A battery temperature control method based on balanced topology

Abstract

The application provides a battery temperature control method based on an equalization topology, and relates to the technical field of battery temperature control.The application obtains multi-dimensional data to calculate a current state of charge to determine an equalization starting condition, so as to avoid loss caused by invalid equalization.A prediction model is constructed to map a control signal and a duty cycle into an actual current, and a predicted state of charge and a predicted temperature are output, so as to realize thermal decoupling and early prediction.Two prediction differences are non-dimensionalized and scaled, an adaptive weight coefficient is introduced to construct a composite optimization objective function, and a dynamic compromise between power convergence and temperature balance is achieved.A firework algorithm initialized by a Tent chaotic mapping is used for iterative optimization to obtain an optimal control signal, so as to prevent falling into a local optimum.A switch tube and a flying induction are turned on according to the optimal control signal to perform charge-discharge energy transfer, and a mutual exclusion switch is closed under a bypass instruction, so as to consider both power equalization and physical thermal isolation of an abnormal heating single cell.

Description

Technical Field

[0001] This invention relates to the field of battery temperature control technology, specifically to a battery temperature control method based on balanced topology. Background Technology

[0002] In the large-scale energy storage industry, battery packs are prone to inconsistencies in the state of individual cells during long-term operation, affecting the overall lifespan and safety of the battery pack. To mitigate this phenomenon, battery balancing technology has become a core approach, primarily consisting of a bottom-level balancing topology and an upper-level balancing control strategy. Regarding the balancing topology, active balancing architectures based on energy storage components such as inductors are widely used due to their lossless energy transfer characteristics. However, with the increase in overall power density, the technological bottleneck is gradually shifting towards the control strategy. Traditional control strategies mostly focus on rapidly reducing voltage or state-of-charge differences between batteries as a single objective. Under high-rate charge and discharge conditions, the balancing current itself generates a significant thermal effect, forcing the control algorithm to handle the complex coupling between electrical parameters and thermodynamic states. The focus of control technology research and development is gradually shifting from single-objective power balancing to multi-objective optimization control that coordinates electro-thermal processes.

[0003] In the prior art, CN115472929A discloses a voltage and temperature equalization control method based on a reconfigurable battery module. This scheme mainly calculates the reversible entropy coefficient by collecting battery temperature, battery voltage, and open-circuit voltage, and uses this to construct a voltage and temperature equalization objective function. In terms of execution logic, this method collects multiple operating temperatures and voltages of the module within a set moving-time window and substitutes them into the objective function to generate multiple battery equalization values, which are then sorted. Subsequently, the operating temperature and voltage corresponding to the minimum equalization value are selected as the globally set equalization temperature and voltage. Finally, a charge-discharge reconfiguration strategy is generated based on this, reducing the voltage and temperature differences between batteries by controlling the charge-discharge states of different modules.

[0004] However, the aforementioned existing technologies have significant limitations in practical applications and cannot effectively cope with complex electrothermal coupling nonlinear conditions. First, at the level of control strategy construction, existing technologies adopt a fixed objective function structure and fail to establish a dynamic constraint relationship between state-of-charge differences and temperature differences. During high-rate charging and discharging, forcibly suppressing state-of-charge differences can trigger local overcurrent heating. Due to the lack of an adaptive weight coefficient adjustment mechanism based on temperature thresholds, the overall system is prone to exacerbating the risk of thermal runaway due to continuous execution of voltage balancing commands. Second, in terms of topology architecture, existing solutions mostly rely on limited energy transfer between adjacent cells. Long-distance balancing requires step-by-step transfer, leading to increased energy loss. At the same time, the scalability of existing topologies is severely insufficient. Increasing the number of batteries often leads to a linear or more complex increase in circuit components, making it difficult to flexibly adapt to battery packs of different sizes, and lacking a physical bypass isolation mechanism for individual batteries with excessively rapid temperature rise.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a battery temperature control method based on balanced topology to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A battery temperature control method based on balanced topology, comprising the following steps: Step 1: Obtain the operating parameters of each individual cell in the balanced topology, calculate the current state of charge of each individual cell using the ampere-hour integration method, and determine whether the state of charge of the battery pack meets the conditions for balanced start-up. Step 2: When the equalization start-up conditions are met, construct prediction models for the state of charge and temperature of individual cells respectively. Combine the candidate control signal to be input with the duty cycle of the equalization topology and map it into the actual charging and discharging current and equalization current received by the battery. Input the data into the two prediction models and calculate the predicted state of charge and predicted temperature of each individual cell at the next moment. Step 3: Calculate the difference between the predicted state of charge and the predicted temperature based on the predicted state of charge and the predicted temperature at the next moment. Introduce an adaptive weighting coefficient whose magnitude is determined by the current temperature difference in segments, and construct an optimization objective function to simultaneously constrain the convergence of the state of charge and the temperature balance. Step 4: Using the fireworks algorithm that initializes the population with Tent chaotic mapping, candidate control signals are generated and the prediction models of charge state and temperature are called. Iterative optimization is performed based on the composite optimization objective function as the evaluation criterion to obtain the optimal control signal. Step 5: According to the optimal control signal, control the switching transistor of the corresponding single cell in the balanced topology to conduct with the flying inductor, and control the parallel mutual exclusion switch of the corresponding single cell to close when the command state is bypass.

[0008] Furthermore, in obtaining the balanced topology, the operating parameters of each cell in the battery pack include real-time voltage, main circuit charging and discharging current, initial state of charge, temperature data, and battery pack bus terminal charging and discharging current. Among them, the temperature data includes the average temperature of each cell, the temperature on both sides, and the ambient temperature. Based on the real-time voltage of each individual battery, the charging and discharging current of the main circuit, the initial state of charge, and the preset rated capacity of each battery, the current state of charge of each individual battery is calculated using the ampere-hour integration method, with the time step between the current time and the last time the state of charge of each battery was calculated and fed back as a constraint. Calculate the difference between the current maximum and minimum state of charge (SOC) of each individual cell and compare it with a preset equalization start threshold. If the equalization start condition is met, proceed with subsequent steps for equalization adjustment; otherwise, determine that no further equalization adjustment is required for the current state. The equalization start condition is that the difference between the current maximum and minimum SOC is greater than or equal to the preset equalization start threshold.

[0009] Furthermore, by combining the candidate control signal to be input with the duty cycle of the equalization topology, the actual main circuit charging and discharging current and equalization current currently received by each individual battery are obtained through mapping. By using the ampere-hour integration method, the change in discharge charge and the change in equalization charge of each individual cell are calculated based on the actual main circuit charging and discharging current and equalization current received by each individual cell, with the time step of the preset control cycle as a constraint. The current state of charge, charge / discharge change, and equilibrium charge change of each individual cell are used as inputs to the state of charge prediction model to obtain the predicted state of charge of each individual cell at the next moment.

[0010] Furthermore, the current main circuit charge / discharge current mapping logic actually received by each individual battery cell includes: If the current single cell control signal is a discharge signal, the main circuit charging and discharging current is the product of the acquired battery pack bus terminal charging and discharging current and the current equalization topology duty cycle, and the current direction is set to reverse. If the current single-cell control signal is a bypass connection signal, the main circuit charging current is directly mapped to 0; If the current single cell control signal is a charging signal, the main circuit charging and discharging current is the product of the acquired battery pack bus terminal charging and discharging current and the current equalization topology duty cycle, but the current direction is set to positive. The current equalization current mapping logic for each individual battery cell includes: If the current single cell control signal is a discharge signal, the equalization current is calculated by multiplying the total voltage of the current discharge cell, the square of the current equalization topology duty cycle, and the preset equalization period as the numerator, and using twice the preset inductance as the denominator, and calculating the ratio of the numerator to the denominator. The current direction is set to reverse. If the current single-cell control signal is connected to bypass, the equalization current is directly mapped to 0; If the current single-cell control signal is a charging signal, the equalization current is calculated by multiplying the square of the current discharge battery's total voltage, the square of the current equalization topology's duty cycle, and the preset equalization period as the numerator, multiplying twice the preset inductance by the current charging battery pack's total voltage as the denominator, and calculating the ratio of the numerator to the denominator. The current direction is set to positive.

