A three-mode predictive switching method and system based on battery terminal voltage equalization topology

Abstract

This invention relates to the field of mode prediction and switching technology, and in particular to a three-mode prediction and switching method and system based on battery terminal voltage-based equalization topology. The method includes constructing three equalization modes based on the built battery equalization topology: cell-to-cell equalization mode, cell-to-pack equalization mode, and pack-to-cell equalization mode. Mode selection is performed by fusing improved Kalman filter dynamic parameter estimation and rolling time-domain optimization, including terminal voltage detection and calculation, voltage transfer parameter estimation, multi-step prediction based on rolling time-domain optimization, and optimal mode selection. This invention significantly reduces the number of components, circuit complexity, and overall size, lowers manufacturing costs and system weight, while reducing potential failure points and improving system reliability and maintainability.

Description

Technical Field

[0001] This invention relates to the field of mode prediction and switching technology, and in particular to a three-mode prediction and switching method and system for an equalization topology based on battery terminal voltage. Background Technology

[0002] Against the backdrop of the accelerated global energy transition towards a green and low-carbon structure, lithium-ion batteries have become the core driving force for the development of various fields such as electric vehicles and energy storage due to their significant advantages, including high energy density, long cycle life, and low self-discharge rate. Because the voltage and capacity of individual battery cells cannot meet the requirements of applications such as electric vehicles, hundreds of battery cells are typically connected in series to meet application needs. However, due to subtle differences in manufacturing processes and variations in operating environments, loads, and usage habits, inconsistencies inevitably exist among individual battery cells in terms of initial capacity, internal resistance, and aging rate. Without effective management, these inconsistencies will gradually accumulate and amplify during cyclic use, causing some cells to be prematurely charged or discharged, resulting in reduced usable capacity, shortened system range, and significantly impacting the overall performance and lifespan of the battery pack. More seriously, inconsistencies in voltage and capacity can lead to overcharging or over-discharging of individual cells during charging and discharging, accelerating internal chemical side reactions and structural degradation. This not only shortens battery life but may also cause thermal runaway, and even lead to safety accidents such as combustion and explosion, posing a serious threat to the entire battery system and even personal safety and property.

[0003] Currently, battery balancing technologies can be mainly divided into passive balancing and active balancing. Passive balancing involves connecting a discharge resistor in parallel across the terminals of a battery cell with a higher voltage, dissipating excess energy as heat to passively level the voltage. This method has a simple circuit structure and low cost, and is therefore still widely used in many current battery management systems. However, it has low energy utilization efficiency, slow balancing speed, and the continuous heating of the resistor during the balancing process can cause local temperature rise in the battery, potentially affecting battery life and system thermal safety. Active balancing utilizes energy storage components such as inductors, transformers, or capacitors, combined with power electronic switching circuits, to achieve directional energy transfer between battery cells or between cells and the entire battery pack. It has high energy utilization and low heat generation, but current active balancing technologies often suffer from problems such as a single balancing mode, large size, and the use of a large number of switching transistors. In related technologies, patent CN202510838157 proposes a dual-layer topology equalization circuit based on a dual multi-winding flyback transformer. However, each battery pack requires two transformer windings, and the battery pack also needs to be equipped with three inductors, resulting in an increase in size. Patent CN202510378061 proposes a battery pack equalization topology, but this structure cannot achieve equalization from battery cell to battery cell. When the voltage distribution of battery cells is relatively uniform, point-to-point energy transfer cannot be achieved, resulting in poor flexibility.

[0004] Specifically, the following problems exist in current active battery balancing technology: (1) Complex topology Currently, equalization schemes based on multi-independent winding transformers often require a transformer winding and switching transistor with a number similar to the number of battery cells to achieve equalization between any individual cells. This results in complex circuit topology, a large number of components, and increased system size and weight, which not only increases costs but also introduces more potential failure points.

[0005] (2) Equilibrium pattern and single path Currently, most topologies have fixed balancing paths and lack the ability to flexibly select multiple modes. For example, they only support unidirectional energy transfer between adjacent cells or between a cell and the entire battery pack. The imbalance state within the battery pack is more complex. If a suitable balancing path cannot be used, it will lead to increased balancing time and reduced efficiency. Summary of the Invention

[0006] To address the aforementioned problems, this invention provides a method and system for predictive switching of three modes in an equalization topology based on battery terminal voltage.

[0007] In a first aspect, the present invention provides a three-mode predictive switching method for an equalization topology based on battery terminal voltage, which adopts the following technical solution: A three-mode predictive switching method for an equalization topology based on battery terminal voltage includes: Obtain the basic and operational parameters of the battery balancing topology; A battery balancing topology is constructed based on the acquired parameters; Based on the constructed battery balancing topology, three balancing modes are built, including battery cell to battery cell balancing mode, battery cell to battery pack balancing mode, and battery pack to battery cell balancing mode. Mode selection is achieved by integrating improved Kalman filter dynamic parameter estimation with rolling time-domain optimization predictive switching methods, including terminal voltage detection and calculation, voltage transfer parameter estimation, multi-step prediction based on rolling time-domain optimization, and optimal mode selection. The optimal equilibrium mode is obtained.

[0008] Furthermore, the battery balancing topology constructed based on the acquired parameters includes setting a three-winding transformer on the topology, with the number of switches being 2n+8, where n is the number of battery cells. Using the three-winding transformer as the core and the battery terminal voltage as the mode selection criterion, energy transfer is achieved. The balancing circuit topology includes: a battery pack, a three-winding transformer, a switch array, a voltage acquisition module, and a control module. The battery pack consists of n battery cells connected in series, and the battery cells are divided into odd and even groups, with the odd-numbered battery cells being B1, B3, …B k+1 , …Bn-1 Even-numbered battery cells are B2, B4, …B k , …B n The windings of the three-winding transformer are defined from top to bottom and left to right as the first winding, second winding, and third winding. Each battery cell is connected to the first and second windings of the three-winding transformer via two sets of anti-series switching transistors (MOSFETs). Each adjacent battery cell shares a pair of anti-series switching transistors. The entire battery pack is connected to the third winding of the three-winding transformer at both ends. Anti-series diodes are connected in series in all three windings of the three-winding transformer to control current flow. The voltage of each battery cell is directly used as the criterion for the equalization mode decision. The voltage acquisition module collects the voltage across each battery cell and converts it into a digital signal for the control module. The control module receives the voltage data, selects the appropriate equalization mode, and provides a PWM control signal to the switching array. In the switching array, from Q... 11 Q 12 , …Q (n+1)1 Q (n+1)2 These 2n+2 switching transistors are paired up and controlled by the same PWM control signal. The control signals controlling the 2n+2 switching transistors are denoted as PWM1, PWM2, ... PWM (n+1) The remaining six switching transistors Q (n+2) …Q (n+7) Controlled by a separate PWM control signal, denoted as PWM. (n+2) , ...PWM (n+7) .

