Parallel battery system power state prediction method considering aging degree

Abstract

The invention discloses a parallel battery system power prediction method considering the aging degree, and the method comprises the following steps: firstly, obtaining a branch equivalent battery monomer SOC initial value SOC0 through a self-adaptive unscented Kalman filter based on a battery monomer model in combination with the measurement current and voltage, and then carrying out the inconsistency classification of the branch equivalent battery monomer SOC initial value SOC0 through an inconsistency classification module; dividing the equivalent single battery into a normal part, a slightly aged part and a seriously aged part, and compensating different aged batteries according to a battery capacity aging model and a difference prediction module to obtain a corrected SOC value SOCi; determining the safety current Is of the battery by using the SOCi, the open-circuit voltage U0 and the design current Ide, predicting the equivalent battery monomer power SOP0 and SOPk by combining the terminal voltage Ut, multiplying the SOP0 by the number m to obtain the normal battery pack power SOPi, and compensating the aged battery power SOPk according to a battery internal resistance aging model and a difference prediction module to obtain the corrected aged battery power SOPj; and finally, superposing the SOPi and the SOPj to obtain a power prediction value SOPf of the battery system.

Description

Technical Field

[0001] The invention belongs to the technical field of battery management and relates to a power state prediction method for a parallel battery system taking aging degree into consideration. Background Art

[0002] With the frequent use of battery energy storage systems (BESS) in grid connection, the prediction of battery system power state (SOP) has become one of the most important technologies in battery applications such as electric vehicles and grid energy storage systems. The SOP prediction of the battery system can help the management system to reasonably schedule the battery charging and discharging process, avoid over-discharging or overcharging, thereby extending the battery life and improving the overall performance of the system. However, the current SOP prediction technology of parallel battery systems is still immature. On the one hand, the SOP of the battery is easily affected by many factors, including voltage, current, state of charge and health status, etc. These factors have dynamic characteristics of multiple time scales, which increases the difficulty of SOP prediction. On the other hand, there is inconsistency between different single cells in the battery pack. This inconsistency will reduce the overall performance of the battery pack, increase the prediction error, and may cause safety hazards.

[0003] Patent (CN117110889A) discloses a power state prediction method for a parallel battery system. The method uses the current difference to judge the inconsistency of the battery cells, and establishes a multi-constrained safety current selector to obtain the safe current value under the constraints of SOC, voltage and current, and then obtain the battery system SOP base value, and then compensate the battery power state with different degrees of inconsistency through neural network and multiplier. However, this method does not take into account the problem of the decrease in the accuracy of current compensation for battery inconsistency under long time scales. In view of this problem, the present invention establishes a battery inconsistency compensation module that considers aging factors, avoiding the problem of decreased prediction accuracy of short time scales such as current and voltage. Patent (CN113807039A) discloses a power state prediction method for a series battery system. First, the SOP base value of the battery system is obtained, and then a current regulator is established considering the inconsistency between battery cells, and then the battery system SOP compensation value is obtained through the BP neural network. The base value and the compensation value are superimposed to obtain the battery system SOP prediction value. However, this method ignores the problem of whether there is inconsistency in the battery cells in the battery system and the size of the inconsistency difference. To address this problem, the present invention proposes to establish an aging inconsistency classification module to detect the difference in the degree of inconsistency of battery cells, and to perform staged SOC and SOP compensation for battery cells with different degrees of aging inconsistency, so as to obtain an accurate power state of the battery system. Summary of the invention

[0004] The problem solved by the present invention is to provide a power state prediction method for a parallel battery system taking into account the degree of aging. On the one hand, the batteries are divided into three categories according to the different degrees of aging inconsistency, so as to compensate for the inconsistency of battery cells over a long time scale; on the other hand, the SOC and SOP compensation are performed on battery cells with different aging degrees through the battery aging model and APSO-random forest, so as to solve the problem of the difference in the inconsistency of battery cells and obtain accurate SOC and SOP compensation values ​​of the parallel battery system.