[0011] Furthermore, the vector sum of the main circuit charging and discharging current and the equalization current actually received by each individual cell is taken as the total effective current flowing through each individual cell. Based on the total effective current of each individual cell and the preset battery characteristic parameters, the Bernardi heat generation model is used to calculate the comprehensive heat generation rate of each individual cell. Based on the preset convective heat transfer coefficient, preset heat transfer coefficient, preset total effective contact area, average temperature of the battery and effective heat conduction distance of each individual battery cell, the conduction heat dissipation rate of each individual battery cell is calculated using Fourier's law of thermal conductivity. The conduction temperature difference used in the calculation is the sum of the average temperature of the battery and the temperature difference between the two sides of the battery. Based on the preset effective heat conduction distance, the average temperature of the battery and the ambient temperature, the convective heat dissipation rate of each individual battery cell is calculated using Newton's law of cooling. The convective temperature difference used in the calculation is the difference between the average temperature of the battery and the ambient temperature. The value after deducting the conductive heat dissipation rate and the convective heat dissipation rate is taken as the effective heat generation rate. The current temperature data of the single cell, the effective heat generation rate, the prediction step size, and the product of the mass and specific heat capacity of the single cell are used as inputs to the temperature prediction model to obtain the predicted temperature at the next moment.

[0012] Furthermore, based on the predicted state of charge (SOC) of each individual cell at the next time step, the average value of the predicted SOC of the battery pack at the next time step is calculated and used as the baseline for the predicted SOC. The difference between the predicted SOC of each individual cell at the next time step and the baseline for the predicted SOC is calculated. The difference between the predicted SOC of each individual cell and the baseline for the predicted SOC is squared and then summed. The summation result is then square-rooted to obtain the difference in the predicted SOC of the battery pack.

[0013] Furthermore, based on the predicted temperature of each individual cell at the next moment, the average predicted temperature of the battery pack at the next moment is calculated and used as the temperature prediction baseline. The difference between the predicted temperature of each individual cell at the next moment and the temperature prediction baseline is calculated. The difference between the predicted temperature of each individual cell and the temperature prediction baseline is squared and then summed. The summation result is then square-rooted to obtain the predicted temperature difference of the battery pack. The predicted temperature difference of the battery pack is calculated based on the predicted temperature of each individual cell at the next moment. The predicted state of charge difference and the predicted temperature difference are then scaled in a dimensionless manner using preset tolerance thresholds for state of charge and temperature as scaling standards, resulting in the processed predicted state of charge difference and predicted temperature difference.

[0014] Furthermore, based on the relative relationship between the processed predicted state of charge difference and the processed predicted temperature difference, dynamic benchmark parameters for balancing the convergence requirement of the state of charge and the temperature balance requirement are determined. The processed predicted temperature difference is compared with two preset comparison thresholds to determine the temperature difference range to which the processed predicted temperature difference belongs. Based on the temperature difference range, the dynamic reference parameters are constrained and adjusted, and the adaptive weighting coefficients of the state of charge and temperature are determined accordingly. Based on the adaptive weighting coefficients of the state of charge, the adaptive weighting coefficients of the temperature, the processed predicted state of charge difference, and the processed predicted temperature difference, a composite optimization objective function that simultaneously constrains the convergence of the state of charge and the temperature balance is constructed.

[0015] Furthermore, the process of determining the optimal control signal includes: Extract the initial population position coordinates generated by the fireworks algorithm within the physical feasible region, and normalize the initial population position coordinates according to the upper and lower bounds of the physical feasible region to obtain the normalized solution within the preset normalization interval. The normalized solution is iteratively updated by the Tent chaotic mapping function configured with preset morphological adjustment coefficients to obtain a new generation of chaotic solutions. Based on the upper and lower bounds of the physical feasible region, the new generation of chaotic solutions is denormalized to the physical feasible region to obtain the initial population position coordinates after chaos optimization. Using the initial population position coordinates optimized by chaos as the initial solution, the Fireworks Algorithm optimization process, which includes explosion, mutation and selection operations, is executed to generate continuous spatial position coordinates; The continuous spatial position coordinates generated in each iteration are converted into candidate control signals according to a preset discretization rule. The candidate control signals include three discrete states: discharge, bypass, and charge. Each candidate control signal is input into the state of charge prediction model and the temperature prediction model to obtain the corresponding prediction results. The composite optimization objective function value is calculated based on the prediction results, and the composite optimization objective function value is used as the fitness of each candidate control signal. When the number of iterations reaches the preset maximum number of iterations, or when the global optimal fitness has not been updated within a preset number of consecutive generations, the optimization process stops, and the candidate control signal corresponding to the current global optimal fitness is output as the optimal control signal.

[0016] Furthermore, based on the optimal control signal, the command state of the corresponding target single cell is controlled: If the optimal control signal corresponding to the target cell is 1, then the command state of the target cell is determined to be discharge, the mutual exclusion switch connected in parallel with the cell is disconnected, and a pulse width modulation signal is applied to the switch of the corresponding topology branch to control the cell to connect to the equalization topology and transfer the electrical energy to the flyover inductor for magnetic energy storage. If the optimal control signal corresponding to the target single cell is -1, then the command state of the target single cell is determined to be charging. The mutual exclusion switch connected in parallel with the target single cell is disconnected, and the switch tube of the corresponding topology branch is controlled to close, so that the magnetic energy stored in the flying inductor is converted into electrical energy and released into the single cell. If the optimal control signal corresponding to the target cell is 0, the command state of the target cell is determined to be bypass. The mutual exclusion switch connected in parallel with the target cell is closed, so that the cell is physically bypassed and isolated. The main circuit current flows through the mutual exclusion switch, so that the cell is simultaneously disconnected from the main charging and discharging circuit and the equalization topology network.

[0017] Compared with the prior art, the beneficial effects of the present invention are: This invention constructs a state-of-charge (SOC) prediction model and a temperature prediction model for individual cells based on acquired voltage, current, initial state of charge (POC), and temperature data. It combines the candidate control signal to be input with the duty cycle of the equalization topology, mapping it to the actual charging / discharging current and equalization current received by the battery. The predicted SOC and temperature for each individual cell at the next moment are then calculated. Based on this, the predicted SOC difference and predicted temperature difference are calculated and scaled without dimension. An adaptive weighting coefficient, whose magnitude is determined by the current temperature difference, is introduced to construct an optimization objective function for synchronously constraining SOC convergence and temperature balance. A fireworks algorithm with a population initialized via Tent chaotic mapping is used to generate candidate control signals. Iterative optimization is performed using the composite optimization objective function as the evaluation criterion to obtain the optimal control signal. This control logic overcomes the shortcomings of existing technologies that only focus on electrical parameters and ignore the thermal effect of equalization current. It physically decouples the charging and discharging circuit from the equalization circuit at the bottom layer of the model. Through an adaptive weighting mechanism, it actively constrains the temperature rise and temperature difference expansion while pursuing rapid convergence of the charge. This effectively suppresses the risk of aggravated heating and local overheating, and achieves dual-objective synergistic optimization of charge and temperature. This invention also controls the switching transistors of corresponding individual cells and the flyover inductors in the equalization topology to conduct based on the optimal control signal obtained through optimization, and controls the parallel mutual exclusion switches of the corresponding individual cells to close when the command state is bypass. Compared with the limitation of existing mainstream equalization schemes where energy needs to be transferred step by step when the target cells are far apart, the reconfigurable circuit of this invention, combined with the flyover inductor, realizes direct energy transfer between any individual cells, overcoming the path loss and efficiency reduction problems in long-distance equalization. At the same time, the hardware isolation mechanism based on the closed mutual exclusion switch allows for flexible physical bypassing of individual cells with excessively rapid temperature rise without adding additional power devices. This action completely removes the abnormally hot battery from the physical impact of the large current in the main charging and discharging circuit, and it no longer participates in the energy transfer of the flyover inductor, cutting off its continuous heat conduction path from the hardware level, realizing direct thermal isolation at the physical level, and improving the overall safety and temperature distribution uniformity of the battery pack. Attached Figure Description

