Multi-time-scale SOC and SOH synergistic estimation method considering temperature effect

Abstract

The present invention discloses a multi-time scale SOC and SOH collaborative estimation method considering the influence of temperature, and the steps are as follows: constructing a lithium battery equivalent circuit model considering the influence of temperature; using a particle swarm optimization algorithm with dynamic inertia weight to identify the lithium battery equivalent circuit model; inputting the experimental data and identification results under constant current pulse conditions into the lithium battery equivalent circuit model for simulation verification; constructing a collaborative estimator considering the influence of temperature to estimate the lithium battery SOC and SOH. The present invention realizes the accurate calculation of the battery SOC and SOH under the condition of changing the battery temperature, and realizes the collaborative calculation of SOC and SOH under different calculation cycles of the same algorithm, eliminating the influence of the calculation errors between the two.

Description

Technical Field

[0001] The present invention belongs to the technical field of battery management, and in particular relates to a multi-time scale SOC and SOH collaborative estimation method considering temperature influence. Background Art

[0002] With the development of electric vehicles and the continuous increase in market share, vehicle power battery management system (BMS) technology has also been continuously developing. Among them, the battery state of charge (SOC) and state of health (SOH) are two key states in battery management, affecting the vehicle power output and battery life. However, existing sensors cannot directly measure these two states, so algorithms are needed to estimate and calculate the two states.

[0003] Temperature is an important factor affecting the actual use of electric vehicle batteries, so a battery model that takes temperature into account must be established. At the same time, most existing SOC and SOH estimation methods only consider the impact of a single SOC estimate, ignoring the mutual influence between the internal states of the battery. However, the state update cycles between SOC and SOH are different, and the same algorithm cycle cannot be used for estimation calculations.

[0004] In view of this, the present invention proposes a SOC and SOH collaborative estimation method with multiple calculation time scales considering the influence of different temperature changes, which is of great significance to the extension of battery life and the energy efficiency and economy of electric vehicles. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, the purpose of the present invention is to provide a multi-time-scale SOC and SOH collaborative estimation method that takes into account the influence of temperature, so as to solve the problem of inaccurate calculation caused by the existing battery model ignoring the influence of temperature, and at the same time solve the problem of different SOC and SOH state update cycles; the method of the present invention realizes the collaborative estimation of SOC and SOH and reduces unnecessary calculation redundancy.

[0006] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0007] The present invention provides a multi-time scale SOC and SOH collaborative estimation method considering temperature influence, and the steps are as follows:

[0008] 1) Construct an equivalent circuit model of lithium battery considering the influence of temperature;

[0009] 2) using a particle swarm optimization algorithm with dynamic inertia weight to identify the lithium battery equivalent circuit model;

[0010] 3) inputting the experimental data under constant current pulse working conditions and the identification results obtained in step 2) into the lithium battery equivalent circuit model established in step 1) for simulation verification;

[0011] 4) Construct a collaborative estimator that takes temperature effects into account to estimate the SOC and SOH of lithium batteries.

[0012] Furthermore, the method for constructing the lithium battery equivalent circuit model in step 1) is as follows:

[0013] Taking the discharge current as positive, combined with Kirchhoff's law, the spatial state equation of the dual RC battery model is obtained as follows:

[0014]

[0015] Where U OC represents the open circuit voltage of the battery, R0 represents the contact resistance between the internal structures of the battery, R1 is the electrochemical polarization resistance, R2 is the concentration polarization internal resistance, C1 is the electrochemical polarization capacitance, C2 is the concentration polarization capacitance, and U t is the terminal voltage of the battery, I L is the load current;

[0016] Using Laplace transform, U in the above formula (1) is oc Further deduction:

[0017]

[0018] The transfer equation of the battery model is:

[0019]

[0020] Where, τ1=R1C1; τ2=R2C2; τ1 and τ2 are the time constants of the RC polarization unit;

[0021] Discretizing the transfer equation, and mapping the transfer equation from the S plane to the Z plane using a bilinear transformation method;

[0022]

[0023] Where ΔT is the sampling interval; z-1 is the delay factor;

[0024] The equation mapped to the Z plane is:

[0025]

[0026] Transforming the above formula (5) into the discrete time domain, we can obtain:

[0027] y k =U oc,k-U t,k =a1y k-1 +a2y k-2 +a3I k +a4I k-1 +a5I k-2 (6)

[0028] Where U oc,k is the open circuit voltage at time k, U t,k is the terminal voltage at time k, y k is the difference between the open circuit voltage at time k and the terminal voltage at time k, and a1-a5 are the algebraic expressions of the model parameters.

