Solid-state electrolyte-based high-energy-density battery equalization control method

Abstract

The application discloses a high-energy-density battery equalization control method based on a solid-state electrolyte and relates to the technical field of battery management. The high-energy-density battery equalization control method based on the solid-state electrolyte comprises the following steps: S1, preprocessing solid-state battery equalization correlation data; S2, dividing an operation window, generating a grouping orthogonal coding duty cycle sequence, and performing multi-layer frequency band analysis and tensor splicing; S3, constructing a solid-state interface composite state space model, generating an interface accommodation characteristic quantity set, and forming a single-body interface accommodation domain; S4, constructing a high-energy-density solid-state electrolyte battery pack equalization energy topology table; generating an equalization branch control instruction; and S5, performing file writing and correcting model correction parameters. The high-energy-density battery equalization control method solves the problem that the existing high-energy-density battery equalization control method does not take interface contact degradation into equalization constraints, is easy to continuously equalize on a high-impedance single body, and causes local heating and performance difference expansion.

Description

Technical Field

[0001] This invention relates to the field of battery management technology, specifically to a method for equalizing and controlling high-energy-density batteries based on solid-state electrolytes. Background Technology

[0002] With the development of new energy technologies, intelligent equipment, and high-power electronic devices, the application scope of high-energy-density batteries in power output, energy storage, and portable energy systems is constantly expanding. Solid-state electrolyte batteries, as an important direction of next-generation battery technology, use solid electrolytes to replace traditional liquid electrolytes. They have advantages such as high potential for energy density improvement, good thermal stability, and high structural safety, making them suitable for applications with high requirements for range, size and weight, safety performance, and cycle life. In battery pack applications, multiple individual cells are typically connected in series or parallel to form a battery pack. Due to differences in manufacturing, operating conditions, and cycle aging, inconsistencies gradually develop between individual cells in terms of voltage, capacity, internal resistance, and state of charge. To improve the overall usable capacity, operational stability, and cycle life of the battery pack, the battery management system usually needs to perform equalization control, adjusting the energy distribution between cells through active or passive equalization methods to maintain good overall consistency of the battery pack during charging, discharging, and long-term cycling.

[0003] For example, invention patent CN106655408B discloses a battery pack balancing control method and control device. During balancing control, the method first acquires the battery module that completes charging first and the battery module that completes discharging first during the discharge process. Then, it determines whether these two battery modules are the same. If they are not the same, the energy from the battery module that completed charging first is transferred to the battery module that completed discharging first. This method balances the energy of the two battery modules, increases the usable capacity of the battery pack, and achieves battery pack balancing. Furthermore, this balancing control method targets key factors affecting the usable capacity of the battery pack, using these key factors as balancing conditions to implement the balancing control strategy. It is highly reliable and effectively ensures the balancing of the battery pack.

[0004] For example, invention patent CN110015188B discloses a battery balancing method, system, vehicle, storage medium, and electronic device. The method includes: obtaining the internal resistance value of each individual cell in a battery pack; determining a reference internal resistance value required for balancing based on the internal resistance values ​​of each individual cell in the battery pack; and determining the individual cell that needs to be balanced based on the internal resistance value of at least one individual cell and the reference internal resistance value. This method determines the individual cell that needs to be balanced based on the internal resistance value of at least one individual cell in the battery pack, rather than fixedly determining it based on the SOC value or load voltage value. Instead, it uses the internal resistance value of the individual cell as the battery information used to determine the individual cell that needs to be balanced, enriching the battery information used to determine the individual cell that needs to be balanced and providing a more diverse method for determining the individual cell that needs to be balanced.

[0005] Current equalization control methods for high-energy-density batteries and solid-state electrolyte battery packs mainly rely on indicators such as cell voltage, state of charge, capacity estimation, or internal resistance for equalization judgment and energy regulation. They typically use the cells that have completed charging or discharging first as the basis for equalization, or determine the cells requiring equalization based on differences in internal resistance. However, during the operation of solid-state electrolyte batteries, cyclic aging, volume changes, and stacking pressure variations can cause interfacial contact degradation in some cells, manifesting as increased interfacial impedance, delayed polarization recovery, and enhanced sensitivity to localized temperature rise. Existing equalization control methods do not use the degree of interfacial contact degradation as a constraint for equalization entry, equalization current limiting, or equalization delay. This can easily lead to the continuous application of equalization current to high-impedance cells, resulting in localized heating, accumulated interfacial stress, and exacerbated voltage recovery lag. Consequently, the performance differences between cells widen in subsequent cycles, affecting the battery pack's lifespan and operational stability.

[0006] Therefore, in order to address the above problems, there is an urgent need for a high-energy-density battery equalization control method based on solid-state electrolytes. Summary of the Invention

[0007] Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a high-energy-density battery equalization control method based on solid-state electrolytes. This method solves the problem that existing high-energy-density battery equalization control methods do not incorporate interface contact degradation into the equalization constraints, and tend to continuously equalize on high-impedance cells, leading to localized heating and increased performance differences.

[0008] Technical solution To achieve the above objectives, the present invention provides the following technical solution: a high-energy-density battery equalization control method based on solid-state electrolyte, comprising: S1, collecting solid-state battery equalization correlation data and preprocessing the solid-state battery equalization correlation data; S2, dividing the operating window based on the solid-state battery equalization correlation data, generating a grouped orthogonal coded duty cycle sequence based on the operating window, and performing multi-layer frequency band analysis and tensor splicing to obtain a single-unit interface response tensor; S3, constructing a solid-state interface composite state space model based on the single-unit interface response tensor, generating a set of interface bearing characteristics; forming a single-unit interface bearing domain based on the set of interface bearing characteristics; S4, constructing a high-energy-density solid-state electrolyte battery pack equalization energy topology table based on the solid-state battery equalization correlation data and the single-unit interface bearing domain; generating equalization branch control commands based on the control barrier function and performing equalization branch switch driving and equalization current control; S5, writing the battery pack historical operating file, and correcting the model correction parameters based on the battery pack historical operating file.

[0009] Furthermore, the specific process for collecting solid-state battery equalization correlation data is as follows: solid-state battery equalization correlation data is simultaneously collected from the battery management system, equalization branch sampling circuit, temperature sampling circuit, and stacking pressure sampling circuit of the high-energy-density solid-state electrolyte battery pack. The solid-state battery equalization correlation data includes the cell location number, equalization branch number, equalization branch connection relationship, sampling timestamp, cell terminal voltage sequence, total battery pack current sequence, cell temperature sequence, equalization branch current sequence, equalization branch voltage sequence, equalization switch drive pulse sequence, stacking pressure sequence, and cycle count.