[0018] Figure 1 This is a circuit diagram of the multi-path dynamic grouping equalization topology of the present invention; Figure 2 This is a schematic diagram of the overall method flow of the present invention; Figure 3 This is a schematic diagram of the data flow of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0020] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0021] Example: Please see Figures 1 to 3 The present invention provides a technical solution: A battery temperature control method based on a balanced topology is provided. The control method is applied to a multi-path dynamic grouping balanced topology, which includes: a battery pack consisting of multiple individual cells connected in series, a flyover inductor, and a reconfigurable circuit. The reconfigurable circuit includes a mutual exclusion switch matrix configured in parallel for each individual cell. In this embodiment, a method as follows is constructed Figure 1The diagram shows a multi-path dynamic grouping equalization topology circuit. The physical framework of this equalization topology is established by a main battery pack trunk circuit consisting of battery cells BT1 to BTn connected in series. To provide physical isolation to the trunk circuit, a pair of mutually exclusive switches is configured for each battery cell within the reconfigurable circuit region. For example, a first mutually exclusive switch S11 corresponding to battery cell BT1 is connected in series in the main circuit of battery cell BT1, and a second mutually exclusive switch S12 corresponding to battery cell BT1 is connected in parallel across this series branch. This series-parallel unit is cascaded to battery cell BTn and its corresponding first mutually exclusive switch Sn1 and second mutually exclusive switch Sn2, forming the trunk network. To achieve direct energy transfer, this equalization topology constructs an energy transfer channel consisting of an upper equalization bus, a lower equalization bus, and a bridging inductor L. Each connection node of the trunk network is connected to the equalization bus via a bidirectional switching assembly composed of two back-to-back, reverse-connected metal-oxide-semiconductor field-effect transistors. Taking the node where the positive electrode of the first single cell BT1 is located as an example, this node is connected to the upper equalization bus through an upper bridge arm assembly containing MOSFETs M11 and M12, and to the lower equalization bus through a lower bridge arm assembly containing MOSFETs M13 and M14. This pattern repeats along the backbone network to the MOSFETs M1 and M2, and MOSFETs M3 and M4 configured at the end nodes, forming a bidirectional switching array. The underlying control strategy, by coordinating the conduction of the above bidirectional switching components, directly connects single cells of any span to both ends of the flying inductor L, achieving low-loss point-to-point direct transmission. When the first single cell BT1 experiences abnormal heating during balancing, the control logic disconnects the first mutual exclusion switch S11 corresponding to the first single cell BT1 and closes the second mutual exclusion switch S12 corresponding to the first single cell BT1. This forces the large charging and discharging current to flow away through the second mutual exclusion switch S12 corresponding to the first single cell BT1, completely isolating the heated cell from the main circuit current impact and energy interaction with the balancing network. This cuts off the heat source from the underlying conductive path, improving the consistency and safety of the overall battery pack operation. Furthermore, this multi-path dynamic grouping balancing topology possesses extremely high physical scalability. When facing battery grouping requirements of different scales, new single cells can be connected simply by horizontally copying and adding the corresponding switching components along the backbone network according to the configuration rules of the aforementioned first mutual exclusion switch, second mutual exclusion switch, and bidirectional switch components. There is no need for structural reconstruction of the core flying inductor L and balancing bus, overcoming the technical bottleneck of the traditional balancing topology where the circuit complexity increases dramatically with the number of batteries. This allows it to adapt flexibly to various scales of energy storage applications.

[0022] Let's take a group of four individual battery cells as an example. Assume the first battery cell BT1 has a high state of charge (SOC), while the second and fourth batteries BT2 and BT4 have low SOCs, triggering an equalization adjustment. During the discharge phase of the equalization cycle, the pulse-width modulation (PWM) signal remains high, driving the source-side MOSFETs M12 and M24 to close. Current flows from the positive terminal of the first battery cell BT1, sequentially through MOSFETs M12 and M11, the flyover inductor L, MOSFETs M24 and M23, and the first mutual exclusion switch S11 corresponding to the first battery cell BT1, forming an excitation circuit. During this discharge phase, the first battery cell BT1 discharges, converting its electrical energy into magnetic energy, which is stored in the flyover inductor L. The inductor current flowing through the flyover inductor L shows a linear upward trend, reaching its peak at the end of this discharge phase. When the subsequent charging phase begins, the pulse width modulation signal drops to a low level, and the control command forces MOSFET M12 to turn off, cutting off the original excitation channel. Simultaneously, the first mutual exclusion switch S31 corresponding to the third cell BT3 is disconnected, and the MOSFET M2 at the receiving end is precisely closed. This continuous action connects the flyover inductor L with the second and fourth cells BT2 to form a freewheeling circuit. The polarity of the flyover inductor L reverses, releasing magnetic energy, causing the second and fourth cells BT2 to enter the charging state. During this charging phase, the inductor current gradually decreases according to a linear decay law. To avoid magnetic saturation induced by the flyover inductor L under high-frequency alternating conditions, the topology defaults to intermittent conduction mode, meaning a reserved dead time must be left after the discharge and charging phases. During the reserved dead time, the inductor current is strictly maintained at zero to ensure that the magnetic energy is completely discharged. Based on the physical boundary defense of this mode, the duty cycle actually applied by the control terminal must be constantly less than the ratio of the total voltage of the charging battery pack to the sum of the total voltage of the discharging battery pack and the total voltage of the charging battery pack.

[0023] The specific steps of the control method include: Step 1: Obtain the operating parameters of each individual cell in the balanced topology, calculate the current state of charge of each individual cell using the ampere-hour integration method, and determine whether the state of charge of the battery pack meets the conditions for balanced start-up. In this embodiment, the operating parameters of each individual cell in the battery pack in the balanced topology include real-time voltage, main circuit charging and discharging current, initial state of charge, temperature data and battery pack bus terminal charging and discharging current. The temperature data includes the average temperature of each individual cell, the temperature on both sides and the ambient temperature. Based on the real-time voltage of each individual battery, the charging and discharging current of the main circuit, the initial state of charge, and the preset rated capacity of each battery, the current state of charge of each individual battery is calculated using the ampere-hour integration method, with the time step between the current time and the last time the state of charge of each battery was calculated and fed back as a constraint. Calculate the difference between the current maximum and minimum state of charge (SOC) of each individual cell and compare it with a preset equalization start threshold. If the equalization start condition is met, proceed with subsequent steps for equalization adjustment; otherwise, determine that no further equalization adjustment is required for the current state. The equalization start condition is that the difference between the current maximum and minimum SOC is greater than or equal to the preset equalization start threshold.