[0005] In order to achieve the above objectives, the present invention proposes the following technical solutions:

[0006] S1. Establish a second-order RC battery cell equivalent circuit model (1) based on experimental data, and obtain the battery cell spatial state equation;

[0007] S2, taking the battery cell spatial state equation as input, combining the battery cell measured voltage value and the battery cell measured current value, and passing through the adaptive unscented Kalman filter (2), obtain the initial value of the battery cell state of charge SOC0;

[0008] S3, taking the maximum available capacity of each branch battery cell and the internal resistance of each branch battery cell as input, the aging inconsistency classification module classifies each branch battery cell into three categories: normal, slightly aged, and severely aged. The specific classification operation is as follows: In the aging inconsistency classification module, the maximum available capacity of each branch battery cell is subtracted from the maximum available capacity of a normal healthy battery to obtain the battery cell capacity deviation value ΔQ1~ΔQ n At the same time, the internal resistance of each branch battery cell is subtracted from the normal healthy battery internal resistance to obtain the battery cell internal resistance deviation value ΔR1~ΔR n , combined with fuzzy logic (3), the aging inconsistency classifier (4) classifies each battery cell into three categories: normal, slightly aged, and severely aged;

[0009] S4. For the three different types of aging battery cells, the difference compensation module is used to perform state of charge SOC compensation and power state SOP compensation. The specific compensation operation is as follows: For normal battery cells, the output SOC compensation value ΔSOC is obtained through the corresponding zero-difference compensator (5). b (b∈[1,n]) is 0; for slightly aged battery cells, the battery cell capacity difference ΔQ p Difference ΔR from the internal resistance of the battery cell p As input, the SOC compensation value ΔSOC is obtained through the battery aging model (6) p (p∈[1,n]) and SOP compensation value ΣΔSOP p (p∈[1,n]); for severely aged battery cells, the battery cell capacity difference ΔQ q、Current I q Difference ΔR from the internal resistance of the battery cell q 、Current I q As input, the SOC compensation value ΔSOC is obtained through the random forest compensator (7) q (q∈[1,n]) and SOP compensation value ΣΔSOP q (q∈[1,n]);

[0010] S5, SOC0, ΔSOC b , ΔSOC p and ΔSOC q As input, the SOC value SOC of each battery cell after compensation is obtained through the SOC summator (8). i , the specific calculation is:

[0011]

[0012] In the formula, λ C,p ,λ C,q is the state of charge compensation coefficient, combined with the open circuit voltage U0 and the design current I de , through the multi-constrained safety current selector (9), the battery safety current I is determined s , combined with the terminal voltage U t , through the battery cell power state prediction module (10), the battery cell power SOP0 and SOP k , where SOP0 is the normal battery power base value, SOP k The base power value of slightly aged and severely aged battery cells;

[0013] S6. According to the normal battery cell power base value SOP0, obtain the normal battery system power state SOP i :SOP i =m×SOP0, where m is the number of normal battery cells in the parallel battery system;

[0014] S7, SOP k ,∑ΔSOP p and ∑ΔSOP q As input, the battery system aging power state compensation value SOP is obtained through the SOP summator (11). j , the compensation is calculated as: Where M is the total number of slightly aged and severely aged batteries, l is the number of slightly aged batteries, u is the number of severely aged batteries, and n=m+l+u, λ P,p With λ P,q is the power compensation coefficient;

[0015] S8, the normal battery system power state SOP iThe power state compensation value SOP of the battery system aging j Superposition to generate the corrected parallel battery system power state prediction value SOP f .