[0010] Furthermore, the specific preprocessing process for the solid-state battery equalization correlation data is as follows: Time alignment is performed on the individual cell terminal voltage sequence, battery pack total current sequence, individual cell temperature sequence, equalization branch current sequence, equalization branch voltage sequence, equalization switch drive pulse sequence, and stacking pressure sequence according to the sampling timestamp; isolated outliers in the individual cell terminal voltage sequence, individual cell temperature sequence, and stacking pressure sequence are removed using the Hampel filtering algorithm; the solid-state battery equalization correlation data corresponding to the sampling timestamp is completed using the cubic spline interpolation algorithm; the conduction duration and drive cycle of the equalization switch drive pulse sequence are statistically analyzed to obtain the equalization switch duty cycle sequence; the individual cell terminal voltage sequence, battery pack total current sequence, individual cell temperature sequence, equalization branch current sequence, equalization branch voltage sequence, and stacking pressure sequence are standardized using the Z-score normalization algorithm; and the equalization switch duty cycle sequence and cycle count are normalized using the max-min normalization algorithm.

[0011] Furthermore, the specific process of dividing the operating window based on the solid-state battery equalization correlation data and generating a grouped orthogonal coded duty cycle sequence based on the operating window is as follows: The total current sequence of the battery pack and the temperature sequence of each cell are divided into operating windows using a Bayesian online change point detection algorithm based on joint observation of current change rate and temperature rise slope, resulting in a load smoothing window, a load transition window, a resting recovery window, and a heat release window. Within the load smoothing window, resting recovery window, and heat release window, the cells to be excited and their corresponding equalization branches are determined according to the cell location number and equalization branch number. Orthogonal coding modulation is then performed based on the equalization switch duty cycle sequence to generate a grouped orthogonal coded duty cycle sequence. The grouped orthogonal coded duty cycle sequence is then sent to the corresponding equalization branch to form an orthogonal coded excitation signal on the solid electrolyte cell. During the execution of the orthogonal coded excitation signal, the cell terminal voltage sequence, equalization branch current sequence, equalization branch voltage sequence, cell temperature sequence, and stacking pressure sequence are retrieved. A digital phase-locked loop amplification and demodulation algorithm is used to demodulate the phase of the grouped orthogonal coded duty cycle sequence and the cell terminal voltage sequence to obtain the cell voltage excitation response component. A least squares coding inversion algorithm with Tikhonov regularization is used to encode and invert the equalization branch current sequence, equalization branch voltage sequence, and cell voltage excitation response component to obtain the cell interface electrical response spectrum.

[0012] Further, the specific process of obtaining the single-cell interface response tensor through multi-band analysis and tensor splicing is as follows: Multi-band analysis is performed on the single-cell interface electrical response spectrum using Daubechies fourth-order wavelet packet decomposition to obtain the ohmic response component in the first band, the interface contact response component in the second band, and the polarization recovery response component in the third band; the thermal response component is extracted from the single-cell temperature sequence and the equilibrium branch current sequence using autoregressive exogenous input structure thermal dynamic parameter identification; the pressure response component is extracted from the stack pressure sequence and the single-cell voltage excitation response component using the Preisach hysteresis operator identification algorithm; the ohmic response component in the first band, the interface contact response component in the second band, the polarization recovery response component in the third band, the thermal response component, and the pressure response component are spliced ​​according to the single-cell cell position number to obtain the single-cell interface response tensor; the single-cell interface response tensor characterizes the electrical, thermal, and pressure responses of the solid electrolyte and electrode interface under equilibrium branch spectrum excitation.

[0013] Furthermore, the specific process of constructing a solid-state interface composite state-space model based on the individual cell interface response tensor and generating the interface bearing characteristic set is as follows: A solid-state interface composite state-space model is constructed based on the electrochemical response layer, solid-state interface contact layer, thermal response layer, pressure hysteresis layer, and observation mapping layer. The individual cell interface response tensor, individual cell terminal voltage sequence, battery pack total current sequence, individual cell temperature sequence, stacking pressure sequence, and cycle number are input into the solid-state interface composite state-space model. The solid-state interface composite state-space model consists of individual cell state vectors composed of reversible contact loss, irreversible contact loss, interface polarization memory, thermal sensitivity, and pressure hysteresis. The solid-state interface composite state-space model is recursively corrected online using unscented Kalman filtering, outputting the interface bearing characteristic set corresponding to each individual cell. The interface bearing characteristic set includes: reversible contact loss, irreversible contact loss, interface polarization... The system includes components for memory, thermal sensitivity, and pressure hysteresis. The second-band interface contact response component participates in the partial mapping correction of reversible and irreversible contact losses. The third-band polarization recovery response component participates in the correction of interface polarization memory, the thermal response component participates in the correction of thermal sensitivity, the pressure response component and the stacked pressure sequence participate in the correction of pressure hysteresis, and the cycle count participates in the cumulative update of irreversible contact losses. A state observation equation is formed based on the formation mechanism of the single-cell terminal voltage and the mapping relationship of the response components in the single-cell interface response tensor. Based on the state observation equation of the solid-state interface composite state-space model, the source decomposition of the single-cell terminal voltage sequence yields the energy effective voltage component, ohmic voltage drop component, interface contact voltage drop component, polarization memory voltage component, and thermo-coupling voltage offset component. The single-cell energy offset is obtained analytically from the energy effective voltage component and the total current sequence of the battery pack using a differential incremental capacity curve.

[0014] Furthermore, the specific process of forming a single interface acceptance domain based on the set of interface acceptance characteristics is as follows: the single interface acceptance domain is obtained by using the interval Taylor reachable set propagation algorithm for reversible contact loss, irreversible contact loss, interface polarization memory, thermal sensitivity and pressure hysteresis; the single interface acceptance domain includes the allowable equalization current amplitude range, the allowable equalization duration range, the allowable equalization energy direction, the allowable polarization residual boundary, the allowable heat load boundary and the allowable pressure disturbance boundary.

[0015] Furthermore, the specific process of constructing a high-energy-density solid-state electrolyte battery pack equalization energy topology table based on solid-state battery equalization correlation data and single-cell interface acceptance domain is as follows: Single-cell battery location numbers are constructed as equalization energy nodes according to the battery pack series connection relationship; equalization branch numbers are constructed as equalization energy edges according to the equalization branch connection relationship; and single-cell energy offset, single-cell interface acceptance domain, equalization branch current sequence, equalization branch voltage sequence, and equalization switch duty cycle sequence are written into the equalization energy topology table to obtain a high-energy-density solid-state electrolyte battery pack equalization energy topology table with interface constraints. Based on the high-energy-density solid-state electrolyte battery pack equalization energy topology table with interface constraints, a candidate equalization action set is established according to the energy offset direction, equalization branch connectivity, allowable equalization current amplitude range, allowable equalization duration range, and allowable equalization energy direction. The candidate equalization action set includes energy output cells, energy receiving cells, equalization branch numbers, equalization current waveforms, equalization duty cycle waveforms, equalization duration, and equalization recovery intervals.