[0009] Furthermore, the cell-to-cell balancing mode includes, based on even-numbered cell B... k During the discharge process, a specific combination of PWM signals output by the control module triggers and controls the switching transistor Q. k1 Q k2 Q (k+1)1 Q (k+1)2 Q (n+4) Q (n+5) Output a high-level drive signal, corresponding to the PWM control signal. k PWM (k+1) PWM (n+4) PWM (n+5) This causes the switching transistors to conduct synchronously, at which point current flows from battery cell B. kThe energy flows out from the positive terminal of the battery, sequentially through the conducting switch and the second winding of the transformer, and finally back to the negative terminal of the battery, forming a closed loop. During this stage, the loop current gradually increases over time, and the battery's electrical energy is converted into magnetic field energy and stored in the transformer's core. Correspondingly, the induced voltage across the relevant windings of the transformer exhibits a polarity of positive at the top and negative at the bottom. After this energy storage process lasts for a time interval Δt1, the control module sets all PWM signals to a low level, driving all corresponding switches to turn off. After completing the discharge stage, the system enters the charging stage, directionally transferring the energy stored in the transformer to the battery cells to be charged. This process also follows a specific switch conduction logic based on the odd and even grouping of the battery cells; during the charging process based on even-numbered battery cells, the switch Q... k1 Q k2 Q (k+1)1 Q (k+1)2 Q (n+2) When the circuit is turned on, the voltage across the first winding of the transformer exhibits a polarity of positive at the bottom and negative at the top. Current flows out from the bottom of the first winding of the transformer and flows through the battery cell B. k The current then returns to the upper end of the first winding of the transformer, forming a loop. As the loop current gradually decreases over time, the magnetic field energy stored in the transformer core is converted into electrical energy and transferred to battery B. k During this process, after a duration of Δt2, the loop current drops to zero; due to Q (n+3) Since the diode is not conducting, it is reverse-biased and cut off after the current is zero, so the battery will not discharge again and the current in the circuit remains zero.

[0010] Furthermore, the battery cell to battery pack balancing mode includes transitioning from a battery cell discharge mode to a battery pack charging mode, wherein the switching transistor Q... (n+7) By controlling the PWM signal (n+7) When the circuit is turned on, the voltage across the third winding of the transformer exhibits a polarity of positive at the top and negative at the bottom. Driven by this polarity, current flows from the upper end of the third winding, through the entire battery pack, and back to the lower end of the third winding, forming a loop. The current in this loop gradually decreases over time, and the magnetic field energy stored in the transformer core is converted into electrical energy and transferred to the entire battery pack until the loop current drops to zero. Because Q... (n+6) Since the diode is not conducting, it is reverse-biased after the current becomes zero, so the battery pack will not discharge again. The current in the circuit remains zero, keeping all the switching transistors in the same state until the current equalization cycle ends.

[0011] Furthermore, the battery pack-to-cell balancing mode includes transitioning from a battery pack discharge mode to a cell charging mode, and the switching transistor Q... (n+6) Q (n+7)After conduction, current flows out from the positive terminal of the battery, passes through the third winding of the transformer, and flows back, forming a closed loop. The loop current gradually increases over time, and energy is transferred from the battery pack to the transformer core. After a set time Δt1 is reached, the control module sets all PWM signals to low level, driving all corresponding switching transistors to turn off. During the charging process based on even-numbered battery cells, the switching transistor Q... k1 Q k2 Q (k+1)1 Q (k+1)2 Q (n+4) When the circuit is turned on, under polarity drive, current flows out from the upper end of the second winding of the transformer and flows through the battery cell B. k The current then returns to the lower end of the second winding of the transformer, forming a loop. The current in this loop gradually decreases over time, and the magnetic field energy stored in the transformer core is converted into electrical energy, which is then transferred to battery B. k In the middle, until the loop current drops to zero.

[0012] Furthermore, the terminal voltage detection and calculation includes first performing terminal voltage detection and calculation, wherein the control module calculates the average voltage V based on the input terminal voltage data of n individual battery cells. avg : Then, the deviation of each battery cell from the average voltage was calculated: To quantify the overall degree of imbalance, an imbalance index S is defined: When all individual cell voltages are exactly equal, S=0; simultaneously, the control module identifies the cell with the highest and lowest terminal voltage, corresponding to cell B respectively. max and B min : .

[0013] Furthermore, the voltage transfer parameter estimation includes defining the single voltage transfer amounts ΔV1 and ΔV2 used in the battery cell-to-pack balancing mode and the battery pack-to-battery cell balancing mode based on the energy transfer speed characteristics of the three-winding transformer balanced topology. The single voltage transfer amount ΔV3 is used in the battery cell-to-battery cell balancing mode. In the battery cell-to-pack balancing mode, energy is transferred from the individual battery cell to the entire battery pack, targeting the battery pack terminal voltage V. pack High, when the third winding of the transformer charges the battery pack, according to the volt-second balance principle: The inductor current drops to zero within time Δt2, thus allowing time for the discharge phase Δt1. In the cell-to-cell equalization mode, because the voltage of a single cell is much lower than the voltage of the battery pack, the inductor current rises and falls slowly, resulting in a smaller voltage transfer amount ΔV3. However, the actual voltage transfer amount ΔV changes with each equalization operation, and the accuracy of ΔV directly affects the mode switching algorithm. Therefore, a Kalman filter is introduced for online estimation. Each parameter is estimated independently using a Kalman filter. ΔV1 is considered a slowly changing quantity affected by various factors and modeled as a random walk process. ,in This represents the true value of ΔV1 during the k-th balancing operation. For process noise, variance Q reflects the severity of parameter fluctuations over time; the actual voltage transfer is measured after each cell-to-pack equalization operation. Let the monomer being balanced be B. max The voltage before equalization is V. max The voltage after equalization is V max ', then the observed value is: The observation equation is expressed as: ,in To observe the noise, the variance R depends on the voltage acquisition accuracy and short-term fluctuations; in the Kalman filter prediction stage, ΔV1 is assumed to be constant, and the covariance of the estimation error is calculated: During the update phase, ΔV1 and the covariance of the estimation error are updated using the measured voltage transfer. , , ,in Here, P is the state estimate, P is the estimation error covariance, K is the Kalman gain, and the initial value is... It can be set to a smaller value. Q and R are determined based on the actual system debugging.

[0014] Furthermore, the multi-step prediction based on rolling time-domain optimization includes real-time estimates of ΔV based on the Kalman filter output, which form the basis of the rolling time-domain optimization prediction. The rolling time-domain optimization method is used to select the optimal mode sequence over the next N equilibrium periods, where N is the prediction time domain, and there are a total of 3 steps in the next N steps. N Possible action sequences Each sequence This represents the equilibrium mode to be executed sequentially over the next N cycles; for each candidate sequence, it starts from the current measured voltage vector. Initially, using the latest ΔV estimate from the Kalman filter output, the voltage evolution after executing the sequence is predicted step by step according to a preset equalization rule; let the predicted voltage vector at step t be... According to the mode selected in step t Update voltage: If In the battery cell to battery pack mode, after one equalization cycle, the voltage change of each cell is as follows: , like In the battery pack-to-cell mode, after one equalization cycle, the voltage change of each cell is as follows: , like In cell-to-cell mode, after one equalization cycle, the voltage change of each cell is as follows: , The voltages of other uninvolved cells remain unchanged. By recursively applying this formula, the predicted voltage sequence from step 1 to step N is obtained. .

