Thermal safety-oriented lithium-ion battery pack layered active equalization control method and system

Abstract

The application provides a thermal safety-oriented lithium ion battery pack layered active equalization control method and system, relates to the technical fields of lithium ion battery management systems, battery active equalization control and battery thermal safety management. On the basis of adaptive static-difference-free layered model predictive control, the application fuses fuzzy thermal safety supervision, inter-group CUK thermal perception routing, CUK current derating and execution perception input reconstruction mechanism to form thermal safety fuzzy supervision adaptive static-difference-free layered model predictive control. On the basis of realizing state of charge convergence, the application comprehensively considers single temperature state, inter-group energy transmission path thermal risk and actual execution deviation to inhibit local thermal concentration, reduce temperature difference accumulation and improve battery pack temperature consistency.

Description

Technical Field

[0001] This invention relates to the fields of lithium-ion battery management systems, active battery balancing control, and battery thermal safety management, specifically to a method and system for active balancing control of lithium-ion battery modules for thermal safety. Background Technology

[0002] Lithium-ion batteries, with their advantages of high energy density, long cycle life, and low self-discharge rate, are widely used in electric vehicles, energy storage systems, portable power supplies, and unmanned equipment. In practical applications, individual cells are typically connected in series, parallel, or series-parallel configurations to form battery packs to meet the demands for high voltage, large capacity, and high power output. However, due to factors such as manufacturing differences, capacity decay, internal resistance variations, differences in connection resistance, uneven heat dissipation, ambient temperature variations, and differences in aging rates, inconsistencies in state of charge, voltage, and temperature are unavoidable within battery packs. Inconsistent state of charge reduces the usable capacity of the battery pack, limiting the overall charge and discharge capacity to the cells that first reach the charging or discharging cutoff condition; inconsistent temperature further affects the internal resistance, reaction rate, and aging rate of individual cells, causing the inconsistencies within the battery pack to continue to increase.

[0003] Existing battery balancing technologies mainly include passive balancing and active balancing. Passive balancing typically dissipates excess energy from cells with high state of charge through bypass resistors. While structurally simple, it suffers from energy waste and localized heat generation. Active balancing, on the other hand, transfers energy between cells or modules using capacitors, inductors, transformers, or DC-DC converters, offering higher energy efficiency compared to passive balancing. For battery packs containing many cells connected in series, single-layer active balancing structures often suffer from long energy transmission paths, numerous switches, reduced balancing efficiency, and increased control complexity. Therefore, layered active balancing structures are gaining increasing attention. Layered active balancing typically includes intra-group balancing and inter-group balancing. Intra-group balancing is used to achieve localized energy regulation between cells within the same battery sub-group, while inter-group balancing is used to achieve energy coordination between different battery sub-groups.

[0004] Most existing active equalization control methods primarily aim to achieve rapid state-of-charge (SOC) uniformity, voltage uniformity, or improved equalization efficiency, while neglecting the temperature distribution differences, localized heat concentrations, and thermal risks associated with inter-group energy transfer paths that arise during the equalization process. Particularly during inter-group active equalization, if a battery subgroup with a higher temperature continues to participate in large current transfer as a source or target group, it may continue to bear a significant thermal burden, further exacerbating temperature inconsistencies. Furthermore, while model predictive control can continuously optimize the equalization input under constraints, in actual equalization systems, the ideal equalization input output by the optimizer still needs to undergo pulse width modulation mapping, inter-group energy transfer path selection, current limiting, and execution layer allocation. Deviations may exist between the actual execution input and the optimization layer input. If the predictive controller still uses the ideal input for the next sampling cycle prediction, the consistency between the model prediction process and the actual execution process will be reduced.

[0005] Therefore, it is necessary to propose a layered active equalization control method and system for lithium-ion battery packs with thermal safety in mind. Based on achieving state of charge convergence, this method comprehensively considers the temperature state of individual cells, the thermal risks of inter-pack energy transfer paths, and actual execution deviations to suppress local heat concentration, reduce temperature difference accumulation, and improve the temperature consistency of the battery pack. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for active balancing control of lithium-ion battery modules for thermal safety, which addresses the problems of existing active balancing methods that mainly focus on rapid consistency of state of charge, do not adequately consider temperature inconsistencies and thermal risks in inter-module energy transfer paths, and have discrepancies between optimized inputs and actual execution inputs.

[0007] To achieve the above objectives, in a first aspect, the present invention provides an active equalization control method for lithium-ion battery components oriented towards thermal safety, comprising: S1. Construct the zero-mean electro-thermal state vector for the current sampling period based on the state of charge (SOC) and temperature of each individual cell; S2. Introduce group-level SOC and temperature equivalent perturbation to form a perturbation vector; combine the perturbation vector with the zero-mean electro-thermal state vector to form an augmented state and construct an augmented prediction model; S3. Determine the dynamic temperature weights, solve the hierarchical model predictive control optimization problem based on the augmented prediction model, augmented state and dynamic temperature weights, and determine the intra-group equilibrium request and inter-group equilibrium request; determine the current inter-group energy transfer path, the current source group and the current target group based on the inter-group equilibrium request, and determine the thermal risk of the candidate path corresponding to the current inter-group energy transfer path as the thermal risk of the current path. S4. Introduce a fuzzy thermal safety monitor to obtain the inter-group current scaling factor and derate the original inter-group request current corresponding to the current inter-group energy transfer path, and generate intra-group execution instructions and inter-group execution instructions. S5. Constrain the intra-group and inter-group execution instructions, and after the intra-group and inter-group execution instructions are actually executed, combine the augmented prediction model to obtain the zero-mean electro-thermal state vector for the next sampling period. S6. Repeat S1 to S5 to achieve active balance control.

[0008] According to the present invention, a lithium-ion battery component layer active equalization control method for thermal safety is provided, wherein step S1 specifically includes: In the kth sampling period, the SOC and temperature of each cell in the lithium-ion battery pack are obtained. Based on the SOC and temperature of each cell, the average SOC of the battery pack, the average temperature of the battery pack, the average SOC of each battery subgroup, the average temperature of each battery subgroup, the SOC range and the global temperature difference are determined, and a zero-mean electro-thermal state vector is constructed. The average SOC and average temperature of the battery pack are expressed as follows:

[0009] The average SOC and average temperature of each battery sub-group are expressed as follows:

[0010] The SOC range and global temperature difference are respectively expressed as:

[0011] The zero-mean electro-thermal state vector is represented as:

[0012] in, Indicates the first Zero-mean electro-thermal state vector for each sampling period; For sampling period index; For single-entity indexes; and The first The sampling period The SOC and temperature of the monomer; and They are respectively from A column vector consisting of the SOC and temperature of the monomer; The total number of individual units; For battery subgroup indexes; For the first Individual index set of each battery subgroup The number of single entities in the single entity index set; for A column vector of all 1s.

[0013] According to the active equalization control method for lithium-ion battery components oriented towards thermal safety provided by the present invention, step S2 specifically includes: When the k-th sampling period is the initial sampling period, the group-level SOC equivalent perturbation and group-level temperature equivalent perturbation of each battery subgroup are initialized to zero. When the k-th sampling period is not the initial sampling period, the execution-aware prediction state generated after reconstructing the equivalent execution input based on the actual intra-group execution instructions and the actual inter-group execution instructions in the previous sampling period is obtained. Based on the state prediction error between the zero-mean electro-thermal state vector and the execution-aware prediction state, the group-level SOC equivalent perturbation and group-level temperature equivalent perturbation of each battery subgroup are updated. Based on the battery group average SOC, battery group average temperature and SOC range obtained in the k-th sampling period, the SOC input gain and equivalent thermal response gain are adaptively corrected, and the electro-thermal coupling input matrix is ​​updated based on the corrected SOC input gain and equivalent thermal response gain. The zero-mean electro-thermal state vector is combined with the group-level SOC equivalent perturbation and group-level temperature equivalent perturbation determined in the current sampling period to form an augmented state. An augmented prediction model is constructed based on the augmented state and the updated electro-thermal coupling input matrix. In non-initial sampling periods, the execution perception prediction state of the previous sampling period is represented as:

[0014]

[0015] The state prediction error between the zero-mean electro-thermal state vector and the execution-aware predicted state is expressed as:

[0016] No. The SOC prediction error and temperature prediction error of each battery subgroup are expressed as follows:

[0017] The group-level SOC equivalent perturbation and the group-level temperature equivalent perturbation are updated recursively according to the following formulas:

[0018]

[0019] The disturbance vector consists of the equivalent SOC disturbance and the equivalent temperature disturbance of each battery sub-group:

[0020] Where m is the number of battery subgroups; Extract the matrix for the physical state; For dimension The execution perception augmented prediction state; To extract from the execution-aware augmented prediction state, the dimension is The zero-mean electro-thermal state vector is used as the execution sensing prediction state; This is the state prediction error vector; and The first The SOC prediction error component and temperature prediction error component corresponding to each unit; and The first SOC prediction error and temperature prediction error for each battery sub-group; and The first Equivalent SOC and equivalent temperature disturbances of each battery sub-group; and These are the exponential smoothing coefficients for group-level SOC perturbation estimation and group-level temperature perturbation estimation, respectively, both of which are greater than 0 and less than 1.

[0021] According to the active equalization control method for lithium-ion battery components oriented towards thermal safety provided by the present invention, the SOC input gain is expressed as:

[0022] The equivalent thermal response gain is expressed as:

[0023] in, and The first SOC input gain and equivalent thermal response gain for each sampling period; This is a correction factor for the equalization efficiency determined based on the current average SOC and average temperature of the battery pack. and These are the nominal SOC input gain and the nominal equivalent thermal response gain, respectively. For reference temperature; This is the correction factor for the average temperature on the equivalent thermal response gain. This is the correction factor for the SOC range on the equivalent thermal response gain; The augmented state is represented as follows:

[0024] The augmented prediction model is expressed as follows:

[0025] in, For augmented state transition matrix; This is an augmented input matrix that includes the electro-thermal coupling input matrix; This is the non-negative control input after positive and negative decomposition.

[0026] According to the active equalization control method for lithium-ion battery components oriented towards thermal safety provided by the present invention, step S3 specifically includes: The dynamic temperature weight is determined based on the SOC range; the hierarchical model predictive control optimization problem is solved based on the augmented prediction model, the augmented state, and the dynamic temperature weight to obtain the optimal control input sequence; the first control input of the current sampling period is extracted from the optimal control input sequence; and the intra-group equalization request corresponding to each intra-group equalization branch in the intra-group active equalization circuit and the inter-group equalization request corresponding to the inter-group active equalization circuit are determined based on the SOC deviation, temperature deviation, and the temperature state prediction result obtained by the augmented prediction model during the solution of the hierarchical model predictive control optimization problem. Candidate source groups, candidate target groups, and inter-group energy transfer paths are determined based on the energy output and energy receiving directions represented by the inter-group equalization requests. The thermal risk of each inter-group energy transfer path is determined based on the average temperature of the candidate source groups, the average temperature of the candidate target groups, the average temperature of the battery pack, and the global temperature difference. A candidate path score is determined based on the inter-group equalization requests corresponding to each inter-group energy transfer path and the candidate path thermal risk. Based on the candidate path scores of each inter-group energy transfer path, the current inter-group energy transfer path, the current source group, and the current target group are determined from the feasible inter-group energy transfer paths, and the thermal risk of the candidate path corresponding to the current inter-group energy transfer path is determined as the current path thermal risk. The dynamic temperature weight is expressed as:

[0027] The objective function of the hierarchical model predictive control optimization problem consists of the stage cost and the terminal cost in the prediction time domain:

[0028] Among them, the augmented prediction model in the first... The zero-mean electro-thermal state vectors predicted in each prediction step are used to calculate the stage cost, which is expressed as:

[0029] The terminal cost is expressed as:

[0030] The control input satisfies:

[0031]

[0032]

[0033] The optimal control input sequence is obtained by solving the hierarchical model predictive control optimization problem:

[0034] in, Base temperature weighting; For prediction in the time domain; For the prediction step index; In the first The prediction obtained in the sampling period is the first Zero-mean electro-thermal state vector for each sampling period; For the first The stage cost of each prediction step; For the cost of the terminal; The state weight matrix includes SOC deviation weight, dynamic temperature deviation weight, and inter-group average SOC difference weight. To control the input weight matrix; This is the terminal weight matrix; and These are the lower limit and upper limit of the control input, respectively; This is the coefficient matrix of the intra-group branch coupling inequality constraints. This is the upper bound vector for the corresponding inequality constraints; This is the coefficient matrix of the inter-group energy coordination equation constraints. This is the corresponding equality constraint target vector; It is a candidate control input sequence composed of the positive and negative decomposition control inputs of each prediction step in the prediction time domain. For the first The optimal control input sequence is obtained by solving for each sampling period; the equivalent equalization input within each group acts simultaneously on the SOC subsystem and the temperature subsystem through the electro-thermal coupling input matrix, so that the control input corresponding to the equalization branch within each group is jointly determined in the same optimization based on the SOC deviation, temperature deviation, and the temperature state prediction result obtained from the augmented prediction model; from Extract and execute only the first control input of the current sampling period, and directly determine the intra-group equivalent equalization input of each group in the first control input as the intra-group equalization request of the corresponding intra-group equalization branch, and determine the inter-group equivalent equalization input in the first control input as the inter-group equalization request.

[0035] According to the present invention, an active equalization control method for lithium-ion battery components oriented towards thermal safety is provided. The battery subgroup is a candidate source group and the first... When a cell subgroup is a candidate target group, the thermal risk of the candidate path for energy transfer between candidate groups is expressed as follows:

[0036] The candidate path score is represented as follows:

[0037] Based on the candidate path scores of each candidate inter-group energy transfer path, the current inter-group energy transfer path is determined from the feasible candidate inter-group energy transfer paths, with priority given to the inter-group energy transfer path with higher candidate path scores; the thermal risk of the candidate path corresponding to the current inter-group energy transfer path is determined as the thermal risk of the current path; wherein... and These are the candidate source group index and the candidate target group index, respectively. and These are the average temperatures of the candidate source group and the average temperatures of the candidate target group, respectively. This represents the average temperature of the battery pack. For global temperature difference; To prevent positive numbers with excessively small denominators; For from the first The battery subgroup to the first Thermal risk of candidate pathways for energy transfer in individual battery subgroups; The non-negative inter-group equalization request amplitude corresponding to the inter-group energy transfer path; This is a preset non-negative temperature penalty coefficient; The candidate paths are scored; the candidate source group and candidate target group corresponding to the current inter-group energy transfer path are the current source group and the current target group, respectively; the thermal risk of the current path is denoted as... .

[0038] According to the active equalization control method for lithium-ion battery components oriented towards thermal safety provided by the present invention, step S4 specifically includes: The hottest group flag is determined based on whether the current inter-group energy transfer path passes through the battery subgroup with the highest average temperature. An inter-group SOC difference index, characterizing the degree of inconsistency in average SOC between different battery subgroups, is determined based on the average SOC of each battery subgroup. The global temperature difference, the current path thermal risk, the hottest group flag, and the inter-group SOC difference index are input into a fuzzy thermal safety monitor to obtain an inter-group current scaling factor. The original inter-group request current corresponding to the current inter-group energy transfer path is derated based on the inter-group current scaling factor and the inter-group current upper limit to obtain a derated inter-group execution request current. Intra-group execution instructions are generated based on the intra-group balancing requests corresponding to each intra-group balancing branch, and inter-group execution instructions are also generated based on the current inter-group energy transfer path, the current source group, the current target group, and the derated inter-group execution request current. The hottest group flag is set to 1 when the current inter-group energy transfer path passes through the battery subgroup with the highest average temperature, and 0 otherwise; the inter-group SOC difference index is expressed as:

[0039] The inter-group current scaling factor output by the fuzzy thermal safety monitor is expressed as:

[0040] The reduced inter-group execution request current is expressed as follows:

[0041] in, This represents the difference between the maximum and minimum average SOC values ​​of each battery subgroup. Thermal risk of the candidate path corresponding to the current inter-group energy transfer path; This is the marker value for the hottest group; This is the fuzzy inference mapping corresponding to the fuzzy thermal safety monitor; This is the inter-group current scaling factor, which takes a value greater than 0 and not greater than 1. The original inter-group requested current corresponds to the current inter-group power transfer path, and the original inter-group requested current is based on the non-negative inter-group equalization request amplitude corresponding to the current inter-group power transfer path. Sure; This is the upper limit of the inter-group current. The derated inter-group execution request current is used; the intra-group balancing requests corresponding to each intra-group balancing branch are mapped to the corresponding intra-group execution instructions, the derated inter-group execution request current is mapped to the execution control quantity of the inter-group active balancing circuit, and the inter-group execution instructions are generated according to the current inter-group power transfer path, the current source group, the current target group, and the execution control quantity; when the inter-group active balancing circuit is an isolated CUK balancing circuit, the execution control quantity includes the CUK main switch duty cycle.

[0042] According to the active equalization control method for lithium-ion battery components oriented towards thermal safety provided by the present invention, step S5 specifically includes: Based on the allowable control boundaries of the intra-group active balancing circuit and the inter-group active balancing circuit, the intra-group execution instructions and the inter-group execution instructions are constrained to obtain the execution layer constraint processing result. After the intra-group execution instructions and the inter-group execution instructions are actually executed, the execution-aware equivalent input is reconstructed based on the actually executed intra-group execution instructions, the actually executed inter-group execution instructions, and the execution layer constraint processing result to obtain the equivalent execution input. The equivalent execution input is substituted into the augmented prediction model to obtain the execution-aware augmented prediction state of the (k+1)th sampling period, and the zero-mean electro-thermal state vector is extracted from the execution-aware augmented prediction state as the execution-aware prediction state of the (k+1)th sampling period. The equivalent execution input is:

[0043] Substituting the equivalent execution input into the augmented prediction model, the first... The execution-aware augmentation prediction state for each sampling period is represented as follows:

[0044] in, It consists of execution information composed of the intra-group execution instructions actually executed, the inter-group execution instructions actually executed, and the results of execution layer constraint processing; This is an execution-aware mapping that maps the execution information to equivalent execution inputs; In accordance with The equivalent execution input obtained by reconstructing the same positive and negative decomposition structure has the same dimension as... same; To perform the perception augmentation prediction state; extract the zero-mean electro-thermal state vector from the perception augmentation prediction state using the physical state extraction matrix, and use it as the perception prediction state; in the... In each sampling period, based on the state prediction error between the measured zero-mean electro-thermal state vector and the executed sensing prediction state, the group-level SOC equivalent perturbation and group-level temperature equivalent perturbation of each battery subgroup are updated; based on the first sampling period... The average SOC, average temperature, and SOC range of the battery pack are determined by re-collecting the SOC and temperature of each individual cell in each sampling period, and the SOC input gain and equivalent thermal response gain are recalculated.

[0045] Secondly, the present invention provides a layered active balancing control system for lithium-ion battery packs oriented towards thermal safety, comprising a battery pack, an intra-pack active balancing circuit, an inter-pack active balancing circuit, a state acquisition module, and a controller; the battery pack comprises multiple battery sub-packs, each battery sub-pack comprising multiple cells connected in series; the intra-pack active balancing circuit is disposed within each battery sub-pack and includes multiple intra-pack balancing branches; the inter-pack active balancing circuit is connected between different battery sub-packs and can form multiple selectable inter-pack energy transfer paths between different battery sub-packs; The state acquisition module is used to acquire the state of charge (SOC) and temperature of each cell in the battery pack in each sampling cycle; the state acquisition module, the intra-pack active balancing circuit and the inter-pack active balancing circuit are all connected to the controller; the controller is used to implement the method of the first aspect.

[0046] According to the present invention, a lithium-ion battery pack stratified active balancing control system for thermal safety is provided. The intra-packet active balancing circuit is an intra-packet improved Buck-Boost balancing circuit disposed within each battery sub-packet. The inter-packet active balancing circuit is an isolated CUK balancing circuit connected between different battery sub-packets via a gating matrix. The intra-packet execution instructions include pulse width modulation instructions corresponding to each intra-packet improved Buck-Boost balancing branch. The inter-packet execution instructions include the current source group identifier, the current target group identifier, and the CUK main switch duty cycle for controlling the gating matrix and the isolated CUK balancing circuit.

[0047] Compared with the prior art, the present invention has at least the following technical effects: 1. This invention adopts a hierarchical active equalization structure that combines an intra-group improved Buck-Boost equalization circuit and an inter-group gating matrix isolation CUK equalization circuit, which can take into account both individual-level local equalization and group-level energy coordination.

[0048] 2. This invention introduces an electro-thermal coupling prediction model into the equalization control, which enables the controller to consider temperature changes while predicting the SOC convergence process. This helps to improve the problem of insufficient consideration of temperature effects when equalization control is based solely on SOC differences.

[0049] 3. This invention uses a thermal risk assessment and fuzzy thermal safety monitoring mechanism for inter-group energy transfer paths to adjust the inter-group CUK balance request for thermal safety, which can reduce the probability that high-temperature subgroups or high-thermal-risk paths will continue to undertake strong inter-group energy transfer actions, thereby suppressing local heat concentration.

[0050] 4. This invention achieves a dynamic trade-off between adjusting the SOC difference between battery packs and suppressing temperature through the CUK current scaling and derating mechanism, which helps to reduce the accumulation of temperature difference and improve the temperature consistency of the battery pack.

[0051] 5. By performing sensing input reconstruction, this invention converts the actual issued PWM instructions and CUK execution instructions into equivalent execution inputs and feeds them back to the next sampling cycle, which helps to reduce the deviation between the optimized input and the actual execution input and improve the consistency between the predictive control process and the actual equalization process. Attached Figure Description

[0052] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0053] In the attached diagram: Figure 1 This is a schematic diagram of the active balancing system for lithium-ion battery components in this invention. Figure 2 This is a schematic diagram of the improved Buck-Boost equalization circuit within the present invention group; Figure 3 This is a schematic diagram of the inter-group gating matrix isolated CUK equalization circuit of the present invention; Figure 4 This is a schematic diagram of the working principle of the improved Buck-Boost equalization circuit within the present invention group, wherein, Figure 4 (a) is a schematic diagram of B1 charging L1 via Q1. Figure 4 (b) is a schematic diagram of L1 discharging into B2 via Q2. Figure 4 (c) is a schematic diagram of the outer layer energy transfer from the B1 and B2 sides to the B3 and B4 sides via Q5, L3 and Q6; Figure 5 This is a schematic diagram illustrating the working principle of the inter-group isolated CUK equalization circuit of the present invention, wherein... Figure 5 (a) is a schematic diagram of the energy transfer path when the switch Q is turned on. Figure 5 (b) is a schematic diagram of the energy transfer path when the switch Q is turned off; Figure 6 This is a flowchart of the TSF-AOF-HMPC control method of the present invention; Figure 7 This is a schematic diagram of the membership function of the TSF fuzzy thermal safety monitor of the present invention; Figure 8 This is a graph showing the SOC variation curves of different control strategies under the static equilibrium condition of the present invention. Figure 8 (a) represents the MPC control strategy. Figure 8 (b) represents the AOF-HMPC control strategy. Figure 8(c) represents the TSF-AOF-HMPC control strategy of this invention; Figure 9 This is a temperature change curve diagram of different control strategies under the static equilibrium condition of the present invention, wherein, Figure 9 (a) represents the MPC control strategy. Figure 9 (b) represents the AOF-HMPC control strategy. Figure 9 (c) represents the TSF-AOF-HMPC control strategy of this invention. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0055] The following detailed description of some embodiments of the present invention will be provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0056] Please see Figure 1 This invention provides a thermal safety-oriented active balancing control method for lithium-ion battery pack layers. This method, based on adaptive zero-steady-state-error hierarchical model predictive control, integrates fuzzy thermal safety supervision, inter-packet CUK thermal sensing routing, CUK current derating, and an execution sensing input reconstruction mechanism to form Thermal-Safe Fuzzy-supervised Adaptive Offset-Free Hierarchical Model Predictive Control (TSF-AOF-HMPC). Another embodiment of this invention provides a thermal safety-oriented active balancing control system for lithium-ion battery pack layers, used to implement the balancing control method of the aforementioned embodiments.

[0057] The lithium-ion battery pack stratified active balancing system includes a battery pack, an intra-pack active balancing circuit, an inter-pack active balancing circuit, a state acquisition module, and a controller. The battery pack comprises multiple battery sub-packs, each sub-pack containing multiple cells connected in series. The intra-pack active balancing circuit is located within each sub-pack and includes multiple intra-pack active balancing branches. The inter-pack active balancing circuit connects different sub-packs and can form multiple selectable inter-pack energy transfer paths between them. The state acquisition module collects the state of charge (SOC) and temperature of each cell in the battery pack during each sampling period. The state acquisition module, intra-pack active balancing circuit, and inter-pack active balancing circuit are all connected to the controller. The controller implements the aforementioned thermal safety-oriented lithium-ion battery pack stratified active balancing control method. Furthermore, the intra-group active balancing circuit is an intra-group improved Buck-Boost balancing circuit set inside each battery sub-group, and the inter-group active balancing circuit is an isolated CUK balancing circuit connected between different battery sub-groups through a gating matrix; the intra-group execution instructions include pulse width modulation instructions corresponding to each intra-group improved Buck-Boost balancing branch, and the inter-group execution instructions include the current source group identifier, the current target group identifier, and the CUK main switch duty cycle used to control the gating matrix and the isolated CUK balancing circuit.

[0058] This active equalization control method for lithium-ion battery components aimed at thermal safety includes the following steps: S1. Construct the zero-mean electro-thermal state vector for the current sampling period based on the SOC and temperature of each individual cell; Specifically, S1 includes: In the kth sampling period, the SOC and temperature of each cell in the lithium-ion battery pack are obtained. Based on the SOC and temperature of each cell, the average SOC of the battery pack, the average temperature of the battery pack, the average SOC of each battery subgroup, the average temperature of each battery subgroup, the SOC range and the global temperature difference are determined, and a zero-mean electro-thermal state vector is constructed. The average SOC and average temperature of the battery pack are expressed as follows:

[0059] The average SOC and average temperature of each battery sub-group are expressed as follows:

[0060] The SOC range and global temperature difference are respectively expressed as:

[0061] The zero-mean electro-thermal state vector is represented as:

[0062] Among them, lithium-ion battery packs include Section tandem monomers and divided into One battery sub-pack; Indicates the first The zero-mean electro-thermal state vector for each sampling period, with dimension [missing value]. ; For sampling period index; For single-entity indexes; and The first The sampling period The SOC and temperature of the monomer; and They are respectively from A column vector consisting of the SOC and temperature of the monomer; The total number of individual units; For battery subgroup indexes; For the first Individual index set of each battery subgroup The number of single entities in the single entity index set; for A column vector of all 1s.

[0063] S2. Introduce group-level SOC and temperature equivalent perturbation to form a perturbation vector; combine the perturbation vector with the zero-mean electro-thermal state vector to form an augmented state and construct an augmented prediction model; Specifically, S2 includes: When the k-th sampling period is the initial sampling period, the group-level SOC equivalent perturbation and group-level temperature equivalent perturbation of each battery subgroup are initialized to zero. When the k-th sampling period is not the initial sampling period, the execution-aware prediction state generated after reconstructing the equivalent execution input based on the actual intra-group execution instructions and the actual inter-group execution instructions in the previous sampling period is obtained. Based on the state prediction error between the zero-mean electro-thermal state vector and the execution-aware prediction state, the group-level SOC equivalent perturbation and group-level temperature equivalent perturbation of each battery subgroup are updated. Based on the battery group average SOC, battery group average temperature and SOC range obtained in the k-th sampling period, the SOC input gain and equivalent thermal response gain are adaptively corrected, and the electro-thermal coupling input matrix is ​​updated based on the corrected SOC input gain and equivalent thermal response gain. The zero-mean electro-thermal state vector is combined with the group-level SOC equivalent perturbation and group-level temperature equivalent perturbation determined in the current sampling period to form an augmented state. An augmented prediction model is constructed based on the augmented state and the updated electro-thermal coupling input matrix. In non-initial sampling periods, the execution perception prediction state of the previous sampling period is represented as:

[0064]

[0065] The state prediction error between the zero-mean electro-thermal state vector and the execution-aware predicted state is expressed as:

[0066] The average of the state-of-charge prediction error and temperature prediction error for the same battery subgroup is calculated separately. The SOC prediction error and temperature prediction error of each battery subgroup are expressed as follows:

[0067] The group-level SOC equivalent perturbation and the group-level temperature equivalent perturbation are updated recursively according to the following formulas:

[0068]

[0069] The disturbance vector consists of the equivalent SOC disturbance and the equivalent temperature disturbance of each battery sub-group:

[0070] Where m is the number of battery subgroups; Extract the physical state matrix, its dimension is ,Depend on identity matrix of order and Zero-dimensional matrix Horizontal splicing is used to extract dimensions from the execution-aware augmented prediction state. The zero-mean electro-thermal state vector; For dimension The execution perception augmented prediction state; To extract from the execution-aware augmented prediction state, the dimension is The zero-mean electro-thermal state vector is used as the execution sensing prediction state; symbol Indicates based on the first Prediction is based on information available in each sampling period; superscript Represents the transpose of a matrix or vector; This is the state prediction error vector; and The first The SOC prediction error component and temperature prediction error component corresponding to each unit; and The first SOC prediction error and temperature prediction error for each battery sub-group; and The first Equivalent SOC and equivalent temperature disturbances of each battery sub-group; and These are the exponential smoothing coefficients for group-level SOC perturbation estimation and group-level temperature perturbation estimation, respectively, both greater than 0 and less than 1; where, for , for .

[0071] The SOC input gain is expressed as:

[0072] The equivalent thermal response gain is expressed as:

[0073] Among them, the The state-of-charge (POC) input submatrix and temperature input submatrix of each battery subgroup are obtained by multiplying the POC input gain and the equivalent thermal response gain with the corresponding intra-group branch-cell correlation matrix, respectively. The intra-group branch-cell correlation matrix is ​​used to characterize the input interaction relationship between each intra-group equalization branch and the corresponding cell. When each battery subgroup includes 4 cells and corresponds to 3 intra-group equivalent equalization inputs, the POC input submatrix and temperature input submatrix are respectively expressed as follows:

[0074]

[0075] For each battery sub-group, a forward group-level equivalent input characterizing energy output and a reverse group-level equivalent input characterizing energy reception are pre-set. An inter-group input sub-matrix is ​​constructed based on each of the forward and reverse group-level equivalent inputs. The inter-group input sub-matrix is ​​then combined with the state-of-charge input sub-matrix and temperature input sub-matrix of each battery sub-group to form the electro-thermal coupling input matrix. The forward and reverse group-level equivalent inputs have opposite effects on the state of charge, and the additional thermal effects generated by both are non-negative. and The first SOC input gain and equivalent thermal response gain for each sampling period; This is a correction factor for the equalization efficiency determined based on the current average SOC and average temperature of the battery pack. and These are the nominal SOC input gain and the nominal equivalent thermal response gain, respectively. For reference temperature; This is the correction factor for the average temperature on the equivalent thermal response gain. The correction coefficient for the SOC range on the equivalent thermal response gain; when the state of charge is expressed as a percentage, the denominator 5 in the equivalent thermal response gain formula represents a 5 percentage point normalized benchmark for the state of charge range. For the first The state-of-charge input submatrix of each battery subgroup; For the first Temperature input submatrices for each battery subgroup; the state-of-charge input gain and the equivalent thermal response gain are both limited to a preset range near their respective nominal values.

[0076] The augmented state is represented as follows:

[0077] The augmented prediction model is expressed as follows:

[0078] in, The augmented state transition matrix is ​​used to describe the predicted evolution of the zero-mean electro-thermal state vector and group-level equivalent perturbations between adjacent sampling periods; An augmented input matrix containing an electro-thermal coupling input matrix is ​​used to describe the effect of the equivalent equalization input of the control layer after positive and negative decomposition on the augmented state. This is the non-negative control input after positive and negative decomposition.

[0079] The equivalent equalization input of the control layer consists of intra-group equivalent equalization input and inter-group equivalent equalization input, and is decomposed into forward equivalent input and reverse equivalent input:

[0080]

[0081]

[0082] in, This is the equivalent balanced input for the signed control layer without positive / negative decomposition. and They are respectively Non-negative positive input components and non-negative negative input components; For the intra-group equivalent equalization input and the inter-group equivalent equalization input, the positive input component and the negative input component have opposite directions of action on the charged state subsystem, while the additional heat effect of both on the temperature subsystem is non-negative. S3. Determine the dynamic temperature weights, solve the hierarchical model predictive control optimization problem based on the augmented prediction model, augmented state and dynamic temperature weights, and determine the intra-group equilibrium request and inter-group equilibrium request; determine the current inter-group energy transfer path, the current source group and the current target group based on the inter-group equilibrium request, and determine the thermal risk of the candidate path corresponding to the current inter-group energy transfer path as the thermal risk of the current path. Specifically, S3 specifically includes: The dynamic temperature weight is determined based on the SOC range; the hierarchical model predictive control optimization problem is solved based on the augmented prediction model, the augmented state, and the dynamic temperature weight to obtain the optimal control input sequence; the first control input of the current sampling period is extracted from the optimal control input sequence; and the intra-group equalization request corresponding to each intra-group equalization branch in the intra-group active equalization circuit and the inter-group equalization request corresponding to the inter-group active equalization circuit are determined based on the SOC deviation, temperature deviation, and the temperature state prediction result obtained by the augmented prediction model during the solution of the hierarchical model predictive control optimization problem. Candidate source groups, candidate target groups, and inter-group energy transfer paths are determined based on the energy output and energy receiving directions represented by the inter-group equalization requests. The thermal risk of each inter-group energy transfer path is determined based on the average temperature of the candidate source groups, the average temperature of the candidate target groups, the average temperature of the battery pack, and the global temperature difference. A candidate path score is determined based on the inter-group equalization requests corresponding to each inter-group energy transfer path and the candidate path thermal risk. Based on the candidate path scores of each inter-group energy transfer path, the current inter-group energy transfer path, the current source group, and the current target group are determined from the feasible inter-group energy transfer paths, and the thermal risk of the candidate path corresponding to the current inter-group energy transfer path is determined as the current path thermal risk. The dynamic temperature weight is expressed as:

[0083] The objective function of the hierarchical model predictive control optimization problem consists of the stage cost and the terminal cost in the prediction time domain:

[0084] Among them, the augmented prediction model in the first... The zero-mean electro-thermal state vectors predicted in each prediction step are used to calculate the stage cost, which is expressed as:

[0085] The terminal cost is expressed as:

[0086] The control input satisfies:

[0087]

[0088]

[0089] The optimal control input sequence is obtained by solving the hierarchical model predictive control optimization problem:

[0090] in, As the base temperature weight, when the state of charge is expressed as a percentage, the denominator 5 in the dynamic temperature weight formula represents a 5 percentage point normalized baseline for the state of charge range. For prediction in the time domain; This is the prediction step index, with a value range of [value range missing]. to ; In the first The prediction obtained in the sampling period is the first Zero-mean electro-thermal state vector for each sampling period; For the first The stage cost of each prediction step; For the cost of the terminal; The state weight matrix includes SOC deviation weight, dynamic temperature deviation weight, and inter-group average SOC difference weight. To control the input weight matrix; This is the terminal weight matrix; and These are the lower limit and upper limit of the control input, respectively; This is the coefficient matrix of the intra-group branch coupling inequality constraints. This is the upper bound vector for the corresponding inequality constraints; This is the coefficient matrix of the inter-group energy coordination equation constraints. This is the corresponding equality constraint target vector; It is a candidate control input sequence composed of the positive and negative decomposition control inputs of each prediction step in the prediction time domain. For the first The optimal control input sequence is obtained by solving for each sampling period; the equivalent equalization input within each group acts simultaneously on the SOC subsystem and the temperature subsystem through the electro-thermal coupling input matrix, so that the control input corresponding to the equalization branch within each group is jointly determined in the same optimization based on the SOC deviation, temperature deviation, and the temperature state prediction result obtained from the augmented prediction model; from Extract and execute only the first control input of the current sampling period, and directly determine the intra-group equivalent equalization input of each group in the first control input as the intra-group equalization request of the corresponding intra-group equalization branch, and determine the inter-group equivalent equalization input in the first control input as the inter-group equalization request.

[0091] No. The battery subgroup is a candidate source group and the first... When a cell subgroup is a candidate target group, the thermal risk of the candidate path for energy transfer between candidate groups is expressed as follows:

[0092] The candidate path score is represented as follows:

[0093] Based on the candidate path scores of each candidate inter-group energy transfer path, the current inter-group energy transfer path is determined from the feasible candidate inter-group energy transfer paths, with priority given to the inter-group energy transfer path with higher candidate path scores; the thermal risk of the candidate path corresponding to the current inter-group energy transfer path is determined as the thermal risk of the current path; wherein... and These are the candidate source group index and the candidate target group index, respectively. and These are the average temperatures of the candidate source group and the average temperatures of the candidate target group, respectively. This represents the average temperature of the battery pack. For global temperature difference; To prevent positive numbers with excessively small denominators; For from the first The battery subgroup to the first Thermal risk of candidate pathways for energy transfer in individual battery subgroups; The non-negative inter-group equalization request amplitude corresponding to the inter-group energy transfer path; This is a preset non-negative temperature penalty coefficient; The candidate paths are scored; the candidate source group and candidate target group corresponding to the current inter-group energy transfer path are the current source group and the current target group, respectively; the thermal risk of the current path is denoted as... .

[0094] S4. Introduce a fuzzy thermal safety monitor to obtain the inter-group current scaling factor and derate the original inter-group request current corresponding to the current inter-group energy transfer path, and generate intra-group execution instructions and inter-group execution instructions. Specifically, S4 includes: The hottest group flag is determined based on whether the current inter-group energy transfer path passes through the battery subgroup with the highest average temperature. An inter-group SOC difference index, characterizing the degree of inconsistency in average SOC between different battery subgroups, is determined based on the average SOC of each battery subgroup. The global temperature difference, the current path thermal risk, the hottest group flag, and the inter-group SOC difference index are input into a fuzzy thermal safety monitor to obtain an inter-group current scaling factor. The original inter-group request current corresponding to the current inter-group energy transfer path is derated based on the inter-group current scaling factor and the inter-group current upper limit to obtain a derated inter-group execution request current. Intra-group execution instructions are generated based on the intra-group balancing requests corresponding to each intra-group balancing branch, and inter-group execution instructions are also generated based on the current inter-group energy transfer path, the current source group, the current target group, and the derated inter-group execution request current. The hottest group flag is set to 1 when the current inter-group energy transfer path passes through the battery subgroup with the highest average temperature, and 0 otherwise; the inter-group SOC difference index is expressed as:

[0095] The inter-group current scaling factor output by the fuzzy thermal safety monitor is expressed as:

[0096] The reduced inter-group execution request current is expressed as follows:

[0097] in, This represents the difference between the maximum and minimum average SOC values ​​of each battery subgroup. Thermal risk of the candidate path corresponding to the current inter-group energy transfer path; This is the marker value for the hottest group; This is the fuzzy inference mapping corresponding to the fuzzy thermal safety monitor; This is the inter-group current scaling factor, which takes a value greater than 0 and not greater than 1. The original inter-group requested current corresponds to the current inter-group power transfer path, and the original inter-group requested current is based on the non-negative inter-group equalization request amplitude corresponding to the current inter-group power transfer path. Sure; This is the upper limit of the inter-group current. The derated inter-group execution request current is used; the intra-group balancing requests corresponding to each intra-group balancing branch are mapped to the corresponding intra-group execution instructions, the derated inter-group execution request current is mapped to the execution control quantity of the inter-group active balancing circuit, and the inter-group execution instructions are generated according to the current inter-group power transfer path, the current source group, the current target group, and the execution control quantity; when the inter-group active balancing circuit is an isolated CUK balancing circuit, the execution control quantity includes the CUK main switch duty cycle.

[0098] S5. Constrain the intra-group and inter-group execution instructions, and after the intra-group and inter-group execution instructions are actually executed, combine the augmented prediction model to obtain the zero-mean electro-thermal state vector for the next sampling period. Specifically, S5 includes: Based on the allowable control boundaries of the intra-group active balancing circuit and the inter-group active balancing circuit, the intra-group execution instructions and the inter-group execution instructions are constrained to obtain the execution layer constraint processing result. After the intra-group execution instructions and the inter-group execution instructions are actually executed, the execution-aware equivalent input is reconstructed based on the actually executed intra-group execution instructions, the actually executed inter-group execution instructions, and the execution layer constraint processing result to obtain the equivalent execution input. The equivalent execution input is substituted into the augmented prediction model to obtain the execution-aware augmented prediction state for the (k+1)th sampling period, and the zero-mean electro-thermal state vector is extracted from the execution-aware augmented prediction state as the execution-aware prediction state for the (k+1)th sampling period. The execution-aware prediction state is used to update the group-level SOC equivalent perturbation and group-level temperature equivalent perturbation for the (k+1)th sampling period. The equivalent execution input is:

[0099] Substituting the equivalent execution input into the augmented prediction model, the first... The execution-aware augmentation prediction state for each sampling period is represented as follows:

[0100] in, It consists of execution information composed of the intra-group execution instructions actually executed, the inter-group execution instructions actually executed, and the results of execution layer constraint processing; This is an execution-aware mapping that maps the execution information to equivalent execution inputs; In accordance with The equivalent execution input obtained by reconstructing the same positive and negative decomposition structure has the same dimension as... same; To perform the perception augmentation prediction state; extract the zero-mean electro-thermal state vector from the perception augmentation prediction state using the physical state extraction matrix, and use it as the perception prediction state; in the... In each sampling period, based on the state prediction error between the measured zero-mean electro-thermal state vector and the executed sensing prediction state, the group-level SOC equivalent perturbation and group-level temperature equivalent perturbation of each battery subgroup are updated; based on the first sampling period... The average SOC, average temperature, and SOC range of the battery pack are determined by re-collecting the SOC and temperature of each individual cell in each sampling period, and the SOC input gain and equivalent thermal response gain are recalculated.

[0101] S6. Repeat S1 to S5 to achieve active balance control.

[0102] In summary, the lithium-ion battery pack stratified active equalization control method and system provided by this invention, based on adaptive zero steady-state error stratified model predictive control, integrates fuzzy thermal safety supervision, inter-pack CUK thermal sensing routing, CUK current derating, and execution sensing input reconstruction mechanism to form thermal safety fuzzy supervision adaptive zero steady-state error stratified model predictive control. It can achieve state of charge convergence while comprehensively considering the temperature state of individual cells, thermal risks of inter-pack energy transfer paths, and actual execution deviations to suppress local heat concentration, reduce temperature difference accumulation, and improve battery pack temperature consistency.

[0103] In one specific embodiment, the lithium-ion battery pack consists of 12 series-connected lithium-ion battery cells, divided into 3 battery sub-groups, each sub-group comprising 4 cells. Each battery sub-group contains an intra-group improved Buck-Boost balancing circuit, while different battery sub-groups are separated by an inter-group gated matrix isolated CUK balancing circuit.

[0104] like Figure 1 As shown, the hierarchical active balancing system includes a battery pack composed of multiple series-connected cells, an intra-pack improved Buck-Boost balancing circuit located within each battery sub-pack, an inter-pack gating matrix isolated CUK balancing circuit connected between different battery sub-packs, a status acquisition module, and a balancing controller. The status acquisition module is used to acquire the SOC and temperature information of each cell. The balancing controller generates intra-pack balancing commands and inter-pack CUK balancing commands based on the SOC and temperature information, thereby realizing local energy regulation within the same battery sub-pack and energy coordination between different battery sub-packs, respectively.

[0105] An improved Buck-Boost balancing circuit within the same battery sub-group is used to achieve localized energy transfer between individual cells. When one cell within the sub-group has a higher State of Charge (SOC) than another with a lower SOC, the controller controls the corresponding switching branch to allow the high-SOC cell to release energy to the energy storage inductor, which then releases energy to the low-SOC cell. This circuit can also achieve energy transfer between different cells within the same sub-group via an outer energy transfer branch, thereby shortening the energy transmission path between non-adjacent cells.

[0106] like Figure 2 As shown, the improved Buck-Boost balancing circuit within the battery sub-group includes multiple switching branches and an energy storage inductor, enabling the establishment of energy transfer paths between adjacent cells and between different cell clusters within the same battery sub-group. The controller adjusts the PWM commands of the corresponding switching branches to cause high-SOC cells or cell clusters to release energy to the energy storage inductor, which then transfers energy to low-SOC cells or cell clusters, thus achieving fine-grained balancing within the sub-group.

[0107] The operation of the improved Buck-Boost equalization circuit within the battery pack can be divided into local energy transfer between individual cells and outer-layer energy transfer between cell clusters. Taking an example where the SOC of cell B1 within the same battery sub-pack is higher than that of cell B2, such as... Figure 4 As shown in (a), the controller turns on the switch Q1 connected to the high-energy cell B1, forming an energy absorption circuit with cell B1, switch Q1, and energy storage inductor L1. At this time, cell B1 releases energy to energy storage inductor L1, the current in energy storage inductor L1 gradually increases, and some of the energy in the high-energy cell B1 is temporarily stored in energy storage inductor L1. Figure 4 As shown in (b), the controller then turns off switch Q1 and turns on switch Q2 connected to the low-energy cell B2, forming an energy release circuit with energy storage inductor L1, switch Q2, and cell B2. At this time, the energy stored in energy storage inductor L1 is released to cell B2, and the current in energy storage inductor L1 gradually decreases, thus realizing the energy transfer from cell B1 to cell B2. Figure 4 As shown in (c), when the energy of the cell cluster containing cells B1 and B2 is higher than that of the cell cluster containing cells B3 and B4, the controller can transfer energy through the outer branch. That is, the energy of cells B1 and B2 is temporarily stored in the energy storage inductor L3 through the switch Q5 and the energy storage inductor L3, and then the energy in the energy storage inductor L3 is released to cells B3 and B4 through the branch containing the switch Q6, thereby achieving energy coordination between different cell clusters within the same battery sub-group.

[0108] The inter-group selection matrix-type isolated CUK equalization circuit includes a battery selection network and an isolated CUK main power stage. The battery selection network selects the source group and target group from multiple battery subgroups, and the isolated CUK main power stage transfers energy between the source group and the target group. When the average SOC of one battery subgroup is higher than that of another, the subgroup with the higher average SOC can be selected as the source group, and the subgroup with the lower average SOC can be selected as the target group, thus achieving inter-group energy coordination through the isolated CUK equalization circuit.

[0109] like Figure 3 As shown, the inter-group gating matrix-isolated CUK equalization circuit determines the current source group and target group through a gating switch, and transfers energy between the source group and the target group through the isolated CUK main power stage. The source group is the battery sub-group with a higher current average SOC, and the target group is the battery sub-group with a lower current average SOC; the inter-group energy transfer intensity can be changed by adjusting the duty cycle of the CUK main switch.

[0110] like Figure 5 As shown, the operation of the inter-group gating matrix isolated CUK equalization circuit includes two states: switch Q is on and off. Taking the average SOC of battery subgroup q1 being higher than that of battery subgroup q2 as an example, the controller uses the gating matrix to connect q1 to the input side of the isolated CUK main power stage and q2 to the output side, thus transferring energy from the source group q1 to the target group q2. Figure 5 As shown in (a), when the switch Q is turned on, the input-side inductance L a and output side inductor L b The capacitor is in a charging state, while the input-side coupling capacitor C1 and the output-side coupling capacitor C2 are in a discharging state. At this time, the source group q1 discharges to the input-side inductor L. a Charging; simultaneously, the input-side coupling capacitor C1 is energized through the primary side of the transformer T, and the induced voltage on the secondary side, together with the output-side coupling capacitor C2, supplies energy to the output-side circuit and to the output-side inductor L. b During charging, diode D2 is in the off state due to reverse voltage. For example... Figure 5 As shown in (b), when the switch Q is turned off, the input-side inductance L a Discharge begins, input-side coupling capacitor C1 and output-side coupling capacitor C2 enter the charging state, diode D2 conducts and forms a freewheeling path on the output side, and output-side inductor L... b Energy is released to the target group q2. Through the periodic on and off of the switch Q, the isolated CUK equalization circuit can achieve continuous inter-group energy transfer between the source group q1 and the target group q2, and the inter-group energy transfer intensity can be changed by adjusting the duty cycle of the switch Q.

[0111] For a 12-cell series unit, the SOC vector and temperature vector are respectively expressed as:

[0112]

[0113] in, Indicates the discrete sampling period; Indicates the cell number, in this embodiment ; and They represent the first The sampling period The SOC and temperature of individual battery cells; and These represent the SOC vector and temperature vector, respectively, composed of 12 individual battery cells; superscript This indicates the matrix transpose.

[0114] The controller calculates the average SOC and average temperature of the battery pack, constructs the SOC deviation by comparing the SOC of each cell with the average SOC, and constructs the temperature deviation by comparing the temperature of each cell with the average temperature, thus forming a zero-mean electro-thermal state vector, expressed as:

[0115] in, Indicates the first Zero-mean electro-thermal state vector for each sampling period; and These represent the average SOC and average temperature of the 12 battery cells, respectively. This represents a 12-dimensional column vector of all ones. The first 12 elements of the zero-mean electro-thermal state vector represent the deviation of the state of charge (SOC) of each individual cell from the average SOC, and the last 12 elements represent the deviation of the temperature of each individual cell from the average temperature. The expressions for the average SOC and average temperature are:

[0116]

[0117]

[0118]

[0119]

[0120] The first 12 elements of the zero-mean electro-thermal state vector represent the deviation of the SOC of each individual cell from the average SOC, and the last 12 elements represent the deviation of the temperature of each individual cell from the average temperature. , , and These are used for subsequent model parameter adaptation, inter-group path evaluation, and fuzzy thermal safety supervision, respectively.

[0121] Through the above state variable calculations, the controller obtains the current zero-mean electro-thermal state vector. SOC range Global temperature difference The average SOC and average temperature of each battery subgroup are also included; the zero-mean electro-thermal state vector serves as the basis for augmented prediction and optimization, while other statistics are used for model parameter adaptation, inter-group path thermal risk assessment, and fuzzy thermal safety monitoring, respectively.

[0122] Through the above zero-mean processing, the controller can directly use SOC deviation and temperature deviation as prediction and optimization objects, thereby simultaneously considering the battery pack SOC inconsistency and temperature inconsistency.

[0123] The equivalent equalization input vector of the control layer is defined as:

[0124] in, This represents the equivalent current input of the three groups of intra-group balancing channels, with each battery subgroup corresponding to three intra-group balancing inputs; This represents the equivalent group-level input of the inter-group CUK equalization layer. This is a 12-dimensional equivalent balanced current vector used by the control layer, in amperes (A), instead of directly issued PWM or CUK duty cycle commands.

[0125] To facilitate the application of nonnegative boundary constraints in the optimization problem, the 12-dimensional equivalent equilibrium input is decomposed into a positive input and a negative input:

[0126]

[0127] in, Indicates the first The 24-dimensional non-negative control variables solved by the optimizer within each sampling period; and These represent the forward equivalent input and the reverse equivalent input, respectively. Represents the zero vector of the corresponding dimension.

[0128] Therefore, the optimizer actually solves for 24-dimensional non-negative control variables, while the execution layer uses a 12-dimensional equivalent balanced input obtained by subtracting the positive input from the negative input.

[0129] Based on the above definition, the discrete-time electro-thermal coupling prediction model of the hierarchical active balancing system is expressed as follows:

[0130] in, This is the state transition matrix, used to describe the effects of SOC maintenance, inter-unit heat conduction, and convective heat dissipation with the environment on state evolution under conditions of no equilibrium. For the first The electro-thermal coupling input matrix for each sampling period is used to describe the coupling effect of the equivalent equalization input within and between groups after positive and negative decomposition on the SOC state and temperature state. This is the non-negative optimized input after positive and negative decomposition. according to and structure.

[0131] In the input matrix, the positive and negative equivalent equalization inputs have opposite effects on the SOC state, while their additional thermal effects on the temperature state are both non-negative. Therefore, the electro-thermal coupling prediction model can distinguish the influence of energy transfer direction on SOC increases and decreases, describe the additional thermal effects caused by the equalization current under different energy transfer directions, and predict the evolution of individual unit SOC and temperature deviations in the prediction time domain within a unified state space.

[0132] The control strategy in this embodiment is based on adaptive zero steady-state error hierarchical model predictive control, and introduces inter-group CUK thermal sensing routing, fuzzy thermal safety supervision, CUK current derating, and execution sensing input reconstruction to achieve coordination between SOC convergence and temperature suppression.

[0133] like Figure 6 As shown, in the first In each sampling period, the SOC and temperature of each individual cell are first collected, and the average SOC and average temperature of the battery pack and each sub-pack are calculated, along with the SOC range. and global temperature difference And construct a zero-mean electro-thermal state vector In the initial sampling period, the group-level perturbation is initialized to zero. Starting from the second sampling period, the current zero-mean electro-thermal state vector is compared with the execution-aware prediction state generated in the previous sampling period. The equivalent perturbations of group-level SOC and group-level temperature are compared and updated; then adaptive corrections are made based on the average SOC, average temperature, and SOC range of the battery pack. and The updated disturbance estimate and adaptive parameters are substituted into the augmented prediction model. The controller constructs and solves the hierarchical model predictive control optimization problem. By predicting the future SOC deviation and temperature deviation under the action of candidate control input sequences, the objective function is calculated and the optimal control sequence is determined. From this, intra-group equilibrium requests and inter-group CUK equilibrium requests are extracted. Then, based on the inter-group CUK equilibrium requests, the thermal risk and candidate path score of the candidate path are calculated to determine the current inter-group energy transfer path and its current path thermal risk. The CUK current is derating and PWM and CUK execution instructions are generated using a fuzzy thermal safety monitor; finally, the equivalent execution input is reconstructed and the first execution instruction is generated. Execution-aware prediction state for each sampling period This is used for perturbation updates in the next sampling period.

[0134] To compensate for modeling errors, parameter variations, and execution deviations, this embodiment introduces group-level disturbance states in the hierarchical model predictive controller. For three battery subgroups, the disturbance vector is defined as:

[0135] Wherein, the first three elements represent the SOC equivalent perturbation of the three battery sub-groups, and the last three elements represent the temperature equivalent perturbation of the three battery sub-groups; in the initial sampling period, let .

[0136] The zero-mean electro-thermal state variables are combined with the perturbation vector to form an augmented state:

[0137] The structure of the augmented state and the augmented prediction model is represented as follows:

[0138]

[0139]

[0140] in, For the first Augmented state for each sampling period; The zero-mean electro-thermal state vector; This is the group-level perturbation vector; The non-negative optimized input is the result of positive-negative decomposition; For augmented state transition matrix; For augmented input matrix; Assign a matrix to the perturbation; It is a sixth-order identity matrix; It is a 6-row, 24-column zero matrix. At each sampling period, the updated perturbation vector and the adaptive input matrix are substituted into the augmented prediction model.

[0141] Starting from the second sampling period, the perturbation estimation is based on the current zero-mean electro-thermal state vector. Compared with the execution-aware prediction state generated in the previous sampling period The error between the two values ​​is updated. The total prediction error is expressed as:

[0142] For the first The average of the SOC error and temperature error of the four individual cells in each battery sub-group is used to obtain the group-level error:

[0143]

[0144] The perturbation recursive update law is expressed as:

[0145]

[0146] in, Indicates the first Total prediction error for each sampling period; This represents the current measured zero-mean electro-thermal state vector; Indicates the first The first sampling period uses the actual equivalent execution input prediction to obtain the first... Each sampling period's state; Indicates the first A set of individual cell numbers for each battery subgroup; ; and They represent the first SOC prediction error and temperature prediction error for each battery sub-group; and They represent the first Equivalent SOC disturbance and equivalent temperature disturbance of each battery sub-pack; and This is the smoothing coefficient. Through this perturbation augmentation mechanism, the prediction model can continuously correct equivalent perturbations using measurement biases, thereby improving the robustness of the model under mismatch conditions.

[0147] After completing the disturbance update, the controller adjusts the current battery pack average temperature, average SOC, and SOC range. ,right Figure 6 The SOC input gain shown and equivalent thermal response gain Perform adaptive correction:

[0148]

[0149]

[0150] in, For the first SOC input gain per sampling period; This is the nominal SOC input gain; and These are its preset lower and upper limits, respectively; The efficiency adjustment coefficient is used to balance the efficiency. This is the equivalent thermal response gain; This is the nominal equivalent thermal response gain; and These are its preset lower and upper limits, respectively; This represents the average temperature of the battery pack. For reference temperature; The average SOC of the battery pack; This is a reference value when SOC is expressed as a percentage. , , and These are the preset adaptive correction coefficients used in this embodiment; This indicates that the corresponding parameter is restricted to a range. The saturation function within.

[0151] In the Within each sampling period, the controller will update the disturbance estimate, and Substitute the augmented prediction model and construct a hierarchical model for predictive control optimization; under constraints of control input boundaries, intra-group branch coupling, and inter-group energy coordination, perform rolling optimization on the candidate control input sequences in the prediction time domain:

[0152]

[0153]

[0154]

[0155] in, For the first The predictive control objective function for each sampling period; To predict the stage cost within the time domain; To predict the terminal cost at the end of the time domain; To predict the length of the time domain; For predicting the step index in the time domain; For candidate control input sequences; This is the optimal control sequence; , and These are the augmented state weight matrix, the control input weight matrix, and the terminal state weight matrix, respectively. The controller only... The first control action is applied to the current sampling period, and the intra-group equalization request is extracted from that control action. Inter-group CUK balance requests .

[0156] The individual temperature deviation term in the objective function employs dynamic temperature weighting.

[0157]

[0158] in, For the first Dynamic temperature weighting for each sampling period; The temperature-weighted baseline value; For the first The SOC range of 12 individual battery cells at each sampling time; For the first The sampling time of the first sampling moment SOC of a single battery cell =1,2,…,12; the denominator 5 is the normalized benchmark value of the SOC range, which corresponds to 5 percentage points when the SOC is expressed as a percentage.

[0159] For intra-group equilibrium, the controller, when solving the hierarchical model predictive control optimization problem, utilizes an electro-thermal coupling predictive model and dynamic temperature weights. In addition to a temperature deviation penalty term, the system comprehensively evaluates the impact of different intra-group balancing inputs on SOC deviation and temperature deviation, and extracts intra-group balancing requests with thermal sensing characteristics from the optimal control sequence. When the temperature of a certain unit is too high, the temperature deviation penalty reduces the tendency of that unit to continue to undertake strong Buck-Boost balancing actions; when the temperature of a certain unit is low and the SOC correction requirement is obvious, the controller maintains the corresponding balancing strength. Thus, intra-group thermal safety regulation is completed in the intra-group balancing request generation stage, and the generated intra-group balancing requests are subsequently mapped to intra-group PWM commands without the need for independent intra-group temperature derating at the execution layer.

[0160] For inter-group CUK equalization requests, the controller determines candidate source groups based on the positive energy output demand in the inter-group CUK equalization requests, and determines candidate target groups based on the negative energy reception demand in the inter-group CUK equalization requests; typically, candidate source groups correspond to battery subgroups with relatively high average SOC, and candidate target groups correspond to battery subgroups with relatively low average SOC.

[0161] To evaluate the thermal burden of the energy transfer path between candidate groups, the controller uses the average temperature of the candidate source group, the average temperature of the candidate target group, the average temperature of the battery pack, and the global temperature difference. Calculate the thermal risk of candidate paths :

[0162] in, For the first The first battery subgroup was used as a candidate source group, the second Path thermal risk when individual battery subgroups are used as candidate target groups; and These are the average temperatures of the candidate source group and the average temperatures of the candidate target group, respectively. This represents the average temperature of the battery pack. For global temperature difference; To prevent positive numbers with excessively small denominators, this embodiment limits the thermal risk of candidate paths to a range. Inside.

[0163] Based on inter-group CUK balance request and candidate path hot risk Calculate candidate path scores :

[0164] in, For the first The battery subgroup to the first Inter-group balance request for energy transfer of individual battery sub-groups; This represents the temperature penalty coefficient. The controller determines the current inter-CUK energy transfer path based on the candidate path score. and define:

[0165] Determine the current inter-group energy transfer path and current path heat risk Then, the controller determines the SOC difference index between battery subgroups based on the average SOC of each battery subgroup. And determine whether the current path passes through the hottest group to obtain the flag value. The fuzzy thermal safety monitor uses global temperature difference. Current path heat risk The flag value and the inter-group SOC difference index are inputs, and the output is the CUK current scaling factor:

[0166] like Figure 7 As shown, the fuzzy thermal safety monitor uses a membership function to represent the global temperature difference. Current path heat risk and intergroup SOC difference indicators Each is mapped to a corresponding fuzzy level, and combined with the flag quantity. Output CUK current scaling factor according to preset fuzzy rules .

[0167]

[0168] When the global temperature difference is large, the current path has a high thermal risk, and the current path passes through the hottest group, the fuzzy thermal safety monitor outputs a smaller CUK current scaling factor; when the inter-group SOC difference is large and the current path has a low thermal risk, it outputs a larger CUK current scaling factor. Thus, the controller can dynamically trade off between the inter-group SOC balancing requirements and the temperature suppression requirements.

[0169] Based on the CUK current scaling factor, the original CUK request current is downrated to obtain the downrated CUK execution request current:

[0170] in, This indicates the CUK execution request current after derating; This indicates the original CUK request current; This indicates the preset upper limit of CUK current; This represents the CUK current scaling factor output by the fuzzy thermal safety monitor. The controller adjusts the current inter-group energy transfer path accordingly. and Generate CUK execution instructions between groups and load balancing requests within groups. Mapped to in-group PWM commands.

[0171] In practical equalization systems, the ideal control output of the optimizer is not directly equivalent to the final execution input. During actual execution, intra-group requests need to be mapped to PWM instructions, and inter-group requests need to be mapped to CUK source group, target group, execution request current, and duty cycle instructions, and may be affected by current limiting, path selection, and derating mechanisms.

[0172] To reduce the mismatch between the ideal input of the optimization layer and the actual execution input of the circuit, an execution-aware input reconstruction mechanism is introduced. Let the... The ideal control input for the optimizer output in each sampling period is: The ideal control input, after undergoing PWM mapping, CUK duty cycle mapping, path selection, current derating, and execution layer constraint processing, forms the actually executed intra-group execution instructions, the actually executed inter-group execution instructions, and the execution layer constraint processing results; these information are then combined to form execution information. And reconstruct the equivalent execution input by performing a perceptual mapping:

[0173] in, It consists of execution information composed of the intra-group execution instructions actually executed, the inter-group execution instructions actually executed, and the results of execution layer constraint processing; This is an execution-aware mapping that maps the execution information to equivalent execution inputs; For the first Each sampling period is based on... The equivalent execution input is obtained by reconstructing the same positive and negative decomposition structure.

[0174] Substituting the equivalent execution input into the augmented prediction model, the execution-perception one-step prediction is expressed as:

[0175] in, In the first The sampling period is based on the real executable input. Augmented state prediction values ​​at each sampling time; and These are the augmented state transition matrix and the augmented input matrix, respectively. This serves as the equivalent execution input after the execution layer reconstruction. The first 24 electro-thermal state components of the augmented state prediction value are taken to obtain the execution-aware prediction state. This state is in the 1st Each sampling period is compared with the newly acquired and constructed zero-mean electro-thermal state vector to update the group-level SOC equivalent perturbation and the group-level temperature equivalent perturbation.

[0176]

[0177] The above balancing process is repeated until the difference between the maximum and minimum SOC values ​​of each cell in the battery pack is less than the preset SOC balancing threshold.

[0178] To further illustrate the balancing effect of this invention under static conditions, a set of static balancing simulation examples are provided. This example uses 12 series-connected lithium-ion battery cells to form a battery pack, and divides them into three sub-packs of four cells each, denoted as Sub-pack 1, Sub-pack 2, and Sub-pack 3. During the static balancing process, the external charging / discharging current of the battery pack is 0A, the ambient temperature is set to 298.15K, and the initial temperature of each battery cell is also set to 298.15K.

[0179] In this embodiment, the inputs to the control method include the state of charge (SOC) vector of each battery cell, the temperature vector, battery subgroup division information, the execution status of the intra-group equalization circuit and the inter-group equalization circuit, as well as the equalization threshold, sampling period, maximum allowable equalization current, and thermal safety adjustment parameters. The initial SOCs of the 12 battery cells B1 to B12 are approximately 51%, 42%, 52%, 45%, 53%, 59%, 63%, 52%, 61%, 60%, 49%, and 56%, respectively. Among them, cells B7, B9, B10, and B6 are in a relatively high SOC state, while cells B2, B4, and B11 are in a relatively low SOC state. From the perspective of the average state at the group level, the average SOC of the second and third battery subgroups is higher than that of the first battery subgroup. Therefore, in the initial stage of static equalization, there is both an energy transfer demand between cells within the group and an inter-group energy transfer demand from battery subgroups with high average SOC to battery subgroups with low average SOC.

[0180] The control method outputs include pulse width modulation (PWM) drive commands for the improved Buck-Boost equalization circuit within each battery sub-group, and source group number, target group number, equalization current, and duty cycle commands for the inter-group isolated CUK equalization circuit. During static equalization, the intra-group outputs control the on / off switching of the corresponding cell branch or cell cluster branch, enabling energy transfer from high-SOC cells to low-SOC cells; the inter-group outputs control the gating matrix and the isolated CUK main power stage, enabling energy transfer from high-average-SOC battery sub-groups to low-average-SOC battery sub-groups. Outputs used to evaluate the equalization effect include the SOC versus time curves for each cell, the temperature versus time curves for each cell, the equalization completion time, the maximum temperature difference during the equalization process, and the final temperature distribution.

[0181] like Figure 8 As shown in (a), (b), and (c), under the three control strategies of Model Predictive Control (MPC), Adaptive Hierarchical Model Predictive Control with No Steady-State Error (AOF-HMPC), and Thermal Safety Fuzzy Supervised Adaptive Hierarchical Model Predictive Control with No Steady-State Error (TSF-AOF-HMPC), the State of Charge (SOC) of the 12 battery cells can gradually converge to the preset equalization threshold range. The static equalization completion times for MPC, AOF-HMPC, and TSF-AOF-HMPC are approximately 599.51 s, 661.56 s, and 905.12 s, respectively. The relatively longer equalization time of TSF-AOF-HMPC is because this strategy not only reduces the CUK equalization intensity when the inter-group energy transfer path has a high thermal risk, but also adjusts the intra-group Buck-Boost equalization action through the temperature weight and constraint terms in the electro-thermal coupling objective function, ensuring that intra-group and inter-group equalization both conform to the thermal safety priority principle, thereby suppressing local heat concentration.

[0182] like Figure 9 As shown in (a), (b), and (c), under the same initial SOC and initial temperature conditions, the maximum temperature difference during the MPC equalization process is approximately 4.22 K, the maximum temperature difference during the AOF-HMPC equalization process is approximately 3.2 K, and the maximum temperature difference during the TSF-AOF-HMPC equalization process is approximately 2.40 K. Therefore, under static equalization conditions, this invention, on the one hand, adjusts the distribution intensity of the Buck-Boost equalization action within the battery pack through temperature weighting and temperature constraints, and on the other hand, uses a thermal safety fuzzy supervisor to derating the CUK equalization current between battery packs. This allows intra-pack equalization and inter-pack equalization to synergistically consider both SOC convergence and temperature suppression, thereby reducing the temperature dispersion within the battery pack, suppressing local heat accumulation, and improving the thermal safety of the equalization process.

[0183] It should be noted that the above embodiment uses 12 series-connected lithium-ion battery cells, 3 battery sub-groups, and 4 cells per group as an example for illustration, but the present invention is not limited to this specific number. For other battery packs composed of multiple series-connected lithium-ion battery cells and divided into multiple battery sub-groups, the hierarchical active balancing control method provided by the present invention can also be used.

[0184] It should also be noted that the intra-group balancing circuit is not limited to the improved Buck-Boost balancing circuit; other active balancing circuits that can achieve energy transfer within the same battery sub-group can also be used as alternatives. The inter-group balancing circuit is not limited to the isolated CUK balancing circuit; isolated DC-DC converters that can achieve energy transfer between different battery sub-groups can also be used as alternatives.

[0185] The fuzzy thermal safety monitor in the above embodiments can also be replaced by a rule table, lookup table function, or other monitor that can output a current scaling factor based on the global temperature difference, path thermal risk, participation of the hottest group, and SOC difference between groups.

[0186] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the embodiments disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. It should be understood that the invention is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for active equalization control of lithium-ion battery components for thermal safety, characterized in that, include: S1. Construct the zero-mean electro-thermal state vector for the current sampling period based on the SOC and temperature of each individual cell; S2. Introduce group-level SOC and temperature equivalent perturbation to form a perturbation vector; combine the perturbation vector with the zero-mean electro-thermal state vector to form an augmented state and construct an augmented prediction model; S3. Determine the dynamic temperature weights, and solve the hierarchical model predictive control optimization problem based on the augmented prediction model, augmented state and dynamic temperature weights, and determine the intra-group equilibrium request and inter-group equilibrium request. Based on the inter-group balance request, the current inter-group energy transfer path, the current source group, and the current target group are determined, and the thermal risk of the candidate path corresponding to the current inter-group energy transfer path is determined as the thermal risk of the current path. S4. Introduce a fuzzy thermal safety monitor to obtain the inter-group current scaling factor and derate the original inter-group request current corresponding to the current inter-group energy transfer path, and generate intra-group execution instructions and inter-group execution instructions. S5. Constrain the intra-group and inter-group execution instructions, and after the intra-group and inter-group execution instructions are actually executed, combine the augmented prediction model to obtain the zero-mean electro-thermal state vector for the next sampling period. S6. Repeat S1 to S5 to achieve active balance control.

2. The active equalization control method for lithium-ion battery components oriented towards thermal safety according to claim 1, characterized in that, S1 specifically includes: In the kth sampling period, the SOC and temperature of each cell in the lithium-ion battery pack are obtained. Based on the SOC and temperature of each cell, the average SOC of the battery pack, the average temperature of the battery pack, the average SOC of each battery subgroup, the average temperature of each battery subgroup, the SOC range and the global temperature difference are determined, and a zero-mean electro-thermal state vector is constructed. The average SOC and average temperature of the battery pack are expressed as follows: The average SOC and average temperature of each battery sub-group are expressed as follows: The SOC range and global temperature difference are respectively expressed as: The zero-mean electro-thermal state vector is represented as: in, Indicates the first Zero-mean electro-thermal state vector for each sampling period; For sampling period index; For single-entity indexes; and The first The sampling period The SOC and temperature of the monomer; and They are respectively from A column vector consisting of the SOC and temperature of the monomer; The total number of individual units; For battery subgroup indexes; For the first Individual index set of each battery subgroup The number of single entities in the single entity index set; for A column vector of all 1s.

3. The active equalization control method for lithium-ion battery components oriented towards thermal safety according to claim 2, characterized in that, S2 specifically includes: When the k-th sampling period is the initial sampling period, the group-level SOC equivalent perturbation and group-level temperature equivalent perturbation of each battery subgroup are initialized to zero. When the k-th sampling period is not the initial sampling period, the execution-aware prediction state generated after reconstructing the equivalent execution input based on the actual intra-group execution instructions and the actual inter-group execution instructions in the previous sampling period is obtained. Based on the state prediction error between the zero-mean electro-thermal state vector and the execution-aware prediction state, the group-level SOC equivalent perturbation and group-level temperature equivalent perturbation of each battery subgroup are updated. Based on the battery group average SOC, battery group average temperature and SOC range obtained in the k-th sampling period, the SOC input gain and equivalent thermal response gain are adaptively corrected, and the electro-thermal coupling input matrix is ​​updated based on the corrected SOC input gain and equivalent thermal response gain. The zero-mean electro-thermal state vector is combined with the group-level SOC equivalent perturbation and group-level temperature equivalent perturbation determined in the current sampling period to form an augmented state. An augmented prediction model is constructed based on the augmented state and the updated electro-thermal coupling input matrix. In non-initial sampling periods, the execution perception prediction state of the previous sampling period is represented as: The state prediction error between the zero-mean electro-thermal state vector and the execution-aware predicted state is expressed as: No. The SOC prediction error and temperature prediction error of each battery subgroup are expressed as follows: The group-level SOC equivalent perturbation and the group-level temperature equivalent perturbation are updated recursively according to the following formulas: The disturbance vector consists of the equivalent SOC disturbance and the equivalent temperature disturbance of each battery sub-group: Where m is the number of battery subgroups; Extract the matrix for the physical state; For dimension The execution perception augmented prediction state; To extract from the execution-aware augmented prediction state, the dimension is The zero-mean electro-thermal state vector is used as the execution sensing prediction state; This is the state prediction error vector; and The first The SOC prediction error component and temperature prediction error component corresponding to each unit; and The first SOC prediction error and temperature prediction error for each battery sub-group; and The first Equivalent SOC and equivalent temperature disturbances of each battery sub-group; and These are the exponential smoothing coefficients for group-level SOC perturbation estimation and group-level temperature perturbation estimation, respectively.