[0024] To accurately perceive the true physical state of the multi-path dynamic grouping equilibrium topology under complex operating conditions and provide precise data support for subsequent electrothermal synergistic optimization, this method strictly defines the scope of multi-dimensional operational data acquisition in the initial stage of control logic. Specifically, this includes the real-time voltage of each individual cell within the battery pack, the charging and discharging current of the main circuit, the initial state of charge, the charging and discharging current at the battery pack bus terminal, and temperature data encompassing the average temperature of each individual cell, the temperatures on both sides, and the ambient temperature. Since the internal electrochemical reaction of a single cell exhibits a deep nonlinear coupling with its external thermodynamic performance, relying solely on a single electrical characteristic cannot reconstruct the true operational picture. Comprehensive acquisition of the above parameters and retention of detailed temperature gradient data directly constitutes the prerequisite boundary condition for subsequently constructing a high-precision heat transfer and convection cooling model. After acquiring the above multi-dimensional physical parameters, the control strategy uses the time step between the current time and the last time the state of charge of each battery was calculated and fed back as a mathematical constraint. Combined with the preset rated capacity of each battery, the ampere-hour integral method is used to continuously integrate and accumulate the real-time acquired main circuit charging and discharging current, thereby deducing the current state of charge of each individual cell. After obtaining the current state of charge (SOC) data of each individual battery cell, the control logic further extracts the maximum and minimum SOC values ​​from the current SOC set of each individual battery cell and calculates the difference between them. This difference is then rigorously compared with a preset equalization initiation threshold to determine whether to allow subsequent adjustment actions. Given that the SOC differences between individual batteries are extremely small, forced intervention in active energy transfer not only fails to bring substantial gains to overall consistency but also causes additional and significant switching losses and reactive power generation due to frequent high-frequency conduction of external switches. Establishing this hard initiation threshold can accurately filter out invalid equalization commands caused by minor state disturbances, ensuring that subsequent complex prediction and optimization mechanisms are activated only when internal differences genuinely exceed the threshold, suppressing unnecessary topology actions and additional heat effects from the source of operation.

[0025] Step 2: When the equalization start-up conditions are met, construct prediction models for the state of charge and temperature of individual cells respectively. Combine the candidate control signal to be input with the duty cycle of the equalization topology and map it into the actual charging and discharging current and equalization current received by the battery. Input the data into the two prediction models and calculate the predicted state of charge and predicted temperature of each individual cell at the next moment. In this embodiment, the main circuit charging and discharging current and equalization current currently actually received by each individual battery are obtained by combining the candidate control signal to be input with the duty cycle of the equalization topology and mapping them. By using the ampere-hour integration method, the change in discharge charge and the change in equalization charge of each individual cell are calculated based on the actual main circuit charging and discharging current and equalization current received by each individual cell, with the time step of the preset control cycle as a constraint. The current state of charge, charge and discharge changes, and equilibrium charge changes of each individual cell are used as inputs to the state of charge prediction model to obtain the predicted state of charge of each individual cell at the next moment. The principle of the prediction model for the state of charge is as follows: in, Indicates the first Predicting the battery state of charge at any given time. Indicates the first The state of charge of the battery at any given time. Indicates the time sequence number. This indicates that within the preset control cycle time step, the change in discharge charge of each individual battery cell is calculated using the ampere-hour integration method. The current equilibrium charge change of each individual cell is calculated using the ampere-hour integration method within the preset control cycle time step. The current main circuit charge / discharge current mapping logic actually received by each individual battery cell includes: If the current single cell control signal is a discharge signal, the main circuit charging and discharging current is the product of the acquired battery pack bus terminal charging and discharging current and the current equalization topology duty cycle, and the current direction is set to reverse. If the current single-cell control signal is a bypass connection signal, the main circuit charging current is directly mapped to 0; If the current single cell control signal is a charging signal, the main circuit charging and discharging current is the product of the acquired battery pack bus terminal charging and discharging current and the current equalization topology duty cycle, but the current direction is set to positive. The main circuit charging and discharging current mapping logic principle is as follows: in, This indicates the charging and discharging current of the battery's main circuit. This indicates the charging and discharging current at the battery pack bus terminal. This indicates the current equilibrium topology duty cycle. This indicates the current control signal for a single battery cell. When the control signal is -1, it indicates a charging signal; when the control signal is 0, it indicates a bypass connection signal; and when the control signal is 1, it indicates a discharging signal. The current equalization current mapping logic for each individual battery cell includes: If the current single cell control signal is a discharge signal, the equalization current is calculated by multiplying the total voltage of the current discharge cell, the square of the current equalization topology duty cycle, and the preset equalization period as the numerator, and using twice the preset inductance as the denominator, and calculating the ratio of the numerator to the denominator. The current direction is set to reverse. If the current single-cell control signal is connected to bypass, the equalization current is directly mapped to 0; If the current single-cell control signal is a charging signal, the equalization current is calculated by multiplying the square of the current total voltage of the discharging battery, the square of the current equalization topology duty cycle, and the preset equalization period as the numerator, multiplying twice the preset inductance by the current total voltage of the charging battery pack as the denominator, and calculating the ratio of the numerator to the denominator. The current direction is set to positive. The principle of the main circuit current balancing mapping logic is as follows: in, Indicates the equalization current. This indicates the total voltage of the currently discharging battery. This indicates the preset equilibrium period. This indicates the preset inductance value. This indicates the total voltage of the current charging battery pack.

[0026] In the aforementioned state of charge prediction model, to ensure a strict correspondence between the predicted charge transfer and the underlying physical switching actions, the control cycle time step is synchronized with the balancing cycle of the balancing topology. In practice, the control cycle time step is set to a positive integer multiple of the balancing cycle. Within one control cycle, an optimal control signal is locked and output once, and the underlying hardware continuously executes PWM switching actions for a corresponding positive integer multiple of a complete cycle accordingly. Given the significant control delay inherent in conventional passive state feedback mechanisms, this method constructs a state-of-charge (SOC) prediction model for individual cells to anticipate the physical consequences of candidate actions, enabling the optimization algorithm to predict the optimal energy transfer path before actually issuing the action command. This model uses the predicted SOC of each individual cell at the next moment as the dependent variable, intuitively reflecting the available energy level of the individual cell after a control cycle, providing data guidance for overall consistency convergence. To calculate this dependent variable, the model introduces the input control signal and the duty cycle of the balancing topology as core independent variables, establishing the quantitative coupling logic between the control end and the underlying topology. When the control signal is -1, representing the individual cell in a charging state receiving energy, the main circuit charging / discharging current is mapped to the product of the bus-end charging / discharging current and the duty cycle, taking a positive value. In this state, the mapping logic of the balancing current uses the product of the square of the total discharge battery voltage, the duty cycle, and the balancing cycle as the numerator, and the product of twice the inductance and the total voltage of the charging battery pack as the denominator, taking a positive ratio. The necessity of constructing this equalization current mapping logic lies in its accurate reproduction of the physical laws of energy transfer in the flying inductor under discontinuous mode, enabling mathematical deduction to accurately reflect the actual energy throughput of the hardware. When the control signal is 1, representing that the single cell is in a discharge state releasing energy, both the main circuit charging / discharging current and the equalization current are set to reverse and take negative values. If the control signal is 0, it means that the single cell is physically bypassed, and the main circuit charging / discharging current and the equalization current are directly mapped to 0. After obtaining the actual received main circuit charging / discharging current and equalization current, the change in charging / discharging charge and the change in equalization charge are calculated using the ampere-hour integral method with the time step of the control cycle as a constraint, and then superimposed with the current initial state of charge to construct a complete prediction model. Among them, the dependent variable predicted state of charge is positively correlated with the initial state of charge as the independent variable and the positive charge change representing energy injection, and negatively correlated with the negative charge change representing energy extraction. This mapping process transforms abstract instructions into substantial charge dynamics, providing data guidance for overall consistency convergence.