[0015] Furthermore, the optimal mode selection includes defining the cost of each prediction step as the imbalance at that moment. The cumulative cost is the weighted sum of the imbalances at each step. ,Pick As a discount factor, it assigns a higher weight to recent imbalance, and 0 < <1: , Compare the cumulative costs of all candidate sequences and select the one with the highest cumulative cost. minimal sequence : Output the first action of the sequence. The corresponding PWM control signal drives the equalization circuit to perform the equalization operation of the selected mode, complete one equalization cycle, and then return to step one to re-acquire the latest terminal voltage data. The actual voltage change of this equalization will be fed back to the Kalman filter to update the ΔV estimate of the corresponding mode. The updated parameters are then used for the next round of rolling time-domain prediction and decision-making.

[0016] Secondly, a three-mode predictive switching system for an equalization topology based on battery terminal voltage includes: The data acquisition module is configured to acquire the basic parameters and operating status parameters of the battery balancing topology. The topology building module is configured to build a battery balancing topology based on the acquired parameters; The mode construction module is configured to build three equalization modes based on the constructed battery equalization topology, including battery cell to battery cell equalization mode, battery cell to battery pack equalization mode and battery pack to battery cell equalization mode. The mode selection module is configured to perform mode selection by fusing improved Kalman filter dynamic parameter estimation with rolling time domain optimization prediction switching method, including terminal voltage detection and calculation, voltage transfer parameter estimation, multi-step prediction based on rolling time domain optimization and optimal mode selection; The output module is configured to obtain the optimal equilibrium mode.

[0017] Thirdly, the present invention provides a computer-readable storage medium storing a plurality of instructions adapted for loading and execution by a processor of a terminal device of the aforementioned method for predictive switching of a balanced topology based on battery terminal voltage.

[0018] Fourthly, the present invention provides a terminal device, including a processor and a computer-readable storage medium, wherein the processor is used to implement various instructions; the computer-readable storage medium is used to store multiple instructions, the instructions being adapted to be loaded and executed by the processor to provide the aforementioned equalization topology three-mode prediction switching method based on battery terminal voltage.

[0019] In summary, the present invention has the following beneficial technical effects: Compared with existing technologies, the "equalization topology design based on battery terminal voltage and its three-mode predictive switching method" provided by this invention has the following significant advantages: (1) The topology is greatly simplified, and the cost and volume are significantly reduced. This invention achieves active balancing of n series-connected batteries using only a single three-winding transformer and a limited number of switching transistors (2n+8 in total). Compared to traditional multi-winding schemes that require windings similar to the number of batteries and a large number of switching transistors, this topology significantly reduces the number of components, circuit complexity, and overall size, lowering manufacturing costs and system weight. It also reduces potential failure points and improves system reliability and maintainability.

[0020] (2) Supports flexible balance of three modes to adapt to complex imbalance scenarios Three balancing modes can be achieved using the same topology: cell-to-cell, cell-to-pack, and pack-to-cell. The system can select the most suitable balancing path based on the actual inconsistent state of the battery pack, overcoming the limitations of traditional solutions with their single mode and fixed path. Especially when extreme deviations occur in battery voltage (overcharging / over-discharging), directional balancing can be quickly initiated, effectively improving system safety.

[0021] (3) High equilibrium efficiency and improved energy utilization. The active energy transfer method avoids the energy dissipation of passive equalization and reduces heat generation. The three-winding transformer, as an energy transfer unit, achieves efficient energy conversion, especially under large voltage differences (such as between battery packs and individual cells). By utilizing the rapid charging and discharging characteristics of the high-voltage side, it can achieve a larger single energy transfer and shorten the equalization time.

[0022] (4) The optimal equilibrium mode improves the equilibrium speed. By collecting the terminal voltage of each battery cell in real time, calculating the imbalance index, and predicting the improvement effect of each mode, the system can dynamically select the optimal balancing mode. This strategy balances safety and overall balancing efficiency, quickly eliminating cells with extreme voltage while achieving overall voltage consistency through point-to-point energy adjustment within a safe range.

[0023] (5) The control logic is clear and easy to implement digital management. All switching transistor drive signals are generated by a unified control module, with clear logical hierarchy and strict sequential switching between modes, facilitating fully digital control via microcontrollers or digital signal processors. The modular design also facilitates integration with battery management systems.

[0024] In summary, this invention has significant improvements in terms of structural simplification and flexible mode, and is suitable for lithium-ion battery pack applications such as electric vehicles and energy storage systems that have high requirements for size, cost, and safety. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the overall structure of the battery balancing topology system according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the even-numbered battery cell discharge mode under the battery cell-to-battery cell equalization mode in an embodiment of the present invention; Figure 3 This is a schematic diagram of the discharge mode of odd-numbered battery cells under the battery cell-to-battery cell equalization mode in an embodiment of the present invention; Figure 4 This is a schematic diagram of the even-numbered battery cell charging mode in the battery cell-to-battery cell equalization mode of an embodiment of the present invention; Figure 5 This is a schematic diagram of the charging mode of odd-numbered battery cells in the battery cell-to-battery cell equalization mode according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the transformer winding current in the battery cell-to-battery cell balancing mode according to an embodiment of the present invention; Figure 7 This is a schematic diagram of the battery pack charging mode according to an embodiment of the present invention; Figure 8 This is a schematic diagram of the battery pack discharge modes according to an embodiment of the present invention; Figure 9This is a schematic diagram of the transformer winding current in the battery cell to battery pack balancing mode according to an embodiment of the present invention; Figure 10 This is a schematic diagram of the even-numbered battery cell charging mode under the battery pack to battery cell equalization mode of this invention embodiment; Figure 11 This is a schematic diagram of the charging mode of odd-numbered battery cells in the battery pack to battery cell equalization mode of an embodiment of the present invention; Figure 12 This is a schematic diagram of the transformer winding current in the battery pack to battery cell balancing mode according to an embodiment of the present invention; Figure 13 This is a flowchart of the equalization mode prediction and switching method according to an embodiment of the present invention; Figure 14 This is a comparison chart of the voltage convergence curves verified in the first group of experiments of this invention. Figure 15 This is a comparison chart of the voltage convergence curves in the second set of experiments of the embodiments of the present invention. Detailed Implementation

[0026] The present invention will be further described in detail below with reference to the accompanying drawings.

[0027] Example 1 Reference Figure 1 This embodiment of a three-mode prediction switching method for equalization topology based on battery terminal voltage includes: Step 1: Obtain the parameters required for the battery balancing topology; 1. Obtain basic battery pack parameters: By inputting data, basic parameters such as the number of battery cells connected in series in the battery pack, the rated capacity C of each battery cell, and the charge / discharge cutoff voltage of each battery cell can be obtained.

[0028] 2. Real-time acquisition of battery operating status parameters: The voltage sampling circuit collects the terminal voltages V1, V2, …V of each individual battery cell in real time. n .