[0016] Furthermore, the specific process of generating equalization branch control commands based on the control barrier function and driving equalization branch switches and controlling equalization current is as follows: A control barrier function is established using the normalized residual margin between the allowable boundary in the individual interface domain and the actual effect of candidate equalization actions. The set of candidate equalization actions is input into the control barrier function for boundary verification. The control barrier function imposes boundary constraints on the allowable equalization current amplitude range, allowable equalization duration range, allowable polarization residual boundary, allowable thermal load boundary, and allowable pressure disturbance boundary. When a candidate equalization action makes the control barrier function value less than zero, the corresponding candidate equalization action is excluded. When a candidate equalization action makes the control barrier function value greater than or equal to zero and less than the barrier holding threshold, a staggered intermittent equalization waveform is generated. When a candidate equalization action makes the control barrier function value greater than or equal to the barrier holding threshold, a continuous equalization waveform is generated. A mixed integer rolling optimization algorithm is used to jointly solve the equalization parameters in the set of candidate equalization actions. The target equalization path, target equalization current waveform, target equalization duty cycle waveform, and target equalization recovery interval are obtained by balancing the energy path and equalization current waveform. Based on the target equalization path, target equalization current waveform, target equalization duty cycle waveform, and target equalization recovery interval, a finite state machine control method and a pulse width modulation method are used to generate equalization branch control commands. The equalization branch control commands are then sent to the corresponding equalization branches to drive the equalization branch switches and control the equalization current. The switches of the corresponding equalization branches are controlled to turn on and off according to the target equalization duty cycle waveform, and the equalization current between the energy output unit and the energy receiving unit is adjusted according to the target equalization path and target equalization current waveform. During the equalization branch switch driving and equalization current control process, the control barrier function value is updated in real time according to the unit terminal voltage sequence, unit temperature sequence, and stacking pressure sequence. If the updated control barrier function value is less than zero, the mixed integer rolling optimization algorithm is re-executed and the target equalization path is updated.

[0017] Further, the specific process of writing the battery pack historical operation file and correcting the model calibration parameters based on the battery pack historical operation file is as follows: After equalization, a battery pack historical operation file is created, and the individual cell interface response tensor, reversible contact loss, irreversible contact loss, interface polarization memory, thermal sensitivity, pressure hysteresis, individual cell interface bearing domain, control barrier function trigger record, target equalization path, target equalization current waveform, target equalization duty cycle waveform, target equalization recovery interval, and equalization branch switch drive and equalization current control log are written into the battery pack historical operation file; based on the battery pack historical operation file, the model calibration parameters of the solid-state interface composite state space model are updated using the expectation-maximization algorithm.

[0018] Beneficial effects The present invention has the following beneficial effects: (1) This invention determines the energy path, constrains the balancing action, and drives the balancing branch in the balancing process of a high-energy-density solid electrolyte battery pack, so that the balancing control process can match the changes in the state of individual cells during the charging and discharging process of the battery pack. This achieves the effect of refined balancing control and stable energy distribution of the battery pack, effectively solving the problem of the battery pack's usable capacity being affected by the single basis of balancing control and the unclear energy regulation path in the prior art.

[0019] (2) In this invention, by identifying the interface response of solid electrolyte monomers under equilibration, and incorporating interface contact, polarization recovery, thermal state and pressure state into a unified judgment, the equilibration control no longer depends solely on the difference in monomer terminal voltage or state of charge, thereby achieving a more complete effect of identifying the true state of solid electrolyte monomers. This effectively solves the problem that the existing technology of equilibration based solely on apparent voltage difference is prone to distorted equilibration judgment.

[0020] (3) In this invention, by combining the balanced energy topology table with the single-unit interface receiving domain, the execution of the balanced action is screened and controlled, so that the balanced branch driving process can take into account both energy difference correction and single-unit receiving capacity, thereby realizing the effect of coordinated control of balanced path selection and solid-state interface protection, effectively solving the problem of uneven energy distribution caused by the lack of consideration of single-unit interface receiving characteristics in the balanced path selection in the prior art.

[0021] (4) In this invention, by forming a single-unit interface bearing domain and using a control barrier function to perform boundary verification on candidate equalization actions, the equalization current action, equalization duration process and recovery process are constrained by the single-unit interface bearing characteristic quantity, thereby achieving the synergistic effect of equalization action safety constraint and interface protection, effectively solving the problem of local heating and interface stress accumulation caused by high impedance single units continuously bearing equalization action in the prior art.

[0022] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0023] Figure 1 This is a flowchart of the high energy density battery equalization control method based on solid electrolyte of the present invention; Figure 2 This is a flowchart of the control barrier function boundary verification and equalization waveform generation of the present invention; Figure 3 This is a thermal visualization diagram of the set of characteristic quantities of the single interface of the present invention; Figure 4 This is a comparison chart of the single-unit energy offset and the control barrier function value of the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Please see Figures 1-4 This invention provides a technical solution: a high-energy-density battery equalization control method based on solid-state electrolyte, comprising the following steps: S1, collecting solid-state battery equalization correlation data and preprocessing the solid-state battery equalization correlation data; S2, dividing the operation window based on the solid-state battery equalization correlation data, generating a grouped orthogonal coded duty cycle sequence based on the operation window, and performing multi-layer frequency band analysis and tensor splicing to obtain a single-unit interface response tensor; S3, constructing a solid-state interface composite state space model based on the single-unit interface response tensor, generating a set of interface bearing characteristics; forming a single-unit interface bearing domain based on the set of interface bearing characteristics; S4, constructing a high-energy-density solid-state electrolyte battery pack equalization energy topology table based on the solid-state battery equalization correlation data and the single-unit interface bearing domain; generating equalization branch control instructions based on the control barrier function and performing equalization branch switch driving and equalization current control; S5, writing the battery pack historical operation file, and correcting the model correction parameters based on the battery pack historical operation file.

[0026] Specifically, the process of collecting solid-state battery equalization correlation data is as follows: Solid-state battery equalization correlation data is simultaneously collected from the battery management system, equalization branch sampling circuit, temperature sampling circuit, and stacking pressure sampling circuit of the high-energy-density solid-state electrolyte battery pack. The synchronous collection uses the battery management system clock as the time reference. A sampling timestamp is appended to the data collected by each sampling circuit. The sampling period is determined based on the driving period of the equalization switch drive pulse sequence, ensuring that the data from each sampling circuit can be aggregated to the same sampling moment according to the sampling timestamp. The solid-state battery equalization correlation data includes the individual cell location number, equalization branch number, equalization branch connection relationship, sampling timestamp, individual cell terminal voltage sequence, total battery pack current sequence, individual cell temperature sequence, equalization branch current sequence, equalization branch voltage sequence, equalization switch drive pulse sequence, stacking pressure sequence, and cycle count. The individual cell location number indicates the position of the individual cell in the series connection relationship of the high-energy-density solid-state electrolyte battery pack. The data is structured as follows: the single-cell voltage sequence is obtained through the synchronous sample-and-hold circuit of the single-cell voltage acquisition chip in the battery management system; the total current sequence of the battery pack is obtained by sampling from a Hall current sensor after low-pass anti-aliasing filtering; the equalization branch current sequence and equalization branch voltage sequence are obtained through differential sampling resistors and isolation operational amplifiers in the equalization branch sampling circuit; the single-cell temperature sequence is obtained by an NTC thermistor after conditioning with a Wheatstone bridge and instrumentation amplifier in the temperature sampling circuit; the equalization switch drive pulse sequence is captured by the MOSFET gate drive signal in the equalization branch sampling circuit after optocoupler isolation; the stacking pressure sequence is obtained by a piezoresistive pressure sensor after charge amplifier and active filtering in the stacking pressure sampling circuit; and the cycle count is extracted from the cumulative value of the charge-discharge cycle counter in the battery management system. The cycle count is defined as one charge-discharge cycle of the high-energy-density solid electrolyte battery pack, and is written together with the single-cell data under the same sampling timestamp into the solid-state battery equalization correlation data.