[0029] Furthermore, the step 2) specifically includes:

[0030] 21) Initialization;

[0031] Define and initialize various parameters in the iteration process; limit the particle's flight speed range, and the upper limit of the range is the particle's maximum flight speed v max ; The lower limit of the interval is the minimum flight speed v of the particle min ; Set the value range of the identification parameter θ according to the actual characteristics of the target system max ,θ min , to prevent the result from overflowing; the number of particles is defined as N1, the maximum number of iterations is defined as M, and the value is set according to actual needs; the initial velocity of the particle v i,j＝0 and the initial position θ i,j＝0 Randomly select in the corresponding interval, where the subscript i represents the particle sequence, j represents the current iteration number of the particle; the particle position vector θ i,j Represents the parameter set of the equivalent circuit model [R0, R1, R2, C1, C2];

[0032] The initial velocity v i,0 and the initial position θ i,0 The value of is as follows:

[0033] v i,0 =rand(0,1) (7)

[0034] θ i,0 =θ min +(θ max -θ min )*rand(0,1) (8)

[0035] In the formula, the function rand(0,1) represents the generation of uniformly distributed random numbers between (0,1);

[0036] 22) Select and calculate the fitness function;

[0037] Set the fitness function to:

[0038] U RC (k) = U OC (k)-U t (k) (9)

[0039]

[0040] In formula (9), U RC (k) is the fitness function corresponding to particle i at iteration number j, n is the length of the test data, y(k) is the equivalent circuit model at the current particle position θ i,j The difference between the open circuit voltage and the terminal voltage under

[0041] It can be seen that when a1-a5 is represented by identification parameters, y(k) can be calculated; let:

[0042]

[0043] We can get:

[0044] fy k =aI k-2 +bI k-1 +cI k -dy k-2 -ey k-1 (12)

[0045] The specific expressions of each coefficient are as follows:

[0046]

[0047] During the algorithm iteration, any particle changes its position vector θ i,j Substitute the various parameters in into the above formula (13) to calculate the corresponding fitness function;

[0048] 23) Update of individual optimal fitness;

[0049] The position vector θ of particle i at the current iteration number j i,j Substitute it into the fitness function, calculate the specific value fitness(i,j) and compare it with its own historical optimal position The corresponding fitness value fitness(i,best) is compared. If fitness(i,j)<fitness(i,best), the current particle position is used to replace its historical optimal position.

[0050] 24) Update the optimal fitness of the group;

[0051] Compare the historical best fitness value fitness(i,best) of each particle at the current number of iterations with the optimal position θ of the group best The corresponding fitness value fitness(best) is compared. If fitness(i,best) is less than fitness(best), the historical optimal position of the particle corresponding to fitness(i,best) is used. Update the optimal position of the group θ best ;

[0052] 25) Particle velocity and position update;

[0053] After completing a round of fitness update, the speed of each particle is updated:

[0054]

[0055] Where ω1 and ω2 are inertia weight factors used to adjust the search interval; M is the number of iterations; c1 and c2 are particle acceleration constants, usually between [0,4]; r1 and r2 are random numbers between [0,1] to ensure the randomness of particle search;

[0056] Update the particle position:

[0057] θ i,j+1 =θ i,j +v i,j+1 (15);

[0058] 26) Determine whether the particle swarm optimization algorithm is finished;

[0059] The termination conditions of the particle swarm optimization algorithm are: the current number of iterations reaches the maximum number of iterations and the value of the optimal fitness function of the group is less than the preset value; when any of the above two conditions is met, the algorithm terminates and the output group optimal position θ best That is the optimal solution for this calculation, otherwise return to step 22).