[0029] Step 2: Construct a battery balancing topology based on the acquired parameters In terms of topology, the number of transformers is reduced to one, requiring only three windings, and the number of switching transistors is 2n+8 (n is the number of battery cells), which greatly reduces the complexity of the topology and reduces the size of the equalization circuit. In addition, three equalization modes are provided for the proposed topology, and the optimal equalization mode can be determined by the predictive switching algorithm based on the imbalance in the battery pack.

[0030] The proposed battery balancing topology uses a three-winding transformer as its core and battery terminal voltage as the mode selection criterion to achieve energy transfer. The balancing circuit topology is as follows: Figure 1 As shown, it includes: a battery pack, a three-winding transformer, a switch array, a voltage acquisition module, and a control module.

[0031] The battery pack consists of n battery cells connected in series. The battery cells are divided into two groups: odd and even. The odd-numbered battery cells are B1, B3, ... B1. k+1 , …B n-1 Even-numbered battery cells are B2, B4, …B k , …B n The windings of a three-winding transformer are defined from top to bottom and from left to right as the first winding, the second winding, and the third winding.

[0032] Each battery cell is connected to the first and second windings of a three-winding transformer via two sets of anti-series switching transistors. The switching transistors can be MOSFETs or IGBTs, etc. Each adjacent battery cell shares a pair of anti-series switching transistors. The two ends of the entire battery pack are connected to the third winding of the three-winding transformer. Anti-series diodes are connected in series in all three windings of the three-winding transformer to control the current flow.

[0033] Since the battery terminal voltage signal can be obtained directly and without hysteresis, the terminal voltage of the battery cell is directly used as the criterion for the equalization mode decision. The voltage acquisition module is used to acquire the terminal voltage of each battery cell and convert it into a digital signal for the control module to use. The control module receives the terminal voltage data, selects the appropriate equalization mode, and provides PWM control signals to the switch array.

[0034] In the switch array, from Q 11 Q 12 , …Q (n+1)1 Q (n+1)2 These 2n+2 switching transistors are paired up and controlled by the same PWM control signal. The control signal controlling these 2n+2 switching transistors is denoted as PWM1, PWM2, ... PWM (n+1) The remaining six switching transistors Q (n+2) …Q (n+7) A separate PWM control signal is required for control, denoted as PWM. (n+2) , ...PWM (n+7) .

[0035] Step 3: Three-mode analysis This topology achieves three balancing methods through different switching strategies: balancing from battery cell to battery cell, from battery cell to battery pack, and from battery pack to battery cell. Each balancing method can be further divided into three modes.

[0036] 1) Cell-to-cell balancing mode This equalization mode enables energy transfer from any cell in the battery pack to any cell in the battery pack.

[0037] Mode 1: Single cell discharge The discharge paths differ for battery cells grouped in even numbers and those grouped in odd numbers.

[0038] Based on even-numbered battery cells (using B) k For example, during the discharge process of a device (where k is any even number not greater than n), the current path is as follows: Figure 2 As shown, this process is triggered and controlled by a specific combination of PWM signals output by the control module. The control module sends signals to the switching transistor Q. k1 Q k2 Q (k+1)1 Q (k+1)2 Q (n+4) Q (n+5) Output high-level drive signal (corresponding to the control signal PWM) k PWM (k+1) PWM (n+4) PWM (n+5) This causes the aforementioned switching transistors to conduct synchronously. At this time, since the magnetic field energy in the three-winding transformer has been completely released during the demagnetization phase of the previous equalization cycle, there is no reverse induced electromotive force at the ends of each winding, and the current flows from battery cell B. k The current flows out from the positive terminal of the battery, sequentially through the conducting switching transistor and the second winding of the transformer, and finally returns to the negative terminal of the battery, forming a closed loop. During this stage, the loop current gradually increases over time, and the battery's electrical energy is converted into magnetic field energy and stored in the transformer's core. Correspondingly, the induced voltage across the relevant windings of the transformer exhibits a polarity of "positive at the top and negative at the bottom." After this energy storage process continues for a preset time interval Δt1, the control module sets all the aforementioned PWM signals to a low level, driving all corresponding switching transistors to turn off, thereby ending the discharge operation of this stage and preparing for the subsequent energy transfer stage.

[0039] Based on odd-numbered battery cells (using B) k+1 For example, in the discharge equalization process (where k+1 is not greater than n), the current path is as follows: Figure 3 As shown, the control module sends a signal to the switching transistor Q. (k+1)1 Q (k+1)2 Q (k+2)1 Q (k+2)2 Q (n+2) Q (n+3) Output high-level drive signal (corresponding to the control signal PWM) (k+1) PWM (k+2) PWM (n+2)PWM (n+3) This causes the aforementioned switching transistors to conduct synchronously. During this stage, current flows from battery cell B... k+1 The positive current flows out, passes through the switched transistor and the first winding of the transformer, and returns to the negative current, forming a loop. Similar to the even-numbered cell discharge mode, the loop current continuously rises during the time interval Δt. The battery's electrical energy is converted into magnetic field energy and stored in the transformer core, with the corresponding winding of the transformer exhibiting a "positive at the top and negative at the bottom" polarity. After the same preset time interval Δt1, the control module cuts off all the above PWM signals and turns off the corresponding switched transistor, thereby ending the discharge phase of the odd-numbered battery cell.

[0040] Mode 2: Battery cell charging After the discharge phase is completed, the system enters the charging phase, transferring the energy stored in the transformer to the individual battery cells to be charged. This process also follows a specific switching logic based on the odd / even grouping of the battery cells.

[0041] During the charging process based on even-numbered battery cells, with B... k For example, its current path is as follows Figure 4 As shown, the switching transistor Q k1 Q k2 Q (k+1)1 Q (k+1)2 Q (n+2) On (corresponding to the PWM control signal) k PWM (k+1) PWM (n+2) According to the law of electromagnetic induction, the voltage across the first winding of the transformer exhibits a polarity of "positive at the bottom and negative at the top." Driven by this polarity, current flows out from the bottom of the first winding of the transformer and through the battery cell B. k The current then returns to the upper end of the first winding of the transformer, forming a loop. As the current in this loop gradually decreases over time, the magnetic field energy stored in the transformer core is converted into electrical energy and transferred to battery B. k This process continues until, after time Δt2, the loop current drops to zero, indicating that the energy stored in the transformer for this path has been completely released.

[0042] During the charging process based on odd-numbered battery cells, with B k+1 For example, its current path is as follows Figure 5 As shown, the switching transistor Q (k+1)1 Q (k+1)2 Q (k+2)1 Q (k+2)2 Q (n+5) On (corresponding to the PWM control signal) (k+1) PWM (k+2) PWM (n+5)At this point, the voltage across the second winding of the transformer exhibits a polarity of "positive at the bottom and negative at the top," and the current flows out from the bottom of the second winding of the transformer, passing through the battery cell B. k+1 It then returns to the upper end of the second winding of the transformer, forming a loop. The remaining analysis process is similar to that of the charging mode of an even-numbered battery cell.