[0027] The vector sum of the main circuit charging / discharging current and equalization current actually received by each individual cell is taken as the total effective current flowing through each individual cell. Based on the total effective current of each individual cell and the preset battery characteristic parameters, the Bernardi heat generation model is used to calculate the comprehensive heat generation rate of each individual cell. The preset battery characteristic parameters include the preset battery ohmic internal resistance, the preset battery polarization internal resistance, the average temperature of the battery, and the temperature entropy coefficient. The calculation principle for the overall heat generation rate of a single battery cell is as follows: in, This indicates the overall heat generation rate of a single cell. This represents the total effective current of a single cell. This indicates the preset internal resistance of the battery in ohms. This indicates the preset battery polarization internal resistance. This indicates the average temperature of the battery. It represents the temperature entropy coefficient, which is determined by testing the thermodynamic open-circuit voltage of a single cell under different temperature gradients. Based on the preset convective heat transfer coefficient, preset heat transfer coefficient, preset total effective contact area, average temperature of the battery and effective heat conduction distance of each individual battery cell, the conduction heat dissipation rate of each individual battery cell is calculated using Fourier's law of thermal conductivity. The conduction temperature difference used in the calculation is the sum of the average temperature of the battery and the temperature difference between the two sides of the battery. Based on the preset effective heat conduction distance, the average temperature of the battery and the ambient temperature, the convective heat dissipation rate of each individual battery cell is calculated using Newton's law of cooling. The convective temperature difference used in the calculation is the difference between the average temperature of the battery and the ambient temperature. The value after deducting the conductive heat dissipation rate and the convective heat dissipation rate is used as the effective heat generation rate. The current temperature data of the single cell, the effective heat generation rate, the prediction step size, and the product of the mass and specific heat capacity of the single cell are used as inputs to the temperature prediction model to obtain the predicted temperature at the next moment. The principle behind the temperature prediction model is as follows: in, Indicates the first Predicted temperature of individual cells at any given time. Indicates the first The temperature of a single cell at any given time. This indicates the thermal conductivity of a single battery cell. This indicates the convective heat dissipation rate of a single battery cell. This indicates the time step of the preset control cycle.

[0028] Considering the thermal effects generated by the main circuit charging / discharging current and equalization current accompanying high-rate charging / discharging and energy transfer, relying solely on electrical level deductions can easily lead to localized heating due to the pursuit of consistent charge levels. Therefore, this method simultaneously constructs a single-cell temperature prediction model that interacts with the electrical model. The purpose of constructing this thermal deduction logic is to enable the optimization algorithm to quantitatively assess in advance whether the currently allocated main circuit charging / discharging current and equalization current will touch the thermal safety boundary, thereby actively suppressing the temperature rise while pursuing charge convergence. This model uses the predicted temperature of each single cell at the next moment as the core dependent variable, clearly characterizing the thermodynamic state of the single cell at future time points, providing a direct evaluation benchmark for constraining localized heating. In terms of variable transmission relationships, the main circuit charging / discharging current and equalization current obtained from the aforementioned mapping are added to obtain the total effective current. This, along with the preset battery ohmic internal resistance and polarization internal resistance, are used as independent variables driving heat generation, and substituted into the heat generation model to calculate the comprehensive heat generation rate of the single cell. Simultaneously, the difference between the average temperature of a single cell and the temperatures on both sides is substituted into the law of thermal conductivity to calculate the conductive heat dissipation rate, and the difference between the average temperature and the ambient temperature is substituted into the cooling law to calculate the convective heat dissipation rate. After establishing all thermal components, the difference between the overall heat generation rate and the sum of the two heat dissipation rates is calculated to obtain the net heat generation rate. This rate is then divided by the product of the mass and specific heat capacity of the single cell, multiplied by the time step, and added to the current average temperature to obtain the predicted temperature. In this model, the dependent variable, predicted temperature, is positively correlated with the total effective current, various internal resistances, and the overall heat generation rate. Conversely, it is negatively correlated with the conductive and convective heat dissipation rates, which characterize heat dissipation. This multiphysics collaborative model quantifies the combined impact of action-induced heat generation and environmental heat exchange on the temperature rise of a single cell, overcoming the technical deficiency of traditional single-objective control methods that ignore superimposed thermal effects.

[0029] By using the state-of-charge prediction model and temperature prediction model of individual cells, the predicted state of charge and predicted temperature of each individual cell at the next moment after the time step of the control cycle are calculated.

[0030] Step 3: Calculate the difference between the predicted state of charge and the predicted temperature based on the predicted state of charge and the predicted temperature at the next moment. Introduce an adaptive weighting coefficient whose magnitude is determined by the current temperature difference in segments, and construct an optimization objective function to simultaneously constrain the convergence of the state of charge and the temperature balance. In this embodiment, based on the predicted state of charge of each individual cell at the next time step, the average value of the predicted state of charge of the battery pack at the next time step is calculated and used as the prediction baseline of the state of charge. The difference between the predicted state of charge of each individual cell at the next time step and the prediction baseline of the state of charge is calculated. The difference between the predicted state of charge of each individual cell and the prediction baseline of the state of charge is squared and then summed. The summation result is then square-rooted to obtain the difference in the predicted state of charge of the battery pack. Based on the predicted temperature of each individual cell at the next moment, the average predicted temperature of the battery pack at the next moment is calculated and used as the temperature prediction baseline. The difference between the predicted temperature of each individual cell at the next moment and the temperature prediction baseline is calculated. The difference between the predicted temperature of each individual cell and the temperature prediction baseline is squared and then summed. The summation result is then square-rooted to obtain the predicted temperature difference of the battery pack. The predicted temperature difference of the battery pack is calculated based on the predicted temperature of each individual cell at the next moment. The predicted state of charge difference and the predicted temperature difference are scaled in a dimensionless manner based on the preset state of charge tolerance threshold and temperature tolerance threshold as scaling standards, so as to obtain the processed predicted state of charge difference and predicted temperature difference. To quantify the degree of differentiation in internal battery parameters after executing candidate control commands in a multi-path dynamic grouping equilibrium topology, this embodiment constructs a differentiation evaluation formula for the overall state of the battery pack. In the derivation logic, the average value of the predicted state of charge (SOC) of each individual battery cell at the next moment is calculated as the independent variable and used as the predicted SOC baseline. The difference between the predicted SOC of each individual battery cell and the predicted baseline is calculated, and the squares of each difference are summed and the square root is taken to obtain the predicted SOC difference. The predicted temperature difference is derived using the same dispersion evaluation formula. The significance of calculating the predicted SOC difference and the predicted temperature difference lies in intuitively reflecting the degree of inconsistency in the distribution of charge and heat in the battery pack over future periods. Since the predicted SOC is expressed as a percentage value while the predicted temperature is expressed as a degree Celsius value, there are significant differences in their physical dimensions and numerical magnitudes. Directly superimposing the predicted SOC difference and the predicted temperature difference for optimization evaluation would dilute the impact of smaller parameters on the results due to the larger parameter magnitude. To eliminate differences in physical dimensions and magnitudes, the derivation logic introduces preset tolerance thresholds for state of charge (SOC) and temperature as independent variables. Division operations are then performed on the predicted SOC and temperature differences to achieve dimensionless scaling, ultimately yielding the processed predicted SOC and temperature differences as dependent variables. These processed SOC and temperature differences clearly characterize the degree of deviation after normalization, establishing a unified mathematical benchmark for subsequent dual-objective collaborative evaluation. During the calculation, the processed predicted SOC and temperature differences are positively correlated with their corresponding predicted SOC and temperature differences; the more severe the deviation between individual cells, the larger the output value of the dependent variables. Conversely, these two dependent variables are negatively correlated with the tolerance thresholds used as the denominator; the more lenient the tolerance thresholds, the smaller the output value of the dependent variables. This standardized scaling logic flattens physical quantities of different dimensions into a unified mathematical evaluation range.