[0060] Furthermore, the step 3) specifically includes:

[0061] Through the parameter identification in step 2), the values ​​of parameters R0, R1, R2, C1, and C2 under different cycles are obtained respectively; in the MATLAB / Simulink environment, the lithium-ion battery equivalent circuit model considering the temperature effect built in step 1) is established through code and Simulink modules, and the experimental data under the constant current pulse condition is brought into the calculation; wherein, the input is the current and the battery temperature; the output is the terminal voltage; the constant current pulse condition test is used to perform simulation at six temperatures of [-10℃0℃10℃20℃30℃40℃].

[0062] Furthermore, the step 4) specifically includes:

[0063] 41) Establish a multi-time scale filtering algorithm model;

[0064] The multi-time scale filtering algorithm model including the implicit state x and the parameter θ is as follows:

[0065]

[0066] In the formula, x k,l t k,l =t k,0 +l×Δt,1≤l≤L Z The system state at the moment, where the dual time scales k and l describe the macroscopic time scale and the microscopic time scale respectively, L Z is the scale conversion limit, that is, a macroscopic time scale equal to L Z A microscopic time scale; u k,l t k,l Input information of the time system; Y k,l t k,l The observation matrix of the time system; w k,l is the system state, and its covariance is p k is a parametric white noise, and its covariance is v k,l To measure white noise, its covariance is R k,l ; Use microscopic time scale to estimate system state and macroscopic time scale to estimate system parameters; for system parameters, their macroscopic scale is between 0 and L Z The value of ~1 remains unchanged, that is,

[0067] 42) Initialization of multi-time scale filtering algorithm;

[0068] Set the parameter observer HIF separately θ and state observer AEKF x Initial parameter values: θ0, λ θ , S θ , R0, x 0，0 , R 0，0 , where θ0, λ θ , S θ They are parameter observer HIF θ Initial parameter values, initial values ​​of the parameter estimation error covariance matrix, initial values ​​of the system noise covariance matrix, parameter observer performance boundaries, and custom matrices; x 0，0 , R 0，0They are state observer AEKF x The initial state of the system, the initial value of the state estimation error covariance matrix, the initial value of the system noise covariance matrix, and the observation noise covariance;

[0069] 43) Parameter observer HIF based on macroscopic time scale θ Time update;

[0070]

[0071] The system parameters θ and parameter estimation error covariance P are θ The corresponding estimated value is obtained

[0072] 44) AEKF, a state observer based on microscopic time scale x Time update;

[0073]

[0074] Start the state observer AEKF x The time update of the system state X is obtained by and its error covariance P x A priori estimate of Based on the prior estimate of the available capacity of the power battery and the prior estimates of the states Update the OCV value of the power battery

[0075] 45) AEKF, a state observer based on microscopic time scale x Measurement updates;

[0076] The state estimation innovation covariance matrix is ​​updated:

[0077]

[0078] Kalman gain matrix:

[0079]

[0080] Adaptive Covariance Matching:

[0081]

[0082] Noise covariance update:

[0083]

[0084]

[0085] System state estimate correction:

[0086]

[0087] State estimation error covariance update:

[0088]

[0089] Microscopic time scale cycle calculation l=1:L Z And scale conversion:

[0090]

[0091]

[0092]

[0093]

[0094] At this point, a cyclic calculation of the microscopic time scale under the macroscopic time scale is completed. The next step is to return to the macroscopic time scale to perform measurement updates of parameter estimates;

[0095] 46) State Observer HIF Based on Macroscopic Time Scale θ Measurement update:

[0096]

[0097] H∞ gain matrix:

[0098]

[0099] System parameter estimate correction:

[0100]

[0101] H∞ characteristic matrix update:

[0102]

[0103] At this point, the multi-time scale estimation of parameters and states at time k is completed, and the state estimation at time (k+1) is ready, and let Get real-time estimation of available capacity and SOC of power battery.

[0104] Beneficial effects of the present invention:

[0105] The present invention realizes accurate calculation of battery SOC and SOH under the condition of changing battery temperature, and simultaneously realizes coordinated calculation of SOC and SOH under different calculation cycles of the same algorithm, eliminating the influence of calculation errors between the two. BRIEF DESCRIPTION OF THE DRAWINGS

[0106] Figure 1 The schematic diagram of the method of the present invention.

[0107] Figure 2 This is the equivalent circuit model diagram of a lithium battery.