[0043] Mode 3: Stationary state After the current in mode two drops to zero, the cell enters this mode. For cells grouped into even numbers, due to Q... (n+3) Since the diode is not conducting, it is reverse-biased and cut off after the current reaches zero. The battery will not discharge again, and the current in the circuit remains zero. For battery cells Q in an odd-numbered group... (n+4) If not conducting, the same applies to the others. Keep all switching transistors in their current states until the end of the current balancing cycle. The transformer winding current in the complete cell-to-cell balancing mode is as follows: Figure 6 As shown.

[0044] 2) Battery cell to battery pack balancing mode This equalization mode enables energy transfer from any individual cell in the battery pack to the entire battery pack.

[0045] Mode 1: Single cell discharge This mode is completely consistent with mode one of battery cell-to-battery cell equilibration.

[0046] Mode 2: Battery pack charging Current path as Figure 7 As shown, the switching transistor Q (n+7) On (corresponding to the PWM control signal) (n+7) According to the law of electromagnetic induction, the voltage across the third winding of the transformer exhibits a polarity of "positive at the top and negative at the bottom." Driven by this polarity, current flows from the upper end of the third winding, through the entire battery pack, and back to the lower end of the third winding, forming a loop. The current in this loop gradually decreases over time, and the magnetic field energy stored in the transformer core is converted into electrical energy and transferred to the entire battery pack. This process continues until the loop current drops to zero, signifying that the energy stored in the transformer has been completely released.

[0047] Mode 3: Stationary state After the current in mode two drops to zero, this mode is entered, due to Q. (n+6) Since the diode is not conducting, it is reverse-biased after the current reaches zero, preventing the battery pack from discharging again. The current in the circuit remains zero, keeping all switching transistors in the same state until the current balancing cycle ends. The complete battery cell to battery pack balancing mode transformer winding current is as follows: Figure 9 As shown.

[0048] 3) Battery pack to individual cell balancing mode This equalization mode enables energy transfer from the entire battery pack to any individual battery cell.

[0049] Mode 1: Battery pack discharge Current path as Figure 8 As shown, the switching transistor Q (n+6) Q (n+7) On (corresponding to the PWM control signal) (n+6) PWM (n+7) The current flows out from the positive terminal of the battery, passes through the third winding of the transformer, and flows back, forming a closed loop. The loop current gradually increases over time, and energy is transferred from the battery pack to the transformer core. At this time, the induced voltage across the third winding of the transformer exhibits a polarity of "positive at the top and negative at the bottom". After reaching the set time Δt1, the control module sets all the above PWM signals to a low level, driving all the corresponding switching transistors to turn off, thereby ending the discharge operation of this stage.

[0050] Mode 2: Battery cell charging The charging paths differ for battery cells grouped in even numbers and in odd numbers, and are different from the battery cell charging modes in the battery cell-to-battery cell balancing mode.

[0051] During the charging process based on even-numbered battery cells, with B... k For example, its current path is as follows Figure 10 As shown, the switching transistor Q k1 Q k2 Q (k+1)1 Q (k+1)2 Q (n+4) On (corresponding to the PWM control signal) k PWM (k+1) PWM (n+4) According to the law of electromagnetic induction, the voltage across the second winding of the transformer exhibits a polarity of "positive at the top and negative at the bottom." Driven by this polarity, current flows out from the upper end of the second winding of the transformer and flows through the battery cell B. k The current then returns to the lower end of the second winding of the transformer, forming a loop. As the current in this loop gradually decreases over time, the magnetic field energy stored in the transformer core is converted into electrical energy and transferred to battery B. k This process continues until the loop current drops to zero, signifying that the energy stored in the transformer for this path has been completely released.

[0052] During the charging process based on odd-numbered battery cells, with B k+1 For example, its current path is as follows Figure 11 As shown, the switching transistor Q (k+1)1 Q (k+1)2 Q (k+2)1 Q (k+2)2 Q(n+3) On (corresponding to the control signal PWM) (k+1) PWM (k+2) PWM (n+3) At this point, the voltage across the first winding of the transformer exhibits a polarity of "positive at the top and negative at the bottom," and the current flows out from the top of the first winding of the transformer, passing through the battery cell B. k+1 The charge then returns to the lower end of the first winding of the transformer, forming a loop. The remaining analysis process is similar to that of the even-numbered battery cell charging mode.

[0053] Mode 3: Stationary state After the current in mode two drops to zero, entering this mode, for battery cells grouped in even numbers, Q (n+5) Not conducting; for battery cells grouped in odd numbers, Q (n+2) Since the diode is not conducting, and the current is zero, it is reverse-biased and the battery will not discharge again. All switching transistors remain in the same state until the current balancing cycle ends. The transformer winding current in the complete battery pack to individual cell balancing mode is as follows: Figure 12 As shown.

[0054] Step 4: Equilibrium Mode Decision and Switching The voltage acquisition module collects data for each battery cell, B1, B2, …B. n The terminal voltages V1, V2, …V n Then, the data is sent to the control module, which determines which equalization mode to use based on the following method, the flowchart of which is shown below. Figure 13 As shown.

[0055] 1) Check if any individual battery cells exceed the set safety threshold. In the control module of this invention, the safe voltage threshold of a single battery cell is pre-stored, including the maximum voltage threshold V. max and minimum voltage threshold V min The control module monitors the voltage of each cell in the battery pack and compares it with the aforementioned safe voltage threshold. If it detects that the voltage of any battery cell is higher than V... max This indicates that the individual cell is overcharged. Continuing to charge may cause the battery to overheat, bulge, or even catch fire and explode. The control module then initiates the battery cell to battery pack equalization mode.

[0056] Conversely, when the voltage of any single battery cell is detected to be lower than V... min The control module determines that the cell is over-discharged, and continued discharge may permanently reduce the battery life or even damage it. The control module then starts the battery pack into cell balancing mode.

[0057] The two balancing modes described above, by transferring energy from cells with excessively high voltage to the entire battery pack, or from the battery pack to cells with excessively low voltage, can disperse extreme risks during the charging and discharging process from individual cells to the entire battery pack, thereby significantly reducing the probability of battery system failure and improving its safety and service life.

[0058] 2) After ensuring all battery cells are within a safe range, the control module enters the optimal balancing mode selection stage. To combine the advantages of large single-transfer energy amounts in the battery cell-to-pack and battery pack-to-cell balancing modes, and the advantage of precise point-to-point energy transfer in the battery cell-to-cell balancing mode, a predictive switching method integrating improved Kalman filter dynamic parameter estimation and rolling time-domain optimization is adopted in this stage. This method can adaptively correct balancing parameters. The rolling optimization module performs multi-step optimization based on these parameters to select the globally optimal balancing action. The actual effect after the action is executed is fed back to the Kalman filter module to correct subsequent predictions, forming a coupled "estimation-prediction-decision" structure, ensuring that the balancing decision always maintains the best match with the current state of the battery. This stage is achieved through the following steps: (1) Terminal voltage detection and calculation The control module calculates the average voltage V based on the input terminal voltage data of n individual battery cells. avg : , Then, the deviation of each individual cell from the average voltage was calculated: , To quantify the overall imbalance of the system, an imbalance index S is defined: , This indicator has the following characteristics: S=0 when all individual cell voltages are completely equal; in addition, it is more sensitive to larger deviations, which meets the requirement of prioritizing the adjustment of the individual cell with the largest deviation in the balancing process.