[0027] In this implementation scheme, by simultaneously acquiring data from the battery management system, the equalization branch sampling circuit, the temperature sampling circuit, and the stacking pressure sampling circuit, it is possible to fully capture the electrochemical state, temperature distribution, equalization current and voltage characteristics, switching timing, stacking pressure state, and cycle usage information of each cell in the high-energy-density solid-state electrolyte battery pack. This step forms a unified, multimodal, and high-precision raw data foundation, providing reliable quantitative support for subsequent equalization control, estimation of interface bearing characteristic sets, construction of solid-state interface composite state-space models, and control barrier constraints, enabling accurate evaluation of cell energy shift and interface bearing characteristic sets.

[0028] Specifically, the preprocessing of solid-state battery equalization correlation data involves: time alignment of the individual cell terminal voltage sequence, battery pack total current sequence, individual cell temperature sequence, equalization branch current sequence, equalization branch voltage sequence, equalization switch drive pulse sequence, and stacking pressure sequence according to the sampling timestamp. The time alignment is implemented using Lagrange quadratic interpolation to resample each sequence at equal intervals, unifying all sequences onto the same time reference axis. This resampled time reference axis serves as the target time grid for subsequent interpolation completion and model input. Isolated outliers in the individual cell terminal voltage sequence, individual cell temperature sequence, and stacking pressure sequence are removed using the Hampel filtering algorithm. The Hampel filtering algorithm identifies outliers by comparing the deviation of each data point from the median of its local window to whether it exceeds three times the absolute deviation of the median, and replaces these outliers with the median. The local window length is determined based on the drive cycle of the equalization switch drive pulse sequence, ensuring that the local window covers at least one complete drive cycle. The solid-state battery equalization correlation data corresponding to the sampling timestamps is supplemented using a cubic spline interpolation algorithm. Interpolation is performed when valid sampling points exist at both ends of consecutive missing segments. Missing data at the beginning or end of the sequence are processed according to adjacent valid sampling points. The cubic spline interpolation algorithm uses the continuity condition of the second derivative of four adjacent valid sampling points to fit a piecewise cubic polynomial under the natural boundary to recover missing values. The conduction duration and drive cycle of the equalization switch drive pulse sequence are statistically analyzed to obtain the equalization switch duty cycle sequence. The conduction duration and drive cycle statistics are obtained by capturing the time interval between consecutive rising edges of the pulse sequence as the drive cycle and the high-level width between the rising edge and the adjacent falling edge as the conduction duration. The individual cell terminal voltage sequence, battery pack total current sequence, individual cell temperature sequence, equalization branch current sequence, equalization branch voltage sequence, and stacking pressure sequence are standardized using the Z-score normalization algorithm. The Z-score standardization algorithm subtracts the mean of each sequence value and divides by the standard deviation to achieve dimensional uniformity and distribution centering. The mean and standard deviation are statistically analyzed for the same sequence within the same operating window. A sequence with a standard deviation of zero is treated as a constant sequence. The standardized data is used for interface response identification and correction of the solid-state interface composite state-space model. The max-min normalization algorithm normalizes the duty cycle sequence and cycle count of the equalization switch. This algorithm subtracts the minimum value of each sequence value and divides by the range to compress the numerical range to the zero-to-one interval. The maximum and minimum values ​​are derived from corresponding sequences within the same operating window; a sequence with a range of zero is treated as a constant sequence.

[0029] This implementation plan achieves unified timing reference, anomaly suppression, missing data repair, switch-driven feature extraction, and numerical scale unification of solid-state battery equalization correlation data. This forms a continuous, synchronous, stable, and comparable equalization control data foundation, reducing the interference of time offset, sensor noise, data breakpoints, and dimensional differences between multi-sampling links on subsequent analysis results. It also improves the synchronization consistency among single-cell terminal voltage sequences, single-cell temperature sequences, equalization branch current sequences, equalization branch voltage sequences, and stacking pressure sequences, enabling solid-state battery equalization correlation data to meet the data quality requirements for interface response identification, single-cell interface response tensor construction, solid-state interface composite state-space model correction, and control barrier function boundary verification.

[0030] Specifically, the process of dividing the operating window based on solid-state battery equalization correlation data and generating a grouped orthogonal coded duty cycle sequence based on the operating window is as follows: The total current sequence of the battery pack and the temperature sequence of each cell are divided into operating windows using a Bayesian online change point detection algorithm based on joint observation of current change rate and temperature rise slope. This results in a load smoothing window, a load transition window, a resting recovery window, and a heat release window. The Bayesian online change point detection algorithm estimates the posterior probability of the change point in real time using the covariance structure of the current change rate and temperature rise slope in the joint likelihood function. Based on the posterior probability of the change point, the sign of the current change rate, and the sign of the temperature rise slope, the start and end boundaries of each operating window are determined, achieving online segmentation of the multi-modal operating state. Within the load smoothing window, the resting recovery window, and the heat release window, the operating window is determined based on the cell location number and equalization... The branch number determines the cell to be excited and its corresponding equalization branch. Orthogonal coding modulation is performed based on the equalization switch duty cycle sequence. Orthogonal coding modulation uses Walsh code groups, which are sets of codes composed of mutually orthogonal binary codewords. These codes are used to distinguish the excitation signals of different equalization branches and serve as orthogonal basis functions. Walsh codewords with zero cross-correlation are assigned to each group of equalization branches according to the equalization branch number corresponding to the cell to be excited. The code length is determined according to the number of equalization branches in the same group. The duty cycle sequence of each group of equalization branches is multiplied bit by bit with its assigned orthogonal code, so that multiple equalization branches can be injected with non-interfering spectral excitation at the same time, generating a grouped orthogonal coded duty cycle sequence. This grouped orthogonal coded duty cycle sequence is then sent to the corresponding equalization branch, so that the equalization branch forms an orthogonal coded excitation signal on the solid electrolyte cell.