[0031] Based on the relative relationship between the processed predicted state of charge difference and the processed predicted temperature difference, dynamic benchmark parameters for balancing the convergence requirement of state of charge and the temperature balance requirement are determined. The processed predicted temperature difference is compared with two preset comparison thresholds to determine the temperature difference range to which the processed predicted temperature difference belongs. Based on the temperature difference range, the dynamic reference parameters are constrained and adjusted, and the adaptive weighting coefficients of the state of charge and temperature are determined accordingly. Based on the adaptive weighting coefficient of the state of charge, the adaptive weighting coefficient of the temperature, the processed predicted state of charge difference, and the processed predicted temperature difference, a composite optimization objective function that simultaneously constrains the convergence of the state of charge and the temperature balance is constructed. The principle of the composite optimization objective function is as follows: in, Indicates the adaptive weighting coefficients for the state of charge. This represents the temperature-adaptive weighting coefficient. This indicates the preset adjustment coefficient. This indicates the difference in predicted state of charge. Indicates the predicted temperature difference. This represents the composite optimization objective function. This represents the function that takes the minimum value. The function that takes the maximum value is the one that takes the minimum value in the composite optimization objective function. This means that in each iteration, a comparison is made and the smaller value is taken after each iteration.

[0032] After unifying the physical dimensions, to overcome the technical deadlock of conventional fixed-weight evaluation models easily inducing single-boundary exceedances under complex operating conditions, this embodiment further constructs a composite optimization objective function that includes mapping logic for adaptive weight coefficients of state of charge and temperature. The purpose of constructing the mapping logic for adaptive weight coefficients of state of charge and temperature is to guide the control action to achieve a dynamic compromise between rapid equalization of charge and suppression of local heating. The composite optimization objective function, as the final dependent variable, directly determines the comprehensive judgment result of the optimization algorithm on the merits of the current candidate control commands. To generate the adaptive weight coefficients of state of charge and temperature, the formula introduces the ratio of the processed predicted state of charge difference to the sum of the processed predicted temperature difference as a dynamic benchmark ratio. Since the heating risk of a single battery cell varies non-linearly in different temperature ranges, the algorithm uses the processed predicted temperature difference as the core independent variable to make segmented judgments on the dynamic benchmark ratio. If the processed predicted temperature difference is within a small safe range, a wider upper and lower limit range is assigned to the dynamic benchmark ratio for comparison. As the processed predicted temperature difference crosses preset first comparison thresholds, second comparison thresholds, and even third comparison thresholds, the corresponding upper and lower limit constraint intervals are progressively tightened, thereby lowering the output upper limit of the dynamic benchmark ratio. The final result extracted through segmented limiting is multiplied by a preset adjustment coefficient to obtain the state-of-charge (POC) adaptive weighting coefficient, and the temperature adaptive weighting coefficient is obtained by subtracting the POC adaptive weighting coefficient from the value 1. Subsequently, the POC adaptive weighting coefficient and the temperature adaptive weighting coefficient are multiplied by the corresponding processed predicted POC difference and processed predicted temperature difference, respectively, and then superimposed to obtain the composite optimization objective function. In the mapping relationship, the degree of deterioration of the processed predicted temperature difference is negatively correlated with the POC adaptive weighting coefficient in a stepwise manner, while the processed predicted temperature difference is positively correlated with the temperature adaptive weighting coefficient. The construction logic of segmented clamping of charge state adaptive weight coefficient and temperature adaptive weight coefficient based on temperature difference ensures that when the temperature difference is small, a high weight is given to charge balance to accelerate charge convergence, while when the temperature difference worsens, the control logic will actively reduce the weight of charge balance and tilt towards temperature constraint, thus eliminating the risk of aggravating local heating due to blindly pursuing charge consistency.

[0033] Step 4: Using the fireworks algorithm that initializes the population with Tent chaotic mapping, candidate control signals are generated and the prediction models of charge state and temperature are called. Iterative optimization is performed based on the composite optimization objective function as the evaluation criterion to obtain the optimal control signal. In this embodiment, the process of determining the optimal control signal includes: Extract the initial population position coordinates generated by the fireworks algorithm within the physical feasible region, and normalize the initial population position coordinates according to the upper and lower bounds of the physical feasible region to obtain the normalized solution within the preset normalization interval. The normalized solution is iteratively updated by the Tent chaotic mapping function configured with preset morphological adjustment coefficients to obtain a new generation of chaotic solutions. Based on the upper and lower bounds of the physical feasible region, the new generation of chaotic solutions is denormalized to the physical feasible region to obtain the initial population position coordinates after chaos optimization. Using the initial population position coordinates optimized by chaos as the initial solution, the Fireworks Algorithm optimization process, which includes explosion, mutation and selection operations, is executed to generate continuous spatial position coordinates; The continuous spatial position coordinates generated in each iteration are converted into candidate control signals according to a preset discretization rule. The candidate control signals include three discrete states: discharge, bypass, and charge. Each candidate control signal is input into the state of charge prediction model and the temperature prediction model to obtain the corresponding prediction results. The composite optimization objective function value is calculated based on the prediction results, and the composite optimization objective function value is used as the fitness of each candidate control signal. When the number of iterations reaches the preset maximum number of iterations, or when the global optimal fitness has not been updated within a preset number of consecutive generations, the optimization process stops, and the candidate control signal corresponding to the current global optimal fitness is output as the optimal control signal.

[0034] The logic for discretizing the continuous spatial position coordinates into candidate control signals including discharge, bypass, and charging states is as follows: If the generated continuous spatial position coordinate values ​​are less than or equal to -0.5, they are discretized into discharge signals; If the generated continuous spatial position coordinate values ​​are greater than -0.5 and less than 0.5, they are discretized into bypass connection signals; If the generated continuous spatial position coordinates are greater than or equal to 0.5, they are discretized into charging signals.