[0059] Simultaneously, the control module identifies the battery cells with the highest and lowest terminal voltages, corresponding to battery cell B respectively. max and B min : , (2) Online estimation of ΔV parameters The battery cell to battery pack balancing mode and the battery pack to battery cell balancing mode are defined to use larger single voltage transfer amounts ΔV1 and ΔV2, while the battery cell to battery cell balancing mode uses a smaller single voltage transfer amount ΔV3. This design is based on the energy transfer speed characteristics of the three-winding transformer balancing topology.

[0060] In the battery cell-to-pack balancing mode, energy is transferred from the individual battery cells to the entire battery pack, due to the battery pack terminal voltage V. pack The voltage is relatively high. When the third winding of the transformer charges the battery pack, according to the volt-second balance principle: , The inductor current can quickly drop to zero within Δt2 time, thus allowing more time for the discharge stage Δt1, enabling individual cells to discharge for a longer period of time and resulting in a larger single voltage transfer. The same principle applies to the battery pack to individual cell balancing mode.

[0061] In the cell-to-cell balancing mode, since the voltage of a single cell is much lower than that of the battery pack, the inductor current rises and falls more slowly. However, the time of each balancing cycle is fixed, so the corresponding voltage transfer amount ΔV3 is also small.

[0062] In reality, due to factors such as battery state of charge, temperature, and aging, the actual voltage transfer amount ΔV in each equalization operation will change. The accuracy of ΔV will directly affect the mode switching algorithm. Therefore, Kalman filtering is introduced to estimate it online. Each parameter is estimated independently using a Kalman filter. Taking ΔV1 as an example, its implementation process is explained.

[0063] Treating ΔV1 as a slowly changing quantity influenced by various factors, we model it as a random walk process: , in This represents the true value of ΔV1 during the k-th balancing operation. The variance Q represents the process noise and reflects the degree of fluctuation of the parameter over time.

[0064] After each battery cell to battery pack equalization operation, the actual voltage transfer amount can be measured. Let the monomer being balanced be B. max The voltage before equalization is V. max The voltage after equalization is V max ', then the observed value is: , The observation equation is expressed as: , in To observe the noise, the variance R depends on the voltage acquisition accuracy and short-term fluctuations.

[0065] In the Kalman filter prediction stage, ΔV1 is assumed to remain constant, and the covariance of the estimation error is calculated: , During the update phase, ΔV1 and the covariance of the estimation error are updated using the measured voltage transfer: , , , in Here, P is the state estimate, P is the estimation error covariance, K is the Kalman gain, and the initial value is... It can be set to a smaller value. Q and R are determined based on the actual system debugging.

[0066] Similarly, independent Kalman filters are established for ΔV2 and ΔV3 respectively. When the corresponding mode is executed, the corresponding parameter estimates are updated using the measured voltage changes. For modes that are not executed, the estimates remain unchanged.

[0067] (3) Multi-step prediction based on rolling time domain optimization The Kalman filter continuously outputs real-time estimates of ΔV for each mode. These parameters form the basis for the rolling time-domain optimization prediction. Based on this, the control module uses the rolling time-domain optimization method to select the optimal mode sequence in the next N equilibrium periods. N is the prediction time domain, which can be 3 to 5 and can be dynamically adjusted according to the system's real-time requirements.

[0068] Since there are only three modes, there are a total of 3 in the next N steps. N Possible action sequences Each sequence This indicates the equilibrium mode that will be executed sequentially over the next N cycles.

[0069] For each candidate sequence, starting from the current measured voltage vector Initially, using the latest ΔV estimate from the Kalman filter output, the voltage evolution after executing the sequence is predicted step by step according to the preset equalization rule.

[0070] Let the predicted voltage vector at step t be... According to the mode selected in step t Update voltage: like In the battery cell to battery pack mode, after one equalization cycle, the voltage change of each cell is as follows: , like In the battery pack-to-cell mode, after one equalization cycle, the voltage change of each cell is as follows: , like In cell-to-cell mode, after one equalization cycle, the voltage change of each cell is as follows: , The voltages of other uninvolved cells remain unchanged. By recursively applying this formula, the predicted voltage sequence from step 1 to step N is obtained. .

[0071] (4) Optimal mode selection The cost of each prediction step is defined as the imbalance at that moment. The cumulative cost is the weighted sum of the imbalances at each step. ,Pick As a discount factor, it assigns a higher weight to recent imbalance, and 0 < <1: , , Compare the cumulative costs of all candidate sequences and select the cumulative cost. minimal sequence : , (5) Instruction execution The control module outputs the first action of the sequence. The corresponding PWM control signal drives the equalization circuit to perform the equalization operation of the selected mode, complete one equalization cycle, and then return to step one to re-acquire the latest terminal voltage data. The actual voltage change of this equalization will be fed back to the Kalman filter to update the ΔV estimate of the corresponding mode. The updated parameters are then used for the next round of rolling time-domain prediction and decision-making.

[0072] When the maximum voltage difference between individual cells is detected to be less than the set value V th When the equilibrium target is considered to have been achieved, the control module stops the equilibrium process to avoid unnecessary energy loss.

[0073] Experimental verification To verify the feasibility of the proposed three-mode battery terminal voltage balancing topology and the effectiveness of its mode prediction switching method, a simulation and experimental verification platform consisting of 12 battery cells (nominal voltage 3.7V, capacity 5.4Ah) connected in series was built. This platform completely reproduces... Figure 1 The circuit topology shown is illustrated, and the corresponding mode prediction switching algorithm is written.

[0074] To demonstrate the advantages of mode selection based on imbalance, two sets of comparative experiments were conducted to compare the balancing effect and efficiency differences between using a single fixed balancing mode and using a mode prediction switching method under the same initial terminal voltage.

[0075] Group 1: The initial voltage distribution of the 12 batteries is as follows: The entire process adopts a cell-to-cell balancing mode, and the voltage convergence curve of the balancing process is as follows: Figure 14 As shown in (a), the entire process adopts a battery pack-to-cell balancing mode, and the voltage convergence curve of the balancing process is as follows. Figure 14 As shown in (b), the voltage convergence curve of the equalization process is as follows, using the mode prediction switching method. Figure 14 As shown in (c), the comparison reveals that, under the mode prediction switching method, in the initial stage of equalization, the system detects cells with significantly low voltage and quickly activates the battery pack-to-cell equalization mode for rapid energy transfer. Once the voltage of all cells deviates little from the average value, the system flexibly switches to the cell-to-cell mode to finely adjust the remaining deviations. It can be seen that the equalization time is reduced by approximately 8% compared to the single equalization mode.