[0031] During the execution of the orthogonal coded excitation signal, the individual terminal voltage sequence, equalization branch current sequence, equalization branch voltage sequence, individual temperature sequence, and stacking pressure sequence are retrieved. The retrieval process is synchronized with the chip rate of the orthogonal coded excitation signal to ensure phase consistency between excitation and response sampling. A digital lock-in amplifier demodulation algorithm is used to demodulate the phase of the grouped orthogonal coded duty cycle sequence and the individual terminal voltage sequence to obtain the individual voltage excitation response component. The digital lock-in amplifier demodulation algorithm uses the orthogonal code modulation sequence as the reference signal and the codeword period of the grouped orthogonal coded duty cycle sequence as the demodulation window. The terminal voltage amplitude and phase response related to the excitation code pattern are extracted through dual-channel orthogonal mixing and low-pass filtering. A least-squares coding inversion algorithm with Tikhonov regularization is used to encode and invert the equalization branch current sequence, equalization branch voltage sequence, and individual voltage excitation response component to obtain the individual interface electrical response spectrum. Tikhonov regularization is a regularization method that suppresses excessive fluctuations in the results by adding a penalty term to the inversion objective. The grouped orthogonal encoded duty cycle sequence forms an encoding matrix. The individual voltage excitation response component is used as the inversion observation. The equalization branch current sequence and equalization branch voltage sequence are used as constraint inputs. The L2 norm penalty term is used as a smoothing term to constrain the sum of squares of the inversion results, constraining the smoothness of the solution. The regularization coefficient is determined according to the balance relationship between the inversion residual and the smoothness of the solution. The individual interface frequency domain impedance characteristics corresponding to each equalization branch are separated from the multi-coded superimposed response signal.

[0032] In this implementation scheme, this step achieves controllable spectral excitation and interface electrical response separation of solid electrolyte cells under stable operating window, avoiding interference from strong disturbance conditions such as load transitions on interface feature extraction, and improving phase consistency and signal discriminability between excitation injection and response sampling. The grouped orthogonal coding duty cycle sequence enables multiple equalization branches to synchronously form excitations without overlapping. The cell voltage excitation response component and cell interface electrical response spectrum can more accurately characterize the frequency domain impedance change of the solid electrolyte and electrode interface, providing a stable, separable, and traceable electrical response basis for subsequent extraction of the second frequency band interface contact response component, the third frequency band polarization recovery response component, and the construction of the cell interface response tensor.

[0033] Specifically, the process of obtaining the single-unit interface response tensor through multi-level frequency band analysis and tensor splicing is as follows: The single-unit interface electrical response spectrum is analyzed using Daubechies fourth-order wavelet packet decomposition. Daubechies fourth-order wavelet packet decomposition involves frequency division analysis of the signal using a compactly supported orthogonal wavelet basis to obtain the ohmic response component in the first frequency band, the interface contact response component in the second frequency band, and the polarization recovery response component in the third frequency band. Daubechies fourth-order wavelet packet decomposition is then performed with a db4 wavelet basis for three-level full decomposition of the spectrum. The db4 wavelet basis is a Daubechies fourth-order wavelet basis used to maintain the ability to characterize local abrupt changes in the frequency band decomposition. Based on the frequency band energy distribution, the low-frequency band is classified as polarization recovery, the mid-frequency band as interface contact, and the high-frequency band as the physical source of the ohmic response. The boundaries of each frequency band are determined by the frequency band order after the three-level full decomposition and mapped according to the difference in response speed: high-frequency band corresponds to fast ohmic response, mid-frequency band corresponds to interface contact change, and low-frequency band corresponds to the polarization recovery process. The single-unit temperature sequence is then analyzed. The thermal dynamic parameters of the equalization branch current sequence are identified using an autoregressive exogenous input structure to extract the thermal response component. The autoregressive exogenous input structure uses the equalization branch current sequence as the exogenous excitation and the individual temperature sequence as the autoregressive response. The identification window is consistent with the operating window, and the structural order is determined based on the stability of the identification residual. The thermal time constant and gain coefficient of temperature change relative to current excitation are obtained through recursive least squares identification. The pressure response component is extracted from the stacked pressure sequence and the individual voltage excitation response component using the Preisach hysteresis operator identification algorithm. The Preisach hysteresis operator identification algorithm is an operator identification method used to describe the input-output hysteresis relationship. The Preisach hysteresis operator identification algorithm takes the stacked pressure sequence as input and the individual voltage excitation response component as output. The first effective sample value of the stacked pressure sequence is used as the initial state for hysteresis identification. The Preisach plane weight function is filled with a first-order inversion curve to obtain the voltage hysteresis loop feature caused by pressure change, and the voltage hysteresis loop feature is used as the pressure response component.

[0034] The ohmic response components of the first frequency band, the interface contact response components of the second frequency band, the polarization recovery response components of the third frequency band, the thermal response components, and the pressure response components are tensor-concatenated according to the cell location number to obtain the cell interface response tensor. The tensor concatenation uses the cell location number as an index to construct a third-order response tensor structure along the frequency band dimension, the physical field dimension, and the time dimension. The frequency band dimension is arranged according to the ohmic response components of the first frequency band, the interface contact response components of the second frequency band, and the polarization recovery response components of the third frequency band; the physical field dimension is arranged according to the electrical response, thermal response, and pressure response; and the time dimension is arranged along the time reference axis after the sampling timestamp is aligned. Before concatenation, each response component adopts the data scale after the aforementioned standardization processing. The cell interface response tensor characterizes the electrical response, thermal response, and pressure response of the solid electrolyte and electrode interface under the equal branch spectrum excitation, and expresses the response components of different physical fields in a unified data framework to form a multidimensional feature descriptor of the interface bearing feature set.

[0035] In this implementation scheme, this step separates the frequency domain and time domain characteristics of the multi-physics response, integrates the electrical, thermal, and pressure responses of a single interface in a unified tensor form, forming a continuous, three-dimensional, and indexable single interface response tensor. This tensor can clearly characterize the Ohmic response, interface bearing characteristics, polarization recovery behavior, thermal dynamics, and pressure hysteresis effect. The components extracted by multi-layer wavelet decomposition, thermal dynamic identification, and Preisach hysteresis operator are uniformly expressed in a third-order tensor, providing a high-precision, multi-dimensional, and traceable feature foundation for subsequent solid-state interface composite state-space model construction, interface bearing domain evaluation, and control barrier function constraints, thereby improving the reliability and pertinence of equilibrium control decisions.

