k (T k The formula is: In the formula, The process noise covariances for SOC, polarization voltage, and hysteresis state are respectively. η T T is the temperature decay coefficient; ref This is a reference temperature, taken as 25℃.

7. The lithium-ion battery state estimation method based on advanced mobile horizons according to claim 1, characterized in that: In S2, the relationship between the observed voltage response equation and the measured voltage is as follows: y(k)=U OCV (SOC,T)+U hyst (k)-I(k)·R2-U1(k)+v k (9); In the formula, y(k) is the battery voltage, v k To observe noise.

8. The lithium-ion battery state estimation method based on advanced mobile horizons according to claim 1, characterized in that: In S3, the moving horizon estimation optimization framework introduces an event-triggered linearization mechanism. The steps to reduce the computational complexity of nonlinear solutions include: S3.1, Define state increment change: In each iteration, calculate the increment between the current state vector and the previous state vector, and use the L2 norm as the criterion for state change. The formula is defined as follows: ||Δx k ||2=||x k -x k-1 ||2 (10); In the formula, Δx k It is the state increment at the current moment, x k-1 It is the state vector from the previous moment; S3.2 Setting the temperature correction trigger threshold: The threshold calculation formula is as follows: e T =ε0+κ T |T k -T ref | (11); In the formula, ε0 is the reference trigger threshold, representing the triggering condition at standard temperature; κ T It is the temperature sensitivity coefficient, reflecting the effect of temperature changes on the trigger threshold; T k It is the current temperature, ε T The temperature-adaptive trigger threshold; S3.3 Triggering Conditions and Linearization Update: When the state increment satisfies the following condition (12), the system triggers a linearization update operation: ||Δx k ||2+γ H |H k -H k-1 |>ε T (12); In the formula, γ H H is the weighting coefficient for the hysteresis effect. k The observation matrix; S3.4, State Covariance Matrix Update: The update formula is as follows:

9. The lithium-ion battery state estimation method based on advanced mobile horizons according to claim 1, characterized in that: S4 includes the following steps: S4.1, Temperature Correction State Increment Estimation in the Forward Recursive Process: The temperature-related measurement noise covariance matrix R was constructed. k (T k ), used to correct the state increment, the calculation formula is: In the formula, Δx k R is the increment at the current moment. k (T k ) is the temperature-dependent measurement noise covariance matrix; S4.2, State increment correction in the backward recursive process: The state correction process adopts the recursive relationship of formula (16): Δx k ′=F k (T k )Δx k+1 (16); In the formula, Δx′ k The corrected increment, Δx k+1 F is the state increment correction term fed back from the next time step during the backward recursion process. k (T k ) is the backward recursive update matrix; S4.3 Temperature-Adaptive Convergence Criterion: The convergence threshold is dynamically adjusted according to the current temperature, and the formula is defined as follows: ||y k -U b ||2<δ(T k ) (17); δ(T k )=δ0+γ T |T k -T ref | (18); In the formula, δ(T) k ) is the convergence threshold; δ0 is the baseline convergence threshold; γ T It is the effect of temperature change on convergence.

Images