[0076] Group 2: The initial voltage distribution of the 12 batteries is as follows: The entire process adopts a cell-to-cell balancing mode, and the voltage convergence curve of the balancing process is as follows: Figure 15 As shown in (a), the entire process adopts a battery cell-to-battery pack balancing mode, and the voltage convergence curve of the balancing process is as follows. Figure 15 As shown in (b), the voltage convergence curve of the equalization process is as follows, using the mode prediction switching method. Figure 15 As shown in (c), the comparison reveals that, under the mode prediction switching method, in the initial stage of equalization, the system detects cells with significantly higher voltages and quickly activates the cell-to-pack equalization mode for rapid energy transfer. Once the voltages of all cells are not significantly deviated from the average value, the system flexibly switches to the cell-to-cell mode to finely adjust the remaining deviations. It can be seen that the equalization time is reduced by approximately 11% compared to the single equalization mode.

[0077] Example 2 This embodiment provides a three-mode predictive switching system for an equalization topology based on battery terminal voltage, including: The data acquisition module is configured to acquire the basic parameters and operating status parameters of the battery balancing topology. The topology building module is configured to build a battery balancing topology based on the acquired parameters; The mode construction module is configured to build three equalization modes based on the constructed battery equalization topology, including battery cell to battery cell equalization mode, battery cell to battery pack equalization mode and battery pack to battery cell equalization mode. The mode selection module is configured to perform mode selection by fusing improved Kalman filter dynamic parameter estimation with rolling time domain optimization prediction switching method, including terminal voltage detection and calculation, voltage transfer parameter estimation, multi-step prediction based on rolling time domain optimization and optimal mode selection; The output module is configured to obtain the optimal equilibrium mode.

[0078] A computer-readable storage medium storing a plurality of instructions adapted for loading and execution by a processor of a terminal device of the aforementioned method for predictive switching of a balanced topology based on battery terminal voltage.

[0079] A terminal device includes a processor and a computer-readable storage medium, the processor being configured to implement various instructions; the computer-readable storage medium being configured to store multiple instructions adapted for loading and execution by the processor of the aforementioned equalization topology three-mode predictive switching method based on battery terminal voltage.

[0080] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A three-mode predictive switching method for an equalization topology based on battery terminal voltage, characterized in that, include: Obtain the basic parameters and operating status parameters of the battery balancing topology; A battery balancing topology is constructed based on the acquired parameters; Based on the constructed battery balancing topology, three balancing modes are built, including battery cell to battery cell balancing mode, battery cell to battery pack balancing mode, and battery pack to battery cell balancing mode. Mode selection is achieved by integrating improved Kalman filter dynamic parameter estimation with rolling time-domain optimization predictive switching methods, including terminal voltage detection and calculation, voltage transfer parameter estimation, multi-step prediction based on rolling time-domain optimization, and optimal mode selection. To obtain the optimal equilibrium mode; The voltage transfer parameter estimation includes defining the single voltage transfer amounts ΔV1 and ΔV2 used in the battery cell-to-pack balancing mode and the battery pack-to-battery cell balancing mode based on the energy transfer speed characteristics of the three-winding transformer balanced topology. The single voltage transfer amount ΔV3 is used in the battery cell-to-battery cell balancing mode. In the battery cell-to-pack balancing mode, energy is transferred from the individual battery cell to the entire battery pack, with the voltage V at the battery pack terminal being... pack High, when the third winding of the transformer charges the battery pack, according to the volt-second balance principle: The inductor current drops to zero within the time Δt2, thus allowing time for the discharge phase Δt1. In the cell-to-cell equalization mode, because the voltage of a single cell is much smaller than the voltage of the battery pack, the inductor current rises and falls slowly, resulting in a small voltage transfer amount ΔV3. However, the actual voltage transfer amount ΔV changes with each equalization operation, and the accuracy of ΔV directly affects the mode switching algorithm. Therefore, a Kalman filter is introduced for online estimation. Each parameter is estimated independently using a Kalman filter, and ΔV1 is treated as a slowly changing quantity influenced by various factors, modeled as a random walk process. ,in This represents the true value of ΔV1 during the k-th balancing operation. For process noise, variance Q reflects the degree of fluctuation of the parameter over time; The actual voltage transfer amount is measured after each battery cell to battery pack equalization operation. Let the equilibriumed monomer be The voltage before equalization is The voltage after equalization is The observed value is: The observation equation is expressed as: ,in To observe the noise, the variance R depends on the voltage acquisition accuracy and short-term fluctuations; in the Kalman filter prediction stage, ΔV1 is assumed to be constant, and the covariance of the estimation error is calculated: During the update phase, ΔV1 and the covariance of the estimation error are updated using the measured voltage transfer. , , ,in Here, P is the state estimate, P is the estimation error covariance, K is the Kalman gain, and the initial value is... It can be set to a smaller value. Q and R are determined based on the actual system debugging.

2. The method for predictive switching of a balanced topology based on battery terminal voltage according to claim 1, characterized in that, The battery balancing topology constructed based on the acquired parameters includes setting up a three-winding transformer on the topology, with the number of switches being 2n+8, where n is the number of battery cells. Using the three-winding transformer as the core and the battery terminal voltage as the mode selection criterion, energy transfer is achieved. The balancing circuit topology includes: a battery pack, a three-winding transformer, a switch array, a voltage acquisition module, and a control module. The battery pack consists of n battery cells connected in series, with the battery cells divided into odd and even groups. The odd-numbered battery cells are B1, B3, …B k+1 , …B n-1 Even-numbered battery cells are B2, B4, …B k , …B n The windings of the three-winding transformer are defined from top to bottom and left to right as the first winding, second winding, and third winding. Each battery cell is connected to the first and second windings of the three-winding transformer via two sets of anti-series switching transistors (MOSFETs). Each adjacent battery cell shares a pair of anti-series switching transistors. The entire battery pack is connected to the third winding of the three-winding transformer at both ends. Anti-series diodes are connected in series in all three windings of the three-winding transformer to control current flow. The voltage of each battery cell is directly used as the criterion for the equalization mode decision. The voltage acquisition module collects the voltage across each battery cell and converts it into a digital signal for the control module. The control module receives the voltage data, selects the appropriate equalization mode, and provides a PWM control signal to the switching array. In the switching array, from Q... 11 Q 12 , …Q (n+1)1 Q (n+1)2 These 2n+2 switching transistors are paired up and controlled by the same PWM control signal. The control signals controlling the 2n+2 switching transistors are denoted as PWM1, PWM2, ... PWM (n+1) The remaining six switching transistors Q (n+2) …Q (n+7) Controlled by a separate PWM control signal, denoted as PWM. (n+2) , ...PWM (n+7) .