[0036] Specifically, the process of constructing a solid-state interface composite state-space model based on the individual cell interface response tensor and generating the set of interface-bearing characteristic quantities is as follows: A solid-state interface composite state-space model is constructed based on an electrochemical response layer, a solid-state interface contact layer, a thermal response layer, a pressure hysteresis layer, and an observation mapping layer. The electrochemical response layer receives the first-band ohmic response component and the third-band polarization recovery response component; the solid-state interface contact layer receives the second-band interface contact response component and the number of cycles; the thermal response layer receives the thermal response component and the individual cell temperature sequence; the pressure hysteresis layer receives the pressure response component and the stacking pressure sequence; and the observation mapping layer establishes a state observation equation based on the individual cell terminal voltage formation mechanism. The individual cell interface response tensor, individual cell terminal voltage sequence, total battery pack current sequence, individual cell temperature sequence, stacking pressure sequence, and number of cycles are input into the solid-state interface composite state-space model. The joint evolution of electrochemical processes, interface degradation, thermal accumulation, and pressure hysteresis is achieved through cross-correlation terms of the individual cell state vectors. These cross-correlation terms are derived from the first-band ohmic response tensor. The synchronous change relationship between the components, the second-band interface contact response component, the third-band polarization recovery response component, the thermal response component, and the pressure response component is formed. The state update input is the single-cell interface response tensor, the total current sequence of the battery pack, the single-cell temperature sequence, the stacking pressure sequence, and the number of cycles. The observation output is the voltage component corresponding to the single-cell terminal voltage sequence. The solid-state interface composite state space model uses reversible contact loss, irreversible contact loss, interface polarization memory, thermal sensitivity, and pressure hysteresis to form the single-cell state vector. All of the above quantities are dimensionless quantities obtained by standardizing the corresponding response components. Among them, reversible contact loss represents the short-time recoverable change of the interface contact response, irreversible contact loss represents the cumulative deviation of the interface contact response with the number of cycles, interface polarization memory represents the hysteresis of the polarization recovery response, thermal sensitivity represents the response of the single-cell temperature sequence relative to the equilibrium branch current sequence, and pressure hysteresis represents the hysteresis of the single-cell voltage excitation response component under the action of the stacking pressure sequence.

[0037] Online calibration of the solid-state interface composite state-space model is performed using unscented Kalman filtering recursion, outputting a set of interface bearing characteristics for each individual component. These characteristics include reversible contact loss, irreversible contact loss, interface polarization memory, thermal sensitivity, and pressure hysteresis. The unscented Kalman filtering recursion propagates between state prediction and observation updates using a deterministic sampling point set, avoiding the analytical derivation of the Jacobian matrix in strongly nonlinear multiphysics coupled models. The second-band interface contact response component participates in the partial mapping calibration of reversible and irreversible contact losses. Specifically, the second-band interface contact response component corresponds to the reversible contact loss based on the difference between adjacent operating windows, and to the irreversible contact loss based on the deviation from the initial interface contact response component. The third-band polarization recovery response... The components participate in the correction of interface polarization memory, where the third-band polarization recovery response component corresponds to the interface polarization memory according to the recovery change between adjacent operating windows; the thermal response component participates in the correction of thermal sensitivity, where the thermal response component corresponds to the thermal sensitivity according to the response change of the unit temperature sequence under the action of the equalization branch current sequence; the pressure response component and the stacked pressure sequence participate in the correction of pressure hysteresis, where the pressure response component corresponds to the pressure hysteresis according to the hysteresis change of the unit voltage excitation response component under the action of the stacked pressure sequence; the number of cycles participates in the cumulative update of irreversible contact loss, where the number of cycles is cumulatively constrained for irreversible contact loss according to the cumulative increase relationship, and each response component is assigned to the update equation of the corresponding characteristic quantity by the observation residual weighting coefficient in the correction process, so as to realize the directional fusion of multi-source feedback information.

[0038] A state observation equation is formed based on the formation mechanism of the single-cell terminal voltage and the mapping relationship of the response components in the single-cell interface response tensor. This equation is used to separate the ohmic voltage drop caused by current, the interface contact voltage drop caused by interface contact degradation, the polarization memory voltage caused by polarization recovery hysteresis, and the thermo-coupled voltage shift caused by temperature and stacking pressure from the single-cell terminal voltage sequence. The state observation equation takes the single-cell state vector as input, maps reversible and irreversible contact losses to interface contact voltage drop components, interface polarization memory to polarization memory voltage components, heat-sensitive quantities and pressure hysteresis to thermo-coupled voltage shift components, and the first-band ohmic response component to ohmic voltage drop components. Based on the state observation equation of the solid-state interface composite state-space model, the source decomposition of the single-cell terminal voltage sequence is performed to obtain the energy effective voltage component, ohmic voltage drop component, interface contact voltage drop component, polarization memory voltage component, and thermo-coupled voltage shift component. The energy effective voltage component and the total current sequence of the battery pack are analyzed using differential incremental capacity curves. The analysis of the individual cell energy offset and the differential incremental capacity curve analysis includes: calculating the individual cell throughput capacity sequence based on the total current sequence of the battery pack and the sampling timestamp; performing differential differentiation on the effective voltage component of the energy in the individual cell throughput capacity sequence to obtain the individual cell incremental capacity curve; extracting the main peak position and main peak area of ​​the individual cell incremental capacity curve, and calculating the offset with the median value of the main peak position and the median value of the main peak area of ​​each cell in the same battery pack to obtain the individual cell energy offset. The median value of the main peak position and the median value of the main peak area serve as the reference benchmark within the same battery pack. The individual cell energy offset is the deviation of the individual cell incremental capacity curve from the reference benchmark. A positive individual cell energy offset indicates that the corresponding cell has an output energy offset relative to the reference benchmark, and a negative individual cell energy offset indicates that the corresponding cell has a receive energy offset relative to the reference benchmark. The individual cell throughput capacity sequence is obtained by integrating the total current sequence of the battery pack with the sampling timestamp over time and converting it to hours. The differential differentiation uses a first-order central difference within a sliding window to suppress noise amplification.

[0039] In this implementation scheme, this step achieves unified modeling, online correction, and voltage source decoupling of the multi-physics state of the individual interface. It integrates electrochemical response, solid-state interface contact degradation, thermal accumulation, and pressure hysteresis into the solid-state interface composite state-space model, enabling continuous estimation and dynamic correction of the reversible contact loss, irreversible contact loss, interface polarization memory, thermal sensitivity, and pressure hysteresis of each individual. By performing source decomposition on the individual terminal voltage sequence using the state observation equation, the effective voltage component of the energy truly involved in energy balance judgment can be distinguished from the ohmic voltage drop component, interface contact voltage drop component, polarization memory voltage component, and thermo-coupling voltage offset component, avoiding misjudging apparent voltage anomalies caused by interface degradation or thermal disturbances as energy inconsistency. This step generates a more accurate set of interface acceptance characteristics and individual energy offsets, providing a reliable state basis for subsequent generation of individual interface acceptance domains, acceptance judgment of balance actions, and control barrier function constraints.