[0108] Figure 3 This is a comparison chart between the model estimated voltage and the test voltage.

[0109] Figure 4 Schematic diagram of model identification error.

[0110] Figure 5 This is the principle flow chart of the multi-time scale filtering algorithm.

[0111] Figure 6 This is the SOC estimation result diagram.

[0112] Figure 7 This is the SOH estimation result diagram. DETAILED DESCRIPTION

[0113] In order to facilitate the understanding of those skilled in the art, the present invention is further described below in conjunction with embodiments and drawings. The contents mentioned in the implementation modes are not intended to limit the present invention.

[0114] Reference Figure 1 As shown, a multi-time scale SOC and SOH collaborative estimation method considering temperature influence of the present invention comprises the following steps:

[0115] 1) Construct a lithium battery equivalent circuit model considering the influence of temperature, referring to Figure 2 As shown;

[0116] Specifically, the method for constructing the lithium battery equivalent circuit model in step 1) is as follows:

[0117] Taking the discharge current as positive, combined with Kirchhoff's law, the spatial state equation of the dual RC battery model is obtained as follows:

[0118]

[0119] Where U OC represents the open circuit voltage of the battery, R0 represents the contact resistance between the internal structures of the battery, R1 is the electrochemical polarization resistance, R2 is the concentration polarization internal resistance, C1 is the electrochemical polarization capacitance, C2 is the concentration polarization capacitance, and U t is the terminal voltage of the battery, I L is the load current;

[0120] Using Laplace transform, Uo in the above formula (1) is c Further deduction:

[0121]

[0122] The transfer equation of the battery model is:

[0123]

[0124] Where, τ1=R1C1; τ2=R2C2; τ1 and τ2 are the time constants of the RC polarization unit;

[0125] Discretizing the transfer equation, and mapping the transfer equation from the S plane to the Z plane using a bilinear transformation method;

[0126]

[0127] Where ΔT is the sampling interval; z-1 is the delay factor;

[0128] The equation mapped to the Z plane is:

[0129]

[0130] Transforming the above formula (5) into the discrete time domain, we can obtain:

[0131] y k =U oc,k -U t,k =a1y k-1 +a2y k-2 +a3I k +a4I k-1 +a5I k-2 (6)

[0132] Where U oc,k is the open circuit voltage at time k, U t,k is the terminal voltage at time k, y k is the difference between the open circuit voltage at time k and the terminal voltage at time k, and a1-a5 are the algebraic expressions of the model parameters.

[0133] 2) Using a particle swarm optimization algorithm with dynamic inertia weight to identify the lithium battery equivalent circuit model; specifically including:

[0134] 21) Initialization;

[0135] Define and initialize various parameters in the iteration process; limit the particle's flight speed range, and the upper limit of the range is the particle's maximum flight speed v max (It should not be too large, otherwise the optimal value will be lost during iteration); the lower limit of the interval is the minimum flight speed v of the particle min (It cannot be too small, otherwise it will reduce the iteration efficiency of the algorithm); set the value range of the identification parameter θ according to the actual characteristics of the target system max ,θ min, to prevent the result from overflowing; the number of particles is defined as N1, the maximum number of iterations is defined as M, and the value is set according to actual needs; the initial velocity of the particle v i,j＝0 and the initial position θ i,j＝0 Randomly select in the corresponding interval, where the subscript i represents the particle sequence, j represents the current iteration number of the particle; the particle position vector θ i,j Represents the parameter set of the equivalent circuit model [R0, R1, R2, C1, C2];

[0136] The initial velocity v i,0 and the initial position θ i,0 The value of is as follows:

[0137] v i,0 =rand(0,1) (7)

[0138] θ i,0 =θ min +(θ max -θ min )*rand(0,1) (8)

[0139] In the formula, the function rand(0,1) represents the generation of uniformly distributed random numbers between (0,1);

[0140] 22) Select and calculate the fitness function;

[0141] Set the fitness function to:

[0142] U RC (k) = U OC (k)-U t (k) (9)

[0143]

[0144] In formula (9), U RC (k) is the fitness function corresponding to particle i at iteration number j, n is the length of the test data, y(k) is the equivalent circuit model at the current particle position θ i,j The difference between the open circuit voltage and the terminal voltage under