3. The method for predictive switching of a balanced topology based on battery terminal voltage according to claim 2, characterized in that, The cell-to-cell balancing mode includes even-numbered cell B k During the discharge process, a specific combination of PWM signals output by the control module triggers and controls the switching transistor Q. k1 Q k2 Q (k+1)1 Q (k+1)2 Q (n+4) Q (n+5) Output a high-level drive signal, corresponding to the PWM control signal. k PWM (k+1) PWM (n+4) PWM (n+5) This causes the switching transistors to conduct synchronously, at which point current flows from battery cell B. k The energy flows out from the positive terminal of the battery, sequentially through the conducting switch and the second winding of the transformer, and finally back to the negative terminal of the battery, forming a closed loop. During this stage, the loop current gradually increases over time, and the battery's electrical energy is converted into magnetic field energy and stored in the transformer's core. Correspondingly, the induced voltage across the relevant windings of the transformer exhibits a polarity of positive at the top and negative at the bottom. After this energy storage process lasts for a time interval Δt1, the control module sets all PWM signals to a low level, driving all corresponding switches to turn off. After completing the discharge stage, the system enters the charging stage, directionally transferring the energy stored in the transformer to the battery cells to be charged. This process also follows a specific switch conduction logic based on the odd and even grouping of the battery cells. During the charging process based on even-numbered battery cells, the switch Q... k1 Q k2 Q (k+1)1 Q (k+1)2 Q (n+2) When the circuit is turned on, the voltage across the first winding of the transformer exhibits a polarity of positive at the bottom and negative at the top. Current flows out from the bottom of the first winding of the transformer and flows through the battery cell B. k The current then returns to the upper end of the first winding of the transformer, forming a loop. As the loop current gradually decreases over time, the magnetic field energy stored in the transformer core is converted into electrical energy and transferred to battery B. k During this process, after a duration of Δt2, the loop current drops to zero; due to Q (n+3) Since the diode is not conducting, it is reverse-biased and cut off after the current reaches zero, so the battery will not discharge again and the current in the circuit remains zero.

4. The method for predictive switching of a balanced topology based on battery terminal voltage according to claim 3, characterized in that, The battery cell to battery pack balancing mode includes transitioning from a battery cell discharge mode to a battery pack charging mode, wherein the switching transistor Q... (n+7) By controlling the PWM signal (n+7) When the circuit is turned on, the voltage across the third winding of the transformer exhibits a polarity of positive at the top and negative at the bottom. Driven by this polarity, current flows from the upper end of the third winding, through the entire battery pack, and back to the lower end of the third winding, forming a loop. The current in this loop gradually decreases over time, and the magnetic field energy stored in the transformer core is converted into electrical energy and transferred to the entire battery pack until the loop current drops to zero. Because Q... (n+6) Since the diode is not conducting, it is reverse-biased after the current becomes zero, so the battery pack will not discharge again. The current in the circuit remains zero, keeping all the switching transistors in the same state until the current equalization cycle ends.

5. The method for predictive switching of a balanced topology based on battery terminal voltage according to claim 4, characterized in that, The battery pack-to-cell balancing mode includes transitioning from a battery pack discharge mode to a cell charging mode, and the switching transistor Q... (n+6) Q (n+7) After conduction, current flows out from the positive terminal of the battery, passes through the third winding of the transformer, and flows back, forming a closed loop. The loop current gradually increases over time, and energy is transferred from the battery pack to the transformer core. After a set time Δt1 is reached, the control module sets all PWM signals to low level, driving all corresponding switching transistors to turn off. During the charging process based on even-numbered battery cells, the switching transistor Q... k1 Q k2 Q (k+1)1 Q (k+1)2 Q (n+4) When the circuit is turned on, under polarity drive, current flows out from the upper end of the second winding of the transformer and flows through the battery cell B. k The current then returns to the lower end of the second winding of the transformer, forming a loop. The current in this loop gradually decreases over time, and the magnetic field energy stored in the transformer core is converted into electrical energy, which is then transferred to battery B. k In the middle, until the loop current drops to zero.

6. The method for predictive switching of three modes in an equalization topology based on battery terminal voltage according to claim 5, characterized in that, The terminal voltage detection and calculation includes first performing terminal voltage detection and calculation, wherein the control module calculates the average voltage V based on the input terminal voltage data of n individual battery cells. avg : Then, the deviation of each battery cell from the average voltage was calculated: To quantify the overall degree of imbalance, an imbalance index S is defined: When all individual cell voltages are exactly equal, S=0; simultaneously, the control module identifies the cell with the highest and lowest terminal voltage, corresponding to cell B respectively. max and B min : .

7. The method for predictive switching of a balanced topology based on battery terminal voltage according to claim 6, characterized in that, The multi-step prediction based on rolling time-domain optimization includes real-time estimates of ΔV based on the Kalman filter output, which form the basis of the rolling time-domain optimization prediction. The rolling time-domain optimization method selects the optimal mode sequence for the next N equilibrium periods, where N is the prediction time domain, and there are 3 steps in the next N steps. N Possible action sequences Each sequence This represents the equilibrium mode to be executed sequentially over the next N cycles; for each candidate sequence, it starts from the current measured voltage vector. Initially, using the latest ΔV estimate from the Kalman filter output, the voltage evolution after executing the sequence is predicted step by step according to a preset equalization rule; let the predicted voltage vector at step t be... According to the mode selected in step t Update voltage: If In the battery cell to battery pack mode, after one equalization cycle, the voltage change of each cell is as follows: , like In the battery pack-to-cell mode, after one equalization cycle, the voltage change of each cell is as follows: , like In cell-to-cell mode, after one equalization cycle, the voltage change of each cell is as follows: , The voltages of other uninvolved cells remain unchanged. By recursively applying this formula, the predicted voltage sequence from step 1 to step N is obtained. .

8. The method for predictive switching of a balanced topology based on battery terminal voltage according to claim 7, characterized in that, The optimal mode selection includes defining the cost of each prediction step as the imbalance at that moment. The cumulative cost is the weighted sum of the imbalances at each step. ,Pick As a discount factor, it assigns a higher weight to recent imbalance, and 0 < <1: , Compare the cumulative costs of all candidate sequences and select the one with the highest cumulative cost. minimal sequence : Output the first action of the sequence. The corresponding PWM control signal drives the equalization circuit to perform the equalization operation of the selected mode, complete one equalization cycle, and then return to step one to re-acquire the latest terminal voltage data. The actual voltage change of this equalization will be fed back to the Kalman filter to update the ΔV estimate of the corresponding mode. The updated parameters are then used for the next round of rolling time-domain prediction and decision-making.

9. A three-mode predictive switching system for equalized topology based on battery terminal voltage, executing the three-mode predictive switching method for equalized topology based on battery terminal voltage as described in claim 1, characterized in that, include: The data acquisition module is configured to acquire the basic parameters and operating status parameters of the battery balancing topology. The topology building module is configured to build a battery balancing topology based on the acquired parameters; The mode construction module is configured to build three equalization modes based on the constructed battery equalization topology, including battery cell to battery cell equalization mode, battery cell to battery pack equalization mode and battery pack to battery cell equalization mode. The mode selection module is configured to perform mode selection by fusing improved Kalman filter dynamic parameter estimation with rolling time domain optimization prediction switching method, including terminal voltage detection and calculation, voltage transfer parameter estimation, multi-step prediction based on rolling time domain optimization and optimal mode selection; The output module is configured to obtain the optimal equilibrium mode.

Images