[0016] In step S1, the battery cell equivalent circuit model (1) consists of a controlled voltage source U0(SOC), an internal resistance R0(t), an electrochemical polarization resistance R s (t), electrochemical polarization capacitance C s (t) and concentration polarization resistance R l (t), concentration polarization capacitance C l (t) composition, the above parameters are related to the state of charge SOC, and the basic definition of SOC is SOC(t) = SOC0-∫I(t)dt / Q n , where SOC0 is the initial SOC value of the battery cell, Q n is the rated capacity. Battery cell parameters U0(SOC), R0(t), R s (t), C s (t), R l (t) and C l The relationship between (t) and SOC is:

[0017]

[0018] In the formula, a0~a5, b0~b5, c0~c2, d0~d2, e0~e2, f0~f2 are all model coefficients, which can be obtained by fitting the battery measurement data. According to the battery monomer equivalent circuit model after parameter fitting, the spatial state equation is obtained:

[0019]

[0020]

[0021] In the formula, ω g is the process noise, ν g To measure noise.

[0022] In step S2, the design process of the adaptive unscented Kalman filter (2) can be carried out according to the document (CN106597304A): (1) initializing the mean value, mean square error and noise statistics of the state variables; (2) obtaining sampling points and corresponding weights; (3) time updating of state estimation and mean square error; (4) calculating the gain matrix; (5) measurement updating of state estimation and mean square error.

[0023] In step S3, the fuzzy logic (3) design process is as follows:

[0024] (1) Obtain the absolute value of the capacity difference and the internal resistance difference |ΔQ through the absolute value modulei |,|ΔR i |, and use it as input to divide the battery into three fuzzy sets, namely small (S), medium (M) and large (L), representing different degrees of aging, and obtain the triangular membership function of the battery cell according to the fuzzy thresholds of capacity difference and internal resistance difference;

[0025] (2) According to different combinations of capacity difference and internal resistance difference, a fuzzy rule base is constructed:

[0026] Rule 1: If the capacity difference is small and the internal resistance difference is small, the battery cell is in a normal state.

[0027] …

[0028] Rule n: If the capacity difference is large and the internal resistance difference is large, the battery cell is in a serious aging state;

[0029] (3) For each rule, the minimum value method is used to calculate the membership of the capacity difference and the internal resistance difference in their respective fuzzy sets, and then the activation degree of each rule is obtained:

[0030] For rule 1, the activation is: μ 规则1 =min(μ small (ΔC),μ small (ΔR)),

[0031] …

[0032] For rule n, the activation is: μ 规则n =min(μ large (ΔC),μ large (ΔR));

[0033] (4) The aggregation process uses the maximum method to combine the results of all rules into an output fuzzy set, that is, the maximum membership degree of all rules is used for the output fuzzy set:

[0034] Membership for the "normal" state: μ normal =max(μ 规则1 ),

[0035] Membership degree for the “slightly aged” state: μ slight-aging =max(μ 规则2 ,μ 规则3 …),

[0036] Membership for the “severe aging” state: μ severe-aging =max(…μ 规则n-1 ,μ 规则n );

[0037] (5) Use the weighted average method in defuzzification to convert the result of fuzzy reasoning into a specific decision value, which is calculated as follows:

[0038]

[0039] The status value can be set to "normal = 1", "slightly aged = 2", "severely aged = 3", so as to obtain the aging classification of the battery.

[0040] In step S4, the battery aging model (6) is designed as follows:

[0041] (1) When the input is the battery cell capacity attenuation difference ΔQ p When the maximum capacity attenuation compensation of the battery cell SOC is calculated as: In the formula, ΔSOC p is the SOC compensation value of the slightly aged battery cell, SOC0 is the initial SOC value of the battery cell, Q max,0 is the initial maximum capacity;

[0042] (2) When the input is the battery cell internal resistance aging difference ΔR p When the battery terminal voltage U t The relationship between U and internal resistance R is: t =U0-I·R, U0 is the open circuit voltage of the battery, I is the working current of the battery, due to the increase of internal resistance, the terminal voltage U t The aging compensation of the internal resistance of the battery cell SOP is calculated as: ΔSOP p =ΔU t I p =(I i ΔR p )·I p =I p 2 ΔR p , where I p This is the operating current of a slightly aged battery cell.