[0145] It can be seen that when a1-a5 is represented by identification parameters, y(k) can be calculated; let:

[0146]

[0147] We can get:

[0148] fy k =aI k-2 +bI k-1 +cI k -dyk-2 -ey k-1 (12)

[0149] The specific expressions of each coefficient are as follows:

[0150]

[0151] During the algorithm iteration, any particle changes its position vector θ i,j Substitute the various parameters in into the above formula (13) to calculate the corresponding fitness function;

[0152] 23) Update of individual optimal fitness;

[0153] The position vector θ of particle i at the current iteration number j i,j Substitute it into the fitness function, calculate the specific value fitness(i,j) and compare it with its own historical optimal position The corresponding fitness value fitness(i,best) is compared. If fitness(i,j)<fitness(i,best), the current particle position is used to replace its historical optimal position.

[0154] 24) Update the optimal fitness of the group;

[0155] Compare the historical best fitness value fitness(i,best) of each particle at the current number of iterations with the optimal position θ of the group best The corresponding fitness value fitness(best) is compared. If fitness(i,best) is less than fitness(best), the historical optimal position of the particle corresponding to fitness(i,best) is used. Update the optimal position of the group θ best ;

[0156] 25) Particle velocity and position update;

[0157] After completing a round of fitness update, the speed of each particle is updated:

[0158]

[0159] Where ω1 and ω2 are inertia weight factors used to adjust the search interval (ω1 = 0.4, ω2 = 0.9); M is the number of iterations, c1 and c2 are particle acceleration constants, usually between [0, 4] (c1 = c2 = 2); r1 and r2 are random numbers between [0, 1] to ensure the randomness of particle search;

[0160] Update the particle position:

[0161] θ i,j+1 =θ i,j +v i,j+1 (15);

[0162] 26) Determine whether the particle swarm optimization algorithm is finished;

[0163] The termination conditions of the particle swarm optimization algorithm are: the current number of iterations reaches the maximum number of iterations and the value of the optimal fitness function of the group is less than the preset value; when any of the above two conditions is met, the algorithm terminates and the output group optimal position θ best That is the optimal solution for this calculation, otherwise return to step 22).

[0164] 3) inputting the experimental data under constant current pulse working conditions and the identification results obtained in step 2) into the lithium battery equivalent circuit model established in step 1) for simulation verification;

[0165] Through the parameter identification in step 2), the values ​​of parameters R0, R1, R2, C1, and C2 under different cycles are obtained respectively; in the MATLAB / Simulink environment, the lithium-ion battery equivalent circuit model considering the temperature effect built in step 1) is established through code and Simulink modules, and the experimental data under the constant current pulse condition is brought into the calculation; wherein, the input is the current and the battery temperature; the output is the terminal voltage; the constant current pulse condition test is used to perform simulation at six temperatures of [-10℃0℃10℃20℃30℃40℃].

[0166] Take the temperature of 40℃ as an example: Figure 3 and Figure 4 As shown, the measured voltage and the simulated voltage are compared; and the corresponding error is basically within 0.005, as shown Figure 4 As shown; at other temperatures [10℃ 0℃ 10℃ 20℃ 30℃], the errors are [0.007 0.008 0.006 0.004 0.008 0.009] respectively.

[0167] 4) Construct a collaborative estimator that considers the temperature effect to estimate the SOC and SOH of lithium batteries; refer to Figure 5 As shown, specifically:

[0168] 41) Establish a multi-time scale filtering algorithm model;

[0169] The multi-time scale filtering algorithm model including the implicit state x and the parameter θ is as follows:

[0170]

[0171] In the formula, x k,l t k,l=t k,0 +l×Δt,1≤l≤L Z The system state at the moment, where the dual time scales k and l describe the macroscopic time scale and the microscopic time scale respectively, L Z is the scale conversion limit, that is, a macroscopic time scale equal to L Z A microscopic time scale; u k,l t k,l Input information of the time system; Y k,l t k,l The observation matrix of the time system; w k,l is the system state, and its covariance is p k is a parametric white noise, and its covariance is v k,l To measure white noise, its covariance is R k,l ; Use microscopic time scale to estimate system state and macroscopic time scale to estimate system parameters; for system parameters, their macroscopic scale is between 0 and L Z The value of ~1 remains unchanged, that is,