[0040] Specifically, the process of forming a single interface bearing domain based on the set of interface bearing characteristics is as follows: The reversible contact loss, irreversible contact loss, interface polarization memory, thermal sensitivity, and pressure hysteresis are used to obtain the single interface bearing domain through an interval Taylor reachable set propagation algorithm. The single interface bearing domain uses the allowable equilibrium current amplitude range, allowable equilibrium duration range, allowable equilibrium energy direction, allowable polarization residual boundary, allowable heat load boundary, and allowable pressure disturbance boundary as its domain boundaries. The interval Taylor reachable set propagation algorithm uses the current estimation interval of each characteristic as its initial value. The current estimation interval of each characteristic is determined by the estimated value recursively output by the unscented Kalman filter and the observation residual weighting coefficient. A Taylor series expansion is performed one step along the solid-state interface composite state-space model, and the interval envelope is taken. Each recursive step corresponds to the time interval between adjacent sampling timestamps, obtaining the reachable range of each characteristic at the next moment. The single interface bearing domain includes the allowable equilibrium current amplitude range, allowable equilibrium duration range, allowable equilibrium energy direction, allowable... The polarization residual boundary, allowable heat load boundary, and allowable pressure disturbance boundary are all derived from the reachable range of their respective characteristic quantities. The transformation rules are as follows: the reachable range related to interface contact is used to narrow the allowable equilibrium current amplitude range; the reachable range of heat-sensitive quantity is used to narrow the allowable equilibrium duration range and allowable heat load boundary; the reachable range of interface polarization memory quantity is used to determine the allowable polarization residual boundary; the reachable range of pressure hysteresis quantity is used to determine the allowable pressure disturbance boundary; the sign of reversible contact loss quantity is used to determine the allowable equilibrium energy direction; the allowable equilibrium current amplitude range is jointly constrained by reversible contact loss quantity, irreversible contact loss quantity, and ohmic response; the allowable equilibrium duration range is constrained by heat-sensitive quantity and heat capacity parameter; the allowable equilibrium energy direction is constrained by the sign of reversible contact loss quantity; the allowable polarization residual boundary is constrained by the recovery time constant of interface polarization memory quantity; the allowable heat load boundary is constrained by heat-sensitive quantity and safe temperature rise threshold; and the allowable pressure disturbance boundary is constrained by the hysteresis width of pressure hysteresis quantity.

[0041] In this implementation scheme, this step expands the set of individual interface acceptance characteristics to the future reachable range using the interval Taylor reachable set propagation algorithm, forming an individual interface acceptance domain. This quantifies the boundary constraints for states such as reversible contact loss, irreversible contact loss, interface polarization memory, thermal sensitivity, and pressure hysteresis. The generated interface acceptance domain clarifies the allowable ranges for equalization current amplitude, equalization duration, energy transfer direction, polarization residue, thermal load, and pressure disturbance. This provides safety constraints for the admission judgment of equalization action, ensuring that the individual interface will not suffer irreversible damage due to excessive current, heat accumulation, or pressure fluctuations during the equalization process, thus achieving coordinated control of energy consistency correction and solid-state interface protection.

[0042] Specifically, the process of constructing a high-energy-density solid-state electrolyte battery pack balanced energy topology table based on solid-state battery equalization correlation data and single-cell interface acceptance domain is as follows: the single-cell battery location numbers are constructed as equalization energy nodes according to the battery pack series connection relationship, the equalization branch numbers are constructed as equalization energy edges according to the equalization branch connection relationship, and the single-cell energy offset, single-cell interface acceptance domain, equalization branch current sequence, equalization branch voltage sequence, and equalization switch duty cycle sequence are written into the equalization energy topology table to obtain a high-energy-density solid-state electrolyte battery pack balanced energy topology table with interface constraints. The balanced energy topology table stores the connection weights of nodes and edges in the form of an adjacency matrix, where the weights are jointly determined by the equalization branch current sequence and the allowable interval of the single-cell interface acceptance domain.

[0043] Based on the energy balancing topology table of a high-energy-density solid-state electrolyte battery pack with interface constraints, a candidate balancing action set is established according to the energy offset direction, balancing branch connectivity, allowable balancing current amplitude range, allowable balancing duration range, and allowable balancing energy direction. Specifically, the energy output cell and energy receiving cell are determined based on the energy offset of the individual cells. The available balancing branch numbers are determined based on the connectivity of the balancing energy edges. The allowable balancing current amplitude range, allowable balancing duration range, and allowable balancing energy direction in the cell interface domain are used as admission conditions to generate the candidate balancing action set. The candidate balancing action set includes the energy output cell, energy receiving cell, balancing branch number, balancing current waveform, balancing duty cycle waveform, balancing duration, and balancing recovery interval. The balancing current waveform includes three selectable waveform templates: square wave, trapezoidal wave, and sine wave. The balancing duty cycle waveform is defined by the ratio sequence of the balancing branch conduction time to the balancing period. The balancing recovery interval is the forced resting time between two adjacent balancing operations and is determined based on the time required for polarization residue, thermal load, and pressure disturbance to fall back to the corresponding allowable boundary after the candidate balancing current waveform ends.

[0044] In this implementation scheme, this step structures the physical and state information of individual cells and balancing branches within the battery pack into a balancing energy topology table. Nodes represent individual cells, and edges represent balancing branches. The energy offset of individual cells, interface domains, and the current, voltage, and duty cycle sequences of balancing branches are encoded as weights, enabling quantifiable modeling of energy transfer between nodes. Based on this, a set of candidate balancing actions is generated according to the energy offset direction, branch connectivity, and interface constraints. The output cell, receiving cell, branch number, current waveform, duty cycle waveform, duration, and recovery interval for each balancing operation are clearly defined. This provides a systematic and operable balancing action decision framework for subsequent mixed-integer rolling optimization and control barrier function constraints.

[0045] Specifically, the process of generating equalization branch control commands based on the control barrier function and driving equalization branch switching transistors and equalization current control is as follows: A control barrier function is established using the normalized margin between the allowable boundaries in the individual interface domain and the actual action corresponding to the candidate equalization action. The allowable boundaries include: the upper limit of the allowable equalization current amplitude, the upper limit of the allowable equalization duration, the upper limit of the allowable polarization residue, the upper limit of the allowable thermal load, the upper limit of the allowable pressure disturbance, and the allowable equalization energy direction. The actual action consists of the current, duration, polarization residue, thermal load, and pressure disturbance expected to be generated by the candidate equalization action, calculated from the candidate equalization current waveform, equalization duration, equalization recovery interval, interface polarization memory, thermal sensitivity, pressure hysteresis, and stacked pressure sequence, respectively. These are then compared with the corresponding allowable boundaries to calculate the normalized margin. Each normalized margin is calculated based on the difference between the allowable upper limit and the corresponding actual action, using the corresponding allowable upper limit as the normalization benchmark. The control barrier function value is the minimum value among the normalized margins and the direction consistency term. The control barrier function is: ; in, For candidate equilibrium actions The corresponding control barrier function value; The upper limit of the allowable equalization current amplitude represents the upper limit value extracted from the allowable equalization current amplitude range in the single-unit interface bearing domain; The candidate equalization current amplitude is derived from the equalization current waveform in the candidate equalization action set. The upper limit of the allowed equilibrium duration represents the upper limit value extracted from the allowed equilibrium duration range in the single-entity interface domain; The duration of the candidate equilibrium is derived from the set of candidate equilibrium actions; The upper limit of permissible polarization residue represents the upper limit value extracted from the permissible polarization residue boundary in the monomer interface acceptance domain; The candidate polarization residual is obtained from the candidate equalization current waveform, equalization recovery interval, and interface polarization memory. The upper limit of allowable heat load represents the upper limit value extracted from the allowable heat load boundary in the unit interface bearing area; The candidate heat load is obtained from the square integral of the current of the candidate equilibrium current waveform and the heat-sensitive quantity. To allow for an upper limit on pressure disturbance, the pressure fluctuation amplitude is extracted from the stacked pressure sequence, and the candidate equilibrium current change rate sequence is obtained from the first difference of the candidate equilibrium current waveform. Then, a first-order hysteresis convolution is performed on the candidate equilibrium current change rate sequence based on the pressure hysteresis, and the absolute value of the peak value is obtained by superimposing the convolution result and the pressure fluctuation amplitude. The candidate pressure disturbance is obtained from the rate of change of the candidate equilibrium current waveform, the pressure hysteresis, and the stacked pressure sequence. For the direction consistency term, when the candidate equilibrium energy direction belongs to the allowable equilibrium energy direction. Set to 1, when the candidate equilibrium energy direction does not belong to the allowed equilibrium energy direction. Take -1.