[0043] In step S4, the random forest compensator (7) is composed of u sub-APSO-RFs connected in parallel, where u is a natural number less than or equal to n, and each sub-APSO-RF is designed as follows:

[0044] (1) Constructing the APSO algorithm structure

[0045] Randomly initialize the population, set the inertia weight range [w_min, w_max], and calculate the individual optimal value p i And the global optimal value g, dynamically adjust the inertia weight w according to the number of iterations t: And update the individual position x i (t+1) and speed v i (t+1), calculate the new individual optimal p i and the global optimal g, and process the boundary conditions at the same time. When the maximum number of iterations T is reached or the global optimal value converges, the algorithm stops and the final global optimal value g and the corresponding optimal solution are output;

[0046] (2) Constructing the RF algorithm structure

[0047] Import the training set and test set data, perform normalization, call the adaptive particle swarm optimization (APSO) algorithm, obtain the optimal number of trees n_trees = Best_pos(1), the optimal number of layers n_layer = Best_pos(2), use the optimized parameters to create a random forest model, obtain the trained random forest model, perform data prediction and denormalization, and finally use the true value and predicted value to evaluate the model, output various error indicators, and determine the compensation value.

[0048] In step S5, the compensation in the SOC summer (8) is calculated as:

[0049]

[0050] In the formula, λ C,p ,λ C,q is the state of charge compensation coefficient.

[0051] In step S5, the multi-constrained safety current selector (9) design process is as follows:

[0052] (1) Calculate the current under the open circuit voltage U0 constraint:

[0053]

[0054] In the formula, and are the discharge current and the charge current, SOC i,g is the estimated SOC value of the battery cell, Δt is the sampling interval, L is the sampling length, U s and U l are the electrochemical polarization voltage and concentration polarization voltage, R i is the internal resistance, R s and R l are the electrochemical polarization resistance and concentration polarization resistance respectively, τ1 and τ2 are the time constants, U t,min and U t,max are the minimum and maximum values ​​of the voltage safe operating range, Q n is the rated capacity of the battery cell, η is the Coulomb coefficient;

[0055] (2) Calculate the state of charge (SOC) i The current under constraint is In the formula, and are the discharge and charge currents, SOC min and SOC max are the minimum and maximum values ​​of the SOC safe operating range respectively;

[0056] (3) Calculate the design current I de The current under constraint is Where C dis and C cha are the battery discharge and charge rates respectively;

[0057] (4) Obtaining multi-constrained safety current I s for In the formula, is the maximum discharge current, is the minimum charging current.

[0058] In step S5, the maximum discharge power and minimum charge power of the battery cell can be obtained through the battery cell power state prediction module (10): Where U t,g is the battery cell terminal voltage.

[0059] In step S6, the sum of the maximum discharge power and the minimum charge power of the battery cell in the normal state SOP i for Where m is the number of normal battery cells in the parallel battery system.

[0060] In step S7, the SOP summer (11) is calculated as: Among them, SOP j is the sum of the powers of all aged batteries, M is the sum of the number of slightly aged and severely aged batteries, l is the number of slightly aged batteries, u is the number of severely aged batteries, and n=m+l+u, λ P,p With λ P,q is the power compensation coefficient.

[0061] In step S8, the corrected power state prediction value of the parallel battery system is: SOP f =SOP j +SOP i . BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 The structure diagram of the power state prediction method of the parallel battery system considering the aging degree;

[0063] Figure 2 This is the equivalent circuit model diagram of a battery cell. DETAILED DESCRIPTION

[0064] The present invention is further described in detail below in conjunction with specific examples, which are intended to explain the present invention rather than to limit it.