[0172] 42) Initialization of multi-time scale filtering algorithm;

[0173] Set the parameter observer HIF separately θ and state observer AEKF x Initial parameter values: θ0, λ θ , S θ , R0, x 0，0 , R 0，0 , where θ0, λ θ , S θ They are parameter observer HIF θ Initial parameter values, initial values ​​of the parameter estimation error covariance matrix, initial values ​​of the system noise covariance matrix, parameter observer performance boundaries, and custom matrices; x 0，0 , R 0，0 They are state observer AEKF x The initial state of the system, the initial value of the state estimation error covariance matrix, the initial value of the system noise covariance matrix, and the observation noise covariance;

[0174] 43) Parameter observer HIF based on macroscopic time scale θ Time update (a priori estimate);

[0175]

[0176] The system parameters θ and parameter estimation error covariance P are θ The corresponding estimated value is obtained

[0177] 44) AEKF, a state observer based on microscopic time scale x Time update (a priori estimate);

[0178]

[0179] Start the state observer AEKF x The time update of the system state X is obtained by and its error covariance P x A priori estimate of Based on the prior estimate of the available capacity of the power battery and the prior estimates of the states Update the OCV value of the power battery

[0180] 45) AEKF, a state observer based on microscopic time scale x Measurement update (a posteriori estimate) of ;

[0181] The state estimation innovation covariance matrix is ​​updated:

[0182]

[0183] Kalman gain matrix:

[0184]

[0185] Adaptive covariance matching (voltage estimation error window function):

[0186]

[0187] Noise covariance update:

[0188]

[0189]

[0190] System state estimate correction:

[0191]

[0192] State estimation error covariance update:

[0193]

[0194] Microscopic time scale cycle calculation l=1:L Zand scale conversion (when l = L Z hour):

[0195]

[0196]

[0197]

[0198]

[0199] At this point, a cyclic calculation of the microscopic time scale under the macroscopic time scale is completed. The next step is to return to the macroscopic time scale to perform measurement update (posterior estimation) of parameter estimation.

[0200] 46) State Observer HIF Based on Macroscopic Time Scale θ The measurement update (a posteriori estimate) of is:

[0201]

[0202] H∞ gain matrix:

[0203]

[0204] System parameter estimate correction:

[0205]

[0206] H∞ characteristic matrix update:

[0207]

[0208] At this point, the multi-time scale estimation of parameters and states at time k is completed, and the state estimation at time (k+1) is ready, and let Get real-time estimation of available capacity and SOC of power battery.

[0209] like Figure 6 and Figure 7 As shown, the method of the present invention can stably estimate the values ​​of SOC and SOH, and the SOC estimation error is within 1.5%, and the SOH estimation error is within 1%, thereby verifying the effectiveness and accuracy of the estimation method of the present invention.

[0210] The present invention has many specific application paths. The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements can be made without departing from the principle of the present invention. These improvements should also be regarded as the protection scope of the present invention.