[0046] like Figure 2 The diagram shows the flowchart for boundary verification and equalization waveform generation of the control barrier function. The candidate equalization action set is input into the control barrier function for boundary verification, and directional consistency constraints are applied to the allowable equalization energy direction. The control barrier function imposes boundary constraints on the allowable equalization current amplitude range, allowable equalization duration range, allowable polarization residual boundary, allowable heat load boundary, and allowable pressure disturbance boundary. The boundary verification combines a hard constraint strategy of directly eliminating negative margin actions after item-by-item margin calculation with an admission mechanism that retains positive margin actions. When a candidate equalization action makes the control barrier function value less than zero, the corresponding candidate equalization action is excluded. A negative control barrier function value indicates that at least one boundary constraint has been broken, and the candidate equalization action is marked as infeasible. A control barrier function value greater than or equal to zero indicates that the candidate equalization action has not broken the allowable boundary in the individual interface bearing domain. When a candidate equalization action makes the control barrier function value greater than or equal to zero and less than the barrier holding threshold, a staggered intermittent equalization waveform is generated. The staggered intermittent equalization waveform inserts a forced cooling intermittent segment with a zero duty cycle during the equalization duration to reduce heat load and polarization accumulation rate. A continuous equalization waveform is generated when the candidate equalization action causes the control barrier function value to be greater than or equal to the barrier holding threshold. The continuous equalization waveform executes uninterruptedly with a constant duty cycle during the equalization duration, and is suitable for operating conditions with ample safety margins.

[0047] A mixed-integer rolling optimization algorithm is used to jointly solve for the equalization energy path and equalization current waveform in the candidate equalization action set, obtaining the target equalization path, target equalization current waveform, target equalization duty cycle waveform, and target equalization recovery interval. The mixed-integer rolling optimization algorithm constructs a mixed decision space with integer selection variables of the equalization energy path and continuous parameter variables of the equalization current waveform. Constraints include the candidate equalization action set, equalization branch connectivity, unit interface bearing domain, and control barrier function value being greater than or equal to zero. The algorithm performs a rolling solution within a finite prediction time domain with the objective of minimizing the weighted sum of energy equalization rate and control barrier function value. Based on the target equalization path, target equalization current waveform, target equalization duty cycle waveform, and target equalization recovery interval, a finite state machine control method and pulse width modulation method are used to generate equalization branch control commands. These commands are then issued to the corresponding equalization branches, controlling the switching transistors of the corresponding equalization branches to turn on and off according to the target equalization duty cycle waveform. Finally, the algorithm balances the current between the energy output unit and the energy receiving unit according to the target equalization path and target equalization current waveform. The flow is adjusted to ensure that the direction of the equalizing current is consistent with the target equalizing path, and that the trend of the equalizing current changes in accordance with the target equalizing current waveform. The switching transistor drive signal of the equalizing branch is generated by pulse width modulation of the target equalizing duty cycle waveform, and the switching transistor drive is paused and resumed according to the target equalizing recovery interval. During the pause of the switching transistor drive, the corresponding equalizing branch is kept in an open state. When the switching transistor drive is resumed, the drive signal continues to be output according to the target equalizing duty cycle waveform. During the switching transistor drive and equalizing current adjustment process, the control barrier function value is updated in real time according to the individual terminal voltage sequence, individual temperature sequence, and stacking pressure sequence. The updated control barrier function value is used as the basis for determining whether the current equalizing branch should continue to drive, pause drive, or be re-optimized. If the updated control barrier function value is less than zero, the mixed integer rolling optimization algorithm is re-executed and the target equalizing path, target equalizing current waveform, target equalizing duty cycle waveform, and target equalizing recovery interval are updated synchronously. Then, the equalizing branch control command is regenerated according to the updated target equalizing path, target equalizing current waveform, target equalizing duty cycle waveform, and target equalizing recovery interval.

[0048] In this embodiment, Table 1 is a set of characteristic data of the interface of a single unit, which records in detail the normalized values ​​of the energy offset, irreversible contact loss, interface polarization memory, thermal sensitivity, pressure hysteresis and control barrier function value of different units during the equalization control process. It is used to quantify the comprehensive constraint characteristics of the unit under the constraint of equalization action. Specifically: For monomer 1, the monomer energy shift is 0.06, irreversible contact loss is 0.08, interface polarization memory is 0.12, thermal sensitivity is 0.16, pressure hysteresis is 0.10, and the control barrier function value is 0.61; for monomer 2, the monomer energy shift is 0.11, irreversible contact loss is 0.15, interface polarization memory is 0.21, thermal sensitivity is 0.25, pressure hysteresis is 0.19, and the control barrier function value is 0.44; for monomer 3, the monomer energy shift is 0.18, irreversible contact loss is 0.27, interface polarization memory is 0.36, thermal sensitivity is 0.41, pressure hysteresis is 0.33, and the control barrier function value is 0.26. The monomer energy shift of monomer 4 is 0.23, the irreversible contact loss is 0.39, the interface polarization memory is 0.51, the thermal sensitivity is 0.57, the pressure hysteresis is 0.48, and the control barrier function value is 0.09; the monomer energy shift of monomer 5 is 0.29, the irreversible contact loss is 0.55, the interface polarization memory is 0.68, the thermal sensitivity is 0.72, the pressure hysteresis is 0.64, and the control barrier function value is -0.05; the monomer energy shift of monomer 6 is -0.14, the irreversible contact loss is 0.18, the interface polarization memory is 0.24, the thermal sensitivity is 0.30, the pressure hysteresis is 0.22, and the control barrier function value is 0.37. Table 1, with its normalized data, intuitively reflects the differences in the monomers under different interface characteristics and