[0065] Step 1: Establish the battery cell equivalent circuit model and its spatial state equation

[0066] The parallel battery system is composed of n=4 battery cells connected in parallel. The rated voltage of the battery cell is 3.7V, the discharge cut-off voltage is 3V, and the rated capacity is 860mAh. The equivalent circuit model of the battery cell is a model containing a second-order RC circuit, which consists of a controlled voltage source U0(SOC), an internal resistance R0(t), an electrochemical polarization resistance R s (t), electrochemical polarization capacitance C s (t) and concentration polarization resistance R l (t), concentration polarization capacitance C l (t) composition, the terminal voltage is

[0067] U t (t) = U0[SOC(t)]-I(t)[R0(t)+R s (t) / R s (t)jωC s (t)+R l (t) / R l (t)jωC l (t)]

[0068] The relationship between the above parameters and SOC is:

[0069]

[0070] In the formula, the model coefficients a0~a5 obtained by fitting the battery experimental data are -0.915, -40.867, 3.632, 0.537, -0.499, 0.522, respectively; the values ​​of b0~b5 are -0.1463, -30.27, 0.1037, 0.0584, 0.1747, 0.1288, respectively; the values ​​of c0~c2 are 0.1063, -62.49, 0.0437, respectively; the values ​​of d0~d2 are -200, -138, 300, respectively; the values ​​of e0~e2 are 0.0712, -61.4, 0.0288, respectively; the values ​​of f0~f2 are -3083, -180, 5088, respectively;

[0071] According to the above mathematical model, the equivalent circuit model of the battery cell is established with SOC0, U s and U l is the state vector, and the battery space state equation is obtained as:

[0072]

[0073] In the formula, ω g is the process noise, ν g To measure noise.

[0074] Step 2: Get the battery cell state of charge SOC0

[0075] The state of charge SOC0 of the battery cell is obtained by an adaptive unscented Kalman filter. The specific steps can be carried out according to the document (CN106597304A): (1) initializing the mean value, mean square error and noise statistical information of the state variables; (2) obtaining sampling points and corresponding weights; (3) time updating of state estimation and mean square error; (4) calculating the gain matrix; (5) measurement updating of state estimation and mean square error.

[0076] Step 3: Detect aging inconsistencies of battery cells

[0077] (1) At the initial moment, the maximum available capacities of the four batteries are Q1=2900 ampere-seconds, Q2=2500 ampere-seconds, Q3=3000 ampere-seconds, and Q4=3100 ampere-seconds, respectively. The maximum available capacity difference of the batteries is obtained as follows: ΔQ1=200 ampere-seconds, ΔQ2=600 ampere-seconds, ΔQ3=100 ampere-seconds, and ΔQ4=0 ampere-seconds. The internal resistances of the four batteries are R1=0.1083Ω, R2=0.1115Ω, R3=0.1072Ω, and R4=0.1060Ω, respectively. The internal resistance difference of the batteries is ΔR1=0.0023Ω, ΔR2=0.0053Ω, ΔR3=0.0012Ω, and ΔR4=0Ω.

[0078] (2) Classification of battery cell aging

[0079] The difference between the maximum available capacity of the battery and the difference between the internal resistance of the battery are input into the fuzzy logic to classify the aging of the battery cells. It can be concluded that battery 1 is a slightly aged battery, battery 2 is a severely aged battery, and batteries 3 and 4 are normal batteries.

[0080] Step 4: Get the battery SOC compensation value

[0081] (1) Slightly aged battery SOC compensation

[0082] The maximum capacity attenuation compensation model of the battery cell SOC is: The input is the maximum available capacity difference of battery 1 ΔQ1, and the SOC compensation value ΔSOC of slightly aged battery 1 is obtained. p ;

[0083] (2) Severely aged battery SOC compensation

[0084] The difference between the operating current and the maximum available capacity of battery 2, ΔQ2, is input into APSO-RF to perform SOC compensation prediction, and the output is ΔSOC q .