Claims

1. A multi-time scale SOC and SOH collaborative estimation method considering temperature influence, characterized in that: Here are the steps: 1) Construct an equivalent circuit model of lithium battery considering the influence of temperature; 2) using a particle swarm optimization algorithm with dynamic inertia weight to identify the lithium battery equivalent circuit model; 3) inputting the experimental data under constant current pulse working conditions and the identification results obtained in step 2) into the lithium battery equivalent circuit model established in step 1) for simulation verification; 4) Construct a collaborative estimator that considers the temperature effect to estimate the SOC and SOH of lithium batteries; The method for constructing the lithium battery equivalent circuit model in step 1) is as follows: Taking the discharge current as positive, combined with Kirchhoff's law, the spatial state equation of the dual RC battery model is obtained as follows: Where U OC represents the open circuit voltage of the battery, R0 represents the contact resistance between the internal structures of the battery, R1 is the electrochemical polarization resistance, R2 is the concentration polarization internal resistance, C1 is the electrochemical polarization capacitance, C2 is the concentration polarization capacitance, and U t is the terminal voltage of the battery, I L is the load current; Using Laplace transform, U in the above formula (1) is oc Further deduction: The transfer equation of the battery model is: Where, τ1=R1C1; τ2=R2C2; τ1 and τ2 are the time constants of the RC polarization unit; Discretizing the transfer equation, and mapping the transfer equation from the S plane to the Z plane using a bilinear transformation method; Where ΔT is the sampling interval; z -1 is the delay factor; The equation mapped to the Z plane is: Transforming the above formula (5) into the discrete time domain, we can obtain: <h2 style=";text-align:left;direction:ltr">y<h2 style=";text-align:left;direction:ltr"> k <h2 style=";text-align:left;direction:ltr"> =U<h2 style=";text-align:left;direction:ltr"> oc,k <h2 style=";text-align:left;direction:ltr"> -U<h2 style=";text-align:left;direction:ltr"> t,k <h2 style=";text-align:left;direction:ltr"> =a1y<h2 style=";text-align:left;direction:ltr"> k-1 <h2 style=";text-align:left;direction:ltr"> +a2y<h2 style=";text-align:left;direction:ltr"> k-2 <h2 style=";text-align:left;direction:ltr"> +a3I<h2 style=";text-align:left;direction:ltr"> k <h2 style=";text-align:left;direction:ltr"> +a4I<h2 style=";text-align:left;direction:ltr"> k-1 <h2 style=";text-align:left;direction:ltr"> +a5I<h2 style=";text-align:left;direction:ltr"> k-2 <h2 style=";text-align:left;direction:ltr"> (6) Where U oc,k is the open circuit voltage at time k, U t,k is the terminal voltage at time k, y k is the difference between the open circuit voltage at time k and the terminal voltage at time k, a1-a5 are the algebraic expressions of the model parameters; The step 4) specifically includes: 41) Establish a multi-time scale filtering algorithm model; The multi-time scale filtering algorithm model including the implicit state x and the parameter θ is as follows: In the formula, x k,l t k,l =t k,0 +l×Δt,1≤l≤L Z The system state at the moment, where the dual time scales k and l describe the macroscopic time scale and the microscopic time scale respectively, L Z is the scale conversion limit, that is, a macroscopic time scale equal to L Z A microscopic time scale; u k,l t k,l Input information of the time system; Y k,l t k,l The observation matrix of the time system; w k,l is the system state, and its covariance is p k is a parametric white noise, and its covariance is v k,l To measure white noise, its covariance is R k,l ; Use microscopic time scale to estimate system state and macroscopic time scale to estimate system parameters; for system parameters, their macroscopic scale is between 0 and L Z The value of ~1 remains unchanged, that is, 42) Initialization of multi-time scale filtering algorithm; Set the parameter observer HIF separately θ and state observer AEKF x Initial parameter values: θ0, λ θ , S θ , R0, x 0，0 , R 0，0 , where θ0, λ θ , S θ They are parameter observer HIF θ Initial parameter values, initial values ​​of the parameter estimation error covariance matrix, initial values ​​of the system noise covariance matrix, parameter observer performance boundaries, and custom matrices; x 0，0 , R 0，0 They are state observer AEKF x The initial state of the system, the initial value of the state estimation error covariance matrix, the initial value of the system noise covariance matrix, and the observation noise covariance; 43) Parameter observer HIF based on macroscopic time scale θ Time update; The system parameters θ and parameter estimation error covariance P are θ The corresponding estimated value is obtained 44) AEKF, a state observer based on microscopic time scale x Time update; Start the state observer AEKF x The time update of the system state X is obtained by and its error covariance P x A priori estimate of Based on the prior estimate of the available capacity of the power battery and the prior estimates of the states Update the OCV value of the power battery 45) AEKF, a state observer based on microscopic time scale x Measurement updates; State estimation innovation covariance matrix update: Kalman gain matrix: Adaptive Covariance Matching: Noise covariance update: System state estimate correction: State estimation error covariance update: Microscopic time scale cycle calculation l=1:L Z And scale conversion: At this point, a cyclic calculation of the microscopic time scale under the macroscopic time scale is completed. The next step is to return to the macroscopic time scale to perform measurement updates of parameter estimates; 46) State Observer HIF Based on Macroscopic Time Scale θ Measurement update: H∞ gain matrix: System parameter estimate correction: H∞ characteristic matrix update: At this point, the multi-time scale estimation of parameters and states at time k is completed, and the state estimation at time (k+1) is ready, and let Get real-time estimated values ​​of the power battery’s available capacity and SOC.