[0035] After establishing the composite optimization objective function, this embodiment introduces a fireworks algorithm with a population initialized using Tent chaotic mapping to perform global optimization operations on candidate control signals. Conventional random initialization methods easily lead to an imbalance in the initial population distribution within the solution space, weakening the algorithm's convergence efficiency. Introducing Tent chaotic mapping to generate the initial population of the fireworks algorithm with a population initialized using Tent chaotic mapping endows the initial candidate control signal set with ergodicity and uniformity within the multidimensional discrete control space. The mathematical characteristics of Tent chaotic mapping avoid the risk of the optimization operation getting trapped in local optima in the early stages of iteration. In the iterative process of the fireworks algorithm with a population initialized using Tent chaotic mapping, each individual fireworks strictly corresponds to a specific set of candidate control signals. The computational logic substitutes each candidate control signal into the physical and mathematical derivation process to obtain the corresponding composite optimization objective function value, and uses the composite optimization objective function value as a fitness index to measure the quality of individual fireworks. Based on the fitness index, the fireworks algorithm with a population initialized using Tent chaotic mapping dynamically allocates the explosion radius and the number of explosion sparks generated for each individual fireworks. Fireworks with smaller composite optimization objective function values ​​can generate a large number of explosive sparks within a smaller blast radius for local depth mining, while fireworks with larger composite optimization objective function values ​​generate a small number of explosive sparks within a wider blast radius to perform large-scale global exploration. With the continuous alternation of explosive sparks and Gaussian mutation sparks, the candidate control signal population undergoes an evolutionary process. When the iteration count reaches the preset maximum number of iterations, the operational logic extracts the set of candidate control signals with the smallest corresponding composite optimization objective function value as the final output control command. A fireworks algorithm that initializes the population using Tent chaotic mapping is employed to obtain candidate control signals, transforming multi-objective collaborative physical parameters into a heuristic mathematical optimization process. This ensures that the finally selected candidate control signals can accurately drive the actions of each switching component in the multi-path dynamic grouping equilibrium topology, suppressing the heating tendency of individual cells while exploring the potential for direct energy transmission.

[0036] Step 5: According to the optimal control signal, control the switching transistor and flying inductor of the corresponding single cell in the balanced topology to turn on, and control the mutual exclusion switch connected in parallel with the corresponding single cell to close when the command state is bypass. In this embodiment, the command state of the corresponding target single battery cell is controlled according to the optimal control signal: If the optimal control signal corresponding to the target cell is 1, then the command state of the target cell is determined to be discharge, the mutual exclusion switch connected in parallel with the cell is disconnected, and a pulse width modulation signal is applied to the switch of the corresponding topology branch to control the cell to connect to the equalization topology and transfer the electrical energy to the flyover inductor for magnetic energy storage. If the optimal control signal corresponding to the target single cell is -1, then the command state of the target single cell is determined to be charging. The mutual exclusion switch connected in parallel with the target single cell is disconnected, and the switch tube of the corresponding topology branch is controlled to close, so that the magnetic energy stored in the flying inductor is converted into electrical energy and released into the single cell. If the optimal control signal corresponding to the target cell is 0, the command state of the target cell is determined to be bypass. The mutual exclusion switch connected in parallel with the target cell is closed, so that the cell is physically bypassed and isolated. The main circuit current flows through the mutual exclusion switch, so that the cell is simultaneously disconnected from the main charging and discharging circuit and the equalization topology network.

[0037] After obtaining the final control command output by the fireworks algorithm after initializing the population using the Tent chaotic mapping, the underlying logic precisely maps the various control signals contained in the final control command to the physical drive levels of each switching component in the multi-path dynamic grouping equalization topology circuit. When the control signal corresponding to a certain single cell is 1, it indicates that the single cell is in a discharge state of discharging energy. The underlying logic maintains the first mutual exclusion switch corresponding to the single cell in the closed state and generates a pulse width modulation signal according to the duty cycle to drive the bidirectional switch component corresponding to the single cell to conduct, so that the single cell is connected to the upper equalization bus and the lower equalization bus to transfer electrical energy to the flyover inductor L. When the control signal corresponding to a certain single cell is -1, it indicates that the single cell is in a charging state of receiving energy. The underlying logic maintains the first mutual exclusion switch corresponding to the single cell in the closed state and controls the corresponding bidirectional switch component to conduct during the discharge stage, guiding the magnetic energy released by the flyover inductor L to be injected into the single cell. If the control signal for a specific battery cell is 0, it indicates that the battery cell has triggered thermal boundary constraints and needs to enter a forced bypass sleep state. The underlying logic issues a command to disconnect the first mutual exclusion switch of the main circuit corresponding to that battery cell and simultaneously close the second mutual exclusion switch connected in parallel, while keeping the bidirectional switch component corresponding to that battery cell in the off state. This mapping mechanism, which transforms various control signals into underlying hardware actions, allows the abstract mathematical optimization results to be accurately implemented in the physical entity. For battery cells with a control signal of 0, hardware bypass reconstruction is performed, which forcibly guides the charging and discharging current of the battery pack's main circuit to flow through the low-impedance second mutual exclusion switch, giving abnormally hot battery cells a dual isolation state from large current impacts and energy interactions. This execution mechanism cuts off the source of continuous heat generation from the battery cell in the physical conductive path, and, in conjunction with the cross-point-to-point direct control of the other battery cells, achieves the comprehensive goal of balancing rapid charge convergence and suppression of local temperature rise in the physical circuit, improving the overall charge consistency and thermal safety of the battery pack.

[0038] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0039] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0040] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0041] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A battery temperature control method based on balanced topology, characterized in that, The specific steps include: Step 1: Obtain the operating parameters of each individual cell in the balanced topology, calculate the current state of charge of each individual cell using the ampere-hour integration method, and determine whether the state of charge of the battery pack meets the conditions for balanced start-up. Step 2: When the equalization start-up conditions are met, construct prediction models for the state of charge and temperature of individual cells respectively. Combine the candidate control signal to be input with the duty cycle of the equalization topology and map it into the actual charging and discharging current and equalization current received by the battery. Input the data into the two prediction models and calculate the predicted state of charge and predicted temperature of each individual cell at the next moment. Step 3: Calculate the difference between the predicted state of charge and the predicted temperature based on the predicted state of charge and the predicted temperature at the next moment. Introduce an adaptive weighting coefficient whose magnitude is determined by the current temperature difference in segments, and construct an optimization objective function to simultaneously constrain the convergence of the state of charge and the temperature balance. Step 4: Using the fireworks algorithm that initializes the population with Tent chaotic mapping, candidate control signals are generated and the prediction models of charge state and temperature are called. Iterative optimization is performed based on the composite optimization objective function as the evaluation criterion to obtain the optimal control signal. Step 5: According to the optimal control signal, control the switching transistor of the corresponding single cell in the balanced topology to conduct with the flying inductor, and control the parallel mutual exclusion switch of the corresponding single cell to close when the command state is bypass.

2. The battery temperature control method based on balanced topology according to claim 1, characterized in that: In obtaining the balanced topology, the operating parameters of each cell in the battery pack include real-time voltage, main circuit charging and discharging current, initial state of charge, temperature data, and battery pack bus terminal charging and discharging current. Among them, the temperature data includes the average temperature of each cell, the temperature on both sides, and the ambient temperature. Based on the real-time voltage of each individual battery, the charging and discharging current of the main circuit, the initial state of charge, and the preset rated capacity of each battery, the current state of charge of each individual battery is calculated using the ampere-hour integration method, with the time step between the current time and the last time the state of charge of each battery was calculated and fed back as a constraint. Calculate the difference between the current maximum and minimum state of charge (SOC) of each individual cell and compare it with a preset equalization start threshold. If the equalization start condition is met, proceed with subsequent steps for equalization adjustment; otherwise, determine that no further equalization adjustment is required for the current state. The equalization start condition is that the difference between the current maximum and minimum SOC is greater than or equal to the preset equalization start threshold.

3. The battery temperature control method based on balanced topology according to claim 2, characterized in that: By combining the candidate control signal to be input with the duty cycle of the equalization topology, the actual main circuit charging and discharging current and equalization current currently received by each individual cell are obtained through mapping. By using the ampere-hour integration method, the change in discharge charge and the change in equalization charge of each individual cell are calculated based on the actual main circuit charging and discharging current and equalization current received by each individual cell, with the time step of the preset control cycle as a constraint. The current state of charge, charge / discharge change, and equilibrium charge change of each individual cell are used as inputs to the state of charge prediction model to obtain the predicted state of charge of each individual cell at the next moment.