[0085] Step 5: Get the battery SOP compensation value

[0086] (1) SOP compensation for slightly aged batteries

[0087] The internal resistance aging compensation model of the battery cell SOP is: ΔSOP p =ΔU t I p =(I i ΔR p )·I p =I p 2 ΔR p , the input is the internal resistance difference ΔR1 of battery 1, and the SOP compensation value ΔSOP of slightly aged battery 1 is obtained p ;

[0088] (2) SOP compensation for severely aged batteries

[0089] The operating current and internal resistance difference ΔR2 of battery 2 are input into APSO-RF to perform SOP compensation prediction, and the output is ΔSOP q .

[0090] Step 6: Battery system SOC compensation calculation

[0091] (1) For normal battery cells 3 and 4, the homodyne compensator outputs the SOC compensation value ΔSOC b =0, then SOC3=SOC4=SOC0;

[0092] (2) For slightly aged battery cell 1, its SOC compensation calculation is SOC1 = SOC0 + λ C,p ΔSOC p ;

[0093] (3) For severely aged battery cell 2, its SOC compensation calculation is SOC2 = SOC0 + λ C,q ΔSOC q .

[0094] Step 7: Obtain the battery system power status base values ​​SOP0 and SOP k

[0095] (1) Calculate the open circuit voltage U o The current under constraint is

[0096]

[0097] In the formula, and are the discharge current and the charge current, SOC i,g is the estimated SOC value of the battery cell, Δt is the sampling interval, L is the sampling length, U s and U l are the electrochemical polarization voltage and concentration polarization voltage, R i is the internal resistance, R s and R l are the electrochemical polarization resistance and concentration polarization resistance respectively, τ1 and τ2 are time constants, and the voltage safe operation range is [U t,min ,U t,max ]=[3,3.7],Q n is the rated capacity of the battery cell, η is the Coulomb coefficient;

[0098] (2) Calculate the state of charge (SOC) i The current under constraint is In the formula, and are the discharge and charge currents respectively, and the SOC safe operating range is [SOC min ,SOC max ]=[0.1,0.9];

[0099] (3) Calculate the design current I de The current under constraint is In the formula, the battery discharge rate C dis =5C, battery charging rate C cha =3C, then Get multi-constrained safety current I s for In the formula, is the maximum discharge current, is the minimum charging current;

[0100] (4) The total power state SOP of normal battery cells in the battery system i for

[0101]

[0102] In the formula, and are the sum of the maximum discharge power and minimum charge power of normal battery cells, and are the maximum discharge power base value and the minimum charge power base value of a normal battery cell, U t,g is the battery cell terminal voltage. The number of normal battery cells in the parallel battery system is 2. The power state base values ​​of slightly aged and severely aged battery cells in the battery system are

[0103] Step 7: Calculation of SOP compensation for aged cells

[0104] The power state base value SOP of slightly aged and severely aged battery cells in the battery system k and compensation value ΔSOP p , ΔSOP q Superposition, compensation calculation is: SOP j =SOP1+SOP2+λ P,p ΔSOP1+λ P,q ΔSOP2, power state compensation coefficient λ P,p =λ P,q .

[0105] Step 8: Obtain the power state prediction value SOP of the parallel battery system f

[0106] Normal battery power state and SOP i Aging Battery Power State SOP j The corrected power state prediction value of the parallel battery system is: SOP f =SOP i +SOP j .