2. The multi-time scale SOC and SOH collaborative estimation method considering temperature influence according to claim 1 is characterized in that: The step 2) specifically includes: 21) Initialization; Define and initialize various parameters in the iteration process; limit the particle's flight speed range, and the upper limit of the range is the particle's maximum flight speed v max ; The lower limit of the interval is the minimum flight speed v of the particle min ; Set the value range of the identification parameter θ according to the actual characteristics of the target system max ,θ min , to prevent the result from overflowing; the number of particles is defined as N1, the maximum number of iterations is defined as M, and the value is set according to actual needs; the initial velocity of the particle v i,j＝0 and the initial position θ i,j＝0 Randomly select in the corresponding interval, where the subscript i represents the particle sequence, j represents the current iteration number of the particle; the particle position vector θ i,j Represents the parameter set of the equivalent circuit model [R0, R1, R2, C1, C2]; The initial velocity v i,0 and the initial position θ i,0 The value of is as follows: V i,0 =row(0,1) (7) i i,0 =θ min +(θ max -θ min )*rand(0,1) (8) In the formula, the function rand(0,1) represents the generation of uniformly distributed random numbers between (0,1); 22) Select and calculate the fitness function; Set the fitness function to: IN RC (k)=U OC (k)-U t (k) (9) In formula (9), U RC (k) is the fitness function corresponding to particle i at iteration number j, n is the length of the test data, y(k) is the equivalent circuit model at the current particle position θ i,j The difference between the open circuit voltage and the terminal voltage under It can be seen that when a1-a5 is represented by identification parameters, y(k) can be calculated; let: We can get: fy k sI k-2 +bI k-1 +cI k -dy k-2 -ey k-1 (12) The specific expressions of each coefficient are as follows: During the algorithm iteration, any particle changes its position vector θ i,j Substitute the various parameters in into the above formula (13) to calculate the corresponding fitness function; 23) Update of individual optimal fitness; The position vector θ of particle i at the current iteration number j i,j Substitute it into the fitness function, calculate the specific value fitness(i,j) and compare it with its own historical optimal position The corresponding fitness value fitness(i,best) is compared. If fitness(i,j)<fitness(i,best), the current particle position is used to replace its historical optimal position. 24) Update the optimal fitness of the group; Compare the historical best fitness value fitness(i,best) of each particle at the current number of iterations with the optimal position θ of the group best The corresponding fitness value fitness(best) is compared. If fitness(i,best) is less than fitness(best), the historical optimal position of the particle corresponding to fitness(i,best) is used. Update the optimal position of the group θ best ; 25) Particle velocity and position update; After completing a round of fitness update, the speed of each particle is updated: Where ω1 and ω2 are inertia weight factors used to adjust the search interval; M is the number of iterations; c1 and c2 are particle acceleration constants, usually between [0, 4]; r1 and r2 are random numbers between [0, 1] to ensure the randomness of particle search; Update the particle position: i i,j+1 =θ i,j +v i,j+1 (15); 26) Determine whether the particle swarm optimization algorithm is finished; The termination conditions of the particle swarm optimization algorithm are: the current number of iterations reaches the maximum number of iterations and the value of the optimal fitness function of the group is less than the preset value; when any of the above two conditions is met, the algorithm terminates and the output group optimal position θ best That is the optimal solution for this calculation, otherwise return to step 22).

3. The multi-time scale SOC and SOH collaborative estimation method considering temperature influence according to claim 2 is characterized in that: The step 3) specifically includes: The values ​​of parameters R0, R1, R2, C1, and C2 under different cycles are obtained by parameter identification in step 2); in the MATLAB / Simulink environment, the lithium-ion battery equivalent circuit model considering the temperature effect built in step 1) is established through code and Simulink modules, and the experimental data under constant current pulse conditions are brought into the calculation; wherein the input is current and battery temperature; the output is terminal voltage; constant current pulse condition test is used to perform simulation at six temperatures of [-10℃0℃10℃20℃30℃40℃].

Images