4. The battery temperature control method based on balanced topology according to claim 3, characterized in that: The current main circuit charge / discharge current mapping logic actually received by each individual battery cell includes: If the current single cell control signal is a discharge signal, the main circuit charging and discharging current is the product of the acquired battery pack bus terminal charging and discharging current and the current equalization topology duty cycle, and the current direction is set to reverse. If the current single-cell control signal is a bypass connection signal, the main circuit charging current is directly mapped to 0; If the current single cell control signal is a charging signal, the main circuit charging and discharging current is the product of the acquired battery pack bus terminal charging and discharging current and the current equalization topology duty cycle, but the current direction is set to positive. The current equalization current mapping logic for each individual battery cell includes: If the current single cell control signal is a discharge signal, the equalization current is calculated by multiplying the total voltage of the current discharge cell, the square of the current equalization topology duty cycle, and the preset equalization period as the numerator, and using twice the preset inductance as the denominator, and calculating the ratio of the numerator to the denominator. The current direction is set to reverse. If the current single-cell control signal is connected to bypass, the equalization current is directly mapped to 0; If the current single-cell control signal is a charging signal, the equalization current is calculated by multiplying the square of the current discharge battery's total voltage, the square of the current equalization topology's duty cycle, and the preset equalization period as the numerator, multiplying twice the preset inductance by the current charging battery pack's total voltage as the denominator, and calculating the ratio of the numerator to the denominator. The current direction is set to positive.

5. The battery temperature control method based on balanced topology according to claim 4, characterized in that: The vector sum of the main circuit charging and discharging current and the equalization current actually received by each individual cell is taken as the total effective current flowing through each individual cell. Based on the total effective current of each individual cell and the preset battery characteristic parameters, the comprehensive heat generation rate of each individual cell is calculated using the Bernardi heat generation model. Based on the preset convective heat transfer coefficient, preset heat transfer coefficient, preset total effective contact area, average temperature of the battery and effective heat conduction distance of each individual battery cell, the conduction heat dissipation rate of each individual battery cell is calculated using Fourier's law of thermal conductivity. The conduction temperature difference used in the calculation is the sum of the average temperature of the battery and the temperature difference between the two sides of the battery. Based on the preset effective heat conduction distance, the average temperature of the battery and the ambient temperature, the convective heat dissipation rate of each individual battery cell is calculated using Newton's law of cooling. The convective temperature difference used in the calculation is the difference between the average temperature of the battery and the ambient temperature. The value after deducting the conductive heat dissipation rate and the convective heat dissipation rate is taken as the effective heat generation rate. The current temperature data of the single cell, the effective heat generation rate, the prediction step size, and the product of the mass and specific heat capacity of the single cell are used as inputs to the temperature prediction model to obtain the predicted temperature at the next moment.

6. The battery temperature control method based on balanced topology according to claim 5, characterized in that: Based on the predicted state of charge (SOC) of each individual cell at the next time step, the average predicted SOC of the battery pack at the next time step is calculated and used as the baseline for predicting SOC. The difference between the predicted SOC of each individual cell at the next time step and the baseline for predicting SOC is calculated. The difference between the predicted SOC of each individual cell and the baseline for predicting SOC is squared and then summed. The summation result is then square-rooted to obtain the difference in predicted SOC of the battery pack.

7. The battery temperature control method based on balanced topology according to claim 6, characterized in that: Based on the predicted temperature of each individual cell at the next moment, the average predicted temperature of the battery pack at the next moment is calculated and used as the temperature prediction baseline. The difference between the predicted temperature of each individual cell at the next moment and the temperature prediction baseline is calculated. The difference between the predicted temperature of each individual cell and the temperature prediction baseline is squared and then summed. The summation result is then square-rooted to obtain the predicted temperature difference of the battery pack. The predicted temperature difference of the battery pack is calculated based on the predicted temperature of each individual cell at the next moment. The predicted state of charge difference and the predicted temperature difference are then scaled in a dimensionless manner using preset tolerance thresholds for state of charge and temperature as scaling standards, resulting in the processed predicted state of charge difference and predicted temperature difference.

8. The battery temperature control method based on balanced topology according to claim 7, characterized in that: Based on the relative relationship between the processed predicted state of charge difference and the processed predicted temperature difference, dynamic benchmark parameters for balancing the convergence requirement of state of charge and the temperature balance requirement are determined. The processed predicted temperature difference is compared with two preset comparison thresholds to determine the temperature difference range to which the processed predicted temperature difference belongs. Based on the temperature difference range, the dynamic reference parameters are constrained and adjusted, and the adaptive weighting coefficients of the state of charge and temperature are determined accordingly. Based on the adaptive weighting coefficients of the state of charge, the adaptive weighting coefficients of the temperature, the processed predicted state of charge difference, and the processed predicted temperature difference, a composite optimization objective function that simultaneously constrains the convergence of the state of charge and the temperature balance is constructed.

9. A battery temperature control method based on balanced topology according to claim 8, characterized in that: The process of determining the optimal control signal includes: Extract the initial population position coordinates generated by the fireworks algorithm within the physical feasible region, and normalize the initial population position coordinates according to the upper and lower bounds of the physical feasible region to obtain the normalized solution within the preset normalization interval. The normalized solution is iteratively updated by the Tent chaotic mapping function configured with preset morphological adjustment coefficients to obtain a new generation of chaotic solutions. Based on the upper and lower bounds of the physical feasible region, the new generation of chaotic solutions is denormalized to the physical feasible region to obtain the initial population position coordinates after chaos optimization. Using the initial population position coordinates optimized by chaos as the initial solution, the Fireworks Algorithm optimization process, which includes explosion, mutation and selection operations, is executed to generate continuous spatial position coordinates; The continuous spatial position coordinates generated in each iteration are converted into candidate control signals according to a preset discretization rule. The candidate control signals include three discrete states: discharge, bypass, and charge. Each candidate control signal is input into the state of charge prediction model and the temperature prediction model to obtain the corresponding prediction results. The composite optimization objective function value is calculated based on the prediction results, and the composite optimization objective function value is used as the fitness of each candidate control signal. When the number of iterations reaches the preset maximum number of iterations, or when the global optimal fitness has not been updated within a preset number of consecutive generations, the optimization process stops, and the candidate control signal corresponding to the current global optimal fitness is output as the optimal control signal.

10. A battery temperature control method based on balanced topology according to claim 9, characterized in that: Based on the optimal control signal, control the command state of the corresponding target single cell: If the optimal control signal corresponding to the target cell is 1, then the command state of the target cell is determined to be discharge, the mutual exclusion switch connected in parallel with the cell is disconnected, and a pulse width modulation signal is applied to the switch of the corresponding topology branch to control the cell to connect to the equalization topology and transfer the electrical energy to the flyover inductor for magnetic energy storage. If the optimal control signal corresponding to the target single cell is -1, then the command state of the target single cell is determined to be charging. The mutual exclusion switch connected in parallel with the target single cell is disconnected, and the switch tube of the corresponding topology branch is controlled to close, so that the magnetic energy stored in the flying inductor is converted into electrical energy and released into the single cell. If the optimal control signal corresponding to the target cell is 0, the command state of the target cell is determined to be bypass. The mutual exclusion switch connected in parallel with the target cell is closed, so that the cell is physically bypassed and isolated. The main circuit current flows through the mutual exclusion switch, so that the cell is simultaneously disconnected from the main charging and discharging circuit and the equalization topology network.

Images