[0107] Finally, it should be noted that the above description is only an example of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for predicting the power state of a parallel battery system considering the degree of aging, wherein the parallel battery system is composed of n branches connected in parallel, each branch being a battery cell, wherein n is a natural number greater than 1, and the method comprises the following steps: S1. Establish a second-order RC battery cell equivalent circuit model based on experimental data and obtain the battery cell spatial state equation; S2, taking the battery cell spatial state equation as input, combining the battery cell measured voltage value and the battery cell measured current value, and obtaining the battery cell state of charge initial value SOC0 through an adaptive unscented Kalman filter; S3, taking the maximum available capacity of each branch battery cell and the internal resistance of each branch battery cell as input, the aging inconsistency classification module classifies each branch battery cell into three categories: normal, slightly aged, and severely aged. The specific classification operation is as follows: In the aging inconsistency classification module, the maximum available capacity of each branch battery cell is subtracted from the maximum available capacity of a normal healthy battery to obtain the battery cell capacity deviation value ΔQ1~ΔQ n At the same time, the internal resistance of each branch battery cell is subtracted from the normal healthy battery internal resistance to obtain the battery cell internal resistance deviation value ΔR1~ΔR n , then combined with fuzzy logic, the aging inconsistency classifier classifies each battery cell into three categories: normal, slightly aged, and severely aged; S4. For the three different types of aging battery cells, the difference compensation module is used to perform SOC compensation and SOP compensation. The specific compensation operation is as follows: For normal battery cells, the output SOC compensation value ΔSOC is obtained through the corresponding zero-difference compensator. b (b∈[1,n]) is 0; for slightly aged battery cells, the battery cell capacity difference ΔQ p Difference ΔR from the internal resistance of the battery cell p As input, the SOC compensation value ΔSOC is obtained through the battery aging model p (p∈[1,n]) and SOP compensation value ∑ΔSOP p (p∈[1,n]); for severely aged battery cells, the battery cell capacity difference ΔQ q 、Current I q Difference ΔR from the internal resistance of the battery cell q 、Current I q As input, the SOC compensation value ΔSOC is obtained through the random forest compensator. q (q∈[1,n]) and SOP compensation value ∑ΔSOP q (q∈[1,n]); S5, SOC0, ΔSOC b , ΔSOC p and ΔSOC q As input, the SOC value of each battery cell after compensation is obtained through the SOC summator. i , the specific calculation is: In the formula, λ C,p ,λ C,q is the state of charge compensation coefficient, combined with the open circuit voltage U0 and the design current I de , the battery safety current I is determined by the multi-constrained safety current selector s , combined with the terminal voltage U t , the battery power state prediction module obtains the battery power SOP0 and SOP k , where SOP0 is the normal battery power base value, SOP k The base power value of slightly aged and severely aged battery cells; S6. According to the normal battery cell power base value SOP0, obtain the normal battery system power state SOP i :SOP i =m×SOP0, where m is the number of normal battery cells in the parallel battery system; S7, SOP k ,∑ΔSOP p and ∑ΔSOP q As input, the battery system aging power state compensation value SOP is obtained through the SOP summator. j , the compensation is calculated as: Where M is the total number of slightly aged and severely aged batteries, l is the number of slightly aged batteries, u is the number of severely aged batteries, and n=m+l+u, λ P,p With λ P,q is the power compensation coefficient; S8, the normal battery system power state SOP i The power state compensation value SOP of the battery system aging j Superposition to generate the corrected parallel battery system power state prediction value SOP f .

2. A power state prediction method for a parallel battery system considering aging degree according to claim 1, characterized in that: The design of the aging classification module is as follows: (1) Obtain the absolute value of the capacity difference and the internal resistance difference |ΔQ through the absolute value module i |,|ΔR i |, and use it as input to divide the battery into three fuzzy sets, namely small (S), medium (M) and large (L), representing different degrees of aging, and obtain the triangular membership function of the battery cell according to the fuzzy thresholds of capacity difference and internal resistance difference; (2) According to different combinations of capacity difference and internal resistance difference, a fuzzy rule base is constructed: Rule 1: If the capacity difference is small and the internal resistance difference is small, the battery cell is in a normal state. … Rule n: If the capacity difference is large and the internal resistance difference is large, the battery cell is in a serious aging state; (3) For each rule, the minimum value method is used to calculate the membership of the capacity difference and the internal resistance difference in their respective fuzzy sets, and then the activation degree of each rule is obtained: For rule 1, the activation is: μ 规则1 =min(μ small (ΔC),μ small (ΔR)), … For rule n, the activation is: μ 规则n =min(μ large (ΔC),μ large (ΔR)); (4) The aggregation process uses the maximum method to combine the results of all rules into an output fuzzy set, that is, the maximum membership degree of all rules is used for the output fuzzy set: Membership for the "normal" state: μ normal =max(μ 规则1 ), Membership for the "slightly aged" state: μ slight-aging =max(μ 规则2 ,μ 规则3 …), Membership for the "severe aging" state: μ severe-aging =max(…μ 规则n-1 ,μ 规则n ); (5) Use the weighted average method in defuzzification to convert the result of fuzzy reasoning into a specific decision value, which is calculated as follows: The status value can be set to "normal = 1", "slightly aged = 2", "severely aged = 3", thereby obtaining the aging classification of the battery.