4. The active equalization control method for lithium-ion battery components oriented towards thermal safety according to claim 3, characterized in that, The SOC input gain is expressed as: The equivalent thermal response gain is expressed as: in, and The first SOC input gain and equivalent thermal response gain for each sampling period; This is a correction factor for the equalization efficiency determined based on the current average SOC and average temperature of the battery pack. and These are the nominal SOC input gain and the nominal equivalent thermal response gain, respectively. For reference temperature; This is the correction factor for the average temperature on the equivalent thermal response gain. This is the correction factor for the SOC range on the equivalent thermal response gain; The augmented state is represented as follows: The augmented prediction model is expressed as follows: in, For augmented state transition matrix; This is an augmented input matrix that includes the electro-thermal coupling input matrix; This is the non-negative control input after positive and negative decomposition.

5. The active equalization control method for lithium-ion battery components oriented towards thermal safety according to claim 4, characterized in that, S3 specifically includes: The dynamic temperature weight is determined based on the SOC range; the hierarchical model predictive control optimization problem is solved based on the augmented prediction model, the augmented state, and the dynamic temperature weight to obtain the optimal control input sequence; the first control input of the current sampling period is extracted from the optimal control input sequence; and the intra-group equalization request corresponding to each intra-group equalization branch in the intra-group active equalization circuit and the inter-group equalization request corresponding to the inter-group active equalization circuit are determined based on the SOC deviation, temperature deviation, and the temperature state prediction result obtained by the augmented prediction model during the solution of the hierarchical model predictive control optimization problem. Candidate source groups, candidate target groups, and inter-group energy transfer paths are determined based on the energy output and energy receiving directions represented by the inter-group equalization requests. The thermal risk of each inter-group energy transfer path is determined based on the average temperature of the candidate source groups, the average temperature of the candidate target groups, the average temperature of the battery pack, and the global temperature difference. A candidate path score is determined based on the inter-group equalization requests corresponding to each inter-group energy transfer path and the candidate path thermal risk. Based on the candidate path scores of each inter-group energy transfer path, the current inter-group energy transfer path, the current source group, and the current target group are determined from the feasible inter-group energy transfer paths, and the thermal risk of the candidate path corresponding to the current inter-group energy transfer path is determined as the current path thermal risk. The dynamic temperature weight is expressed as: The objective function of the hierarchical model predictive control optimization problem consists of the stage cost and the terminal cost in the prediction time domain: Among them, the augmented prediction model in the first... The zero-mean electro-thermal state vectors predicted in each prediction step are used to calculate the stage cost, which is expressed as: The terminal cost is expressed as: The control input satisfies: The optimal control input sequence is obtained by solving the hierarchical model predictive control optimization problem: in, Base temperature weighting; For prediction in the time domain; For the prediction step index; In the first The prediction obtained in the sampling period is the first Zero-mean electro-thermal state vector for each sampling period; For the first The stage cost of each prediction step; For the cost of the terminal; The state weight matrix includes SOC deviation weight, dynamic temperature deviation weight, and inter-group average SOC difference weight. To control the input weight matrix; This is the terminal weight matrix; and These are the lower limit and upper limit of the control input, respectively; This is the coefficient matrix of the intra-group branch coupling inequality constraints. This is the upper bound vector for the corresponding inequality constraints; This is the coefficient matrix of the inter-group energy coordination equation constraints. This is the corresponding equality constraint target vector; It is a candidate control input sequence composed of the positive and negative decomposition control inputs of each prediction step in the prediction time domain. For the first The optimal control input sequence is obtained by solving for each sampling period; the equivalent equalization input within each group acts simultaneously on the SOC subsystem and the temperature subsystem through the electro-thermal coupling input matrix, so that the control input corresponding to the equalization branch within each group is jointly determined in the same optimization based on the SOC deviation, temperature deviation, and the temperature state prediction result obtained from the augmented prediction model; from Extract and execute only the first control input of the current sampling period, and directly determine the intra-group equivalent equalization input of each group in the first control input as the intra-group equalization request of the corresponding intra-group equalization branch, and determine the inter-group equivalent equalization input in the first control input as the inter-group equalization request.

6. The active equalization control method for lithium-ion battery components oriented towards thermal safety according to claim 5, characterized in that, No. The battery subgroup is a candidate source group and the first... When a cell subgroup is a candidate target group, the thermal risk of the candidate path for energy transfer between candidate groups is expressed as follows: The candidate path score is represented as follows: Based on the candidate path scores of each candidate inter-group energy transfer path, the current inter-group energy transfer path is determined from the feasible candidate inter-group energy transfer paths, with priority given to the inter-group energy transfer path with higher candidate path scores; the thermal risk of the candidate path corresponding to the current inter-group energy transfer path is determined as the thermal risk of the current path; wherein... and These are the candidate source group index and the candidate target group index, respectively. and These are the average temperatures of the candidate source group and the average temperatures of the candidate target group, respectively. This represents the average temperature of the battery pack. For global temperature difference; To prevent positive numbers with excessively small denominators; For from the first The battery subgroup to the first Thermal risk of candidate pathways for energy transfer in individual battery subgroups; The non-negative inter-group equalization request amplitude corresponding to the inter-group energy transfer path; This is a preset non-negative temperature penalty coefficient; The candidate paths are scored; the candidate source group and candidate target group corresponding to the current inter-group energy transfer path are the current source group and the current target group, respectively; the thermal risk of the current path is denoted as... .

7. The active equalization control method for lithium-ion battery components oriented towards thermal safety according to claim 6, characterized in that, S4 specifically includes: The hottest group flag is determined based on whether the current inter-group energy transfer path passes through the battery subgroup with the highest average temperature. An inter-group SOC difference index, characterizing the degree of inconsistency in average SOC between different battery subgroups, is determined based on the average SOC of each battery subgroup. The global temperature difference, the current path thermal risk, the hottest group flag, and the inter-group SOC difference index are input into a fuzzy thermal safety monitor to obtain an inter-group current scaling factor. The original inter-group request current corresponding to the current inter-group energy transfer path is derated based on the inter-group current scaling factor and the inter-group current upper limit to obtain a derated inter-group execution request current. Intra-group execution instructions are generated based on the intra-group balancing requests corresponding to each intra-group balancing branch, and inter-group execution instructions are also generated based on the current inter-group energy transfer path, the current source group, the current target group, and the derated inter-group execution request current. The hottest group flag is set to 1 when the current inter-group energy transfer path passes through the battery subgroup with the highest average temperature, and 0 otherwise; the inter-group SOC difference index is expressed as: The inter-group current scaling factor output by the fuzzy thermal safety monitor is expressed as: The reduced inter-group execution request current is expressed as follows: in, This represents the difference between the maximum and minimum average SOC values ​​of each battery subgroup. Thermal risk of the candidate path corresponding to the current inter-group energy transfer path; This is the marker value for the hottest group; This is the fuzzy inference mapping corresponding to the fuzzy thermal safety monitor; This is the inter-group current scaling factor; The original inter-group requested current corresponds to the current inter-group power transfer path, and the original inter-group requested current is based on the non-negative inter-group equalization request amplitude corresponding to the current inter-group power transfer path. Sure; This is the upper limit of the inter-group current. The derated inter-group execution request current is used; the intra-group balancing requests corresponding to each intra-group balancing branch are mapped to the corresponding intra-group execution instructions, the derated inter-group execution request current is mapped to the execution control quantity of the inter-group active balancing circuit, and the inter-group execution instructions are generated according to the current inter-group power transfer path, the current source group, the current target group, and the execution control quantity; when the inter-group active balancing circuit is an isolated CUK balancing circuit, the execution control quantity includes the CUK main switch duty cycle.

8. The active equalization control method for lithium-ion battery components oriented towards thermal safety according to claim 7, characterized in that, S5 specifically includes: Based on the allowable control boundaries of the intra-group active balancing circuit and the inter-group active balancing circuit, the intra-group execution instructions and the inter-group execution instructions are constrained to obtain the execution layer constraint processing result. After the intra-group execution instructions and the inter-group execution instructions are actually executed, the execution-aware equivalent input is reconstructed based on the actually executed intra-group execution instructions, the actually executed inter-group execution instructions, and the execution layer constraint processing result to obtain the equivalent execution input. The equivalent execution input is substituted into the augmented prediction model to obtain the execution-aware augmented prediction state of the (k+1)th sampling period, and the zero-mean electro-thermal state vector is extracted from the execution-aware augmented prediction state as the execution-aware prediction state of the (k+1)th sampling period. The equivalent execution input is: Substituting the equivalent execution input into the augmented prediction model, the first... The execution-aware augmentation prediction state for each sampling period is represented as follows: in, It consists of execution information composed of the intra-group execution instructions actually executed, the inter-group execution instructions actually executed, and the results of execution layer constraint processing; This is an execution-aware mapping that maps the execution information to equivalent execution inputs; In accordance with The equivalent execution input obtained by reconstructing the same positive and negative decomposition structure has the same dimension as... same; To perform the perception augmentation prediction state; extract the zero-mean electro-thermal state vector from the perception augmentation prediction state using the physical state extraction matrix, and use it as the perception prediction state; in the... In each sampling period, based on the state prediction error between the measured zero-mean electro-thermal state vector and the executed sensing prediction state, the group-level SOC equivalent perturbation and group-level temperature equivalent perturbation of each battery subgroup are updated; based on the first sampling period... The average SOC, average temperature, and SOC range of the battery pack are determined by re-collecting the SOC and temperature of each individual cell in each sampling period, and the SOC input gain and equivalent thermal response gain are recalculated.

9. An active equalization control system for lithium-ion battery components oriented towards thermal safety, characterized in that, The system includes a battery pack, an intra-pack active balancing circuit, an inter-pack active balancing circuit, a status acquisition module, and a controller. The battery pack includes multiple battery sub-packs, and each battery sub-pack includes multiple cells connected in series. The intra-pack active balancing circuit is located inside each battery sub-pack and includes multiple intra-pack balancing branches. The inter-pack active balancing circuit is connected between different battery sub-packs and can form multiple selectable inter-pack energy transfer paths between different battery sub-packs. The status acquisition module is used to acquire the SOC and temperature of each cell in the battery pack in each sampling cycle; the status acquisition module, the intra-pack active balancing circuit and the inter-pack active balancing circuit are all connected to the controller; the controller is used to implement the method as described in any one of claims 1-8.

10. The active equalization control system for lithium-ion battery components oriented towards thermal safety according to claim 9, characterized in that, The intra-group active balancing circuit is an intra-group improved Buck-Boost balancing circuit disposed within each of the battery sub-groups. The inter-group active balancing circuit is an isolated CUK balancing circuit connected between different battery sub-groups via a gating matrix. The intra-group execution instructions include pulse width modulation instructions corresponding to each intra-group improved Buck-Boost balancing branch. The inter-group execution instructions include the current source group identifier, the current target group identifier, and the CUK main switch duty cycle for controlling the gating matrix and the isolated CUK balancing circuit.

Images