3. The power state prediction method of a parallel battery system considering aging degree according to claim 1, characterized in that: The difference compensation module obtains the SOC compensation value ΔSOC through the battery aging model under the slight aging state p and ΔSOP of SOP compensation value p The design is as follows: (1) When the input is the battery cell capacity attenuation difference ΔQ p When the maximum capacity attenuation compensation value of the battery cell SOC is: In the formula, ΔSOC p is the SOC compensation value of the slightly aged battery cell, SOC0 is the initial SOC value of the battery cell, Q max,0 is the initial maximum capacity; (2) When the input is the battery cell internal resistance aging difference ΔR p When the battery terminal voltage U t The relationship between U and internal resistance R is: t =U0-I·R, I is the working current of the battery. Due to the increase of internal resistance, the terminal voltage U t Will decrease, resulting in a decrease in output power, so the compensation goal is to estimate the terminal voltage difference, and then derive the power compensation amount. The internal resistance aging compensation value of the battery cell SOP is: ΔSOP p =ΔU t I p =(I i ΔR p )·I p =I p 2 ΔR p , where I p This is the operating current of a slightly aged battery cell.

4. The power state prediction method of a parallel battery system considering aging degree according to claim 1, characterized in that: The difference compensation module obtains the SOC compensation value ΔSOC through the random forest compensator under the severe aging state q and ΔSOP of SOP compensation value q , the random forest compensator is composed of u sub-APSO-RFs in parallel, where u is a natural number less than or equal to n, and each sub-APSO-RF is designed as follows: (1) Constructing the APSO algorithm structure Randomly initialize the population, set the inertia weight range [w_min, w_max], and calculate the individual optimal value p i and the global optimal value g, dynamically adjust the inertia weight according to the number of iterations t And update the individual position x i (t+1) and speed v i (t+1), calculate the new individual optimal p i and the global optimal g, and process the boundary conditions at the same time. When the maximum number of iterations T is reached or the global optimal value converges, the algorithm stops and the final global optimal value g and the corresponding optimal solution are output; (2) Constructing the RF algorithm structure Import the training set and test set data, perform normalization, call the adaptive particle swarm optimization algorithm (APSO), obtain the optimal number of trees n_trees = Best_pos(1), the optimal number of layers n_layer = Best_pos(2), use the optimized parameters to create a random forest model, obtain the trained random forest model, perform data prediction and denormalization, and finally use the true value and predicted value to evaluate the model, output various error indicators, and determine the SOP compensation value ∑ΔSOP q .

5. The method for predicting power status of a parallel battery system considering aging degree according to claim 1, characterized in that: Normal battery system power state SOP i The calculation of is as follows: In the formula, and are the maximum discharge power and minimum charge power of the normal battery system, and are the maximum discharge power base value and the minimum charge power base value of a normal battery cell, U t,g is the battery cell terminal voltage, is the maximum discharge current, is the minimum charging current, and m is the number of normal battery cells in the parallel battery system.

6. The method for predicting power status of a parallel battery system considering aging degree according to claim 1, characterized in that: In each branch of the parallel system, x battery cells may be connected in series, where x is a natural number greater than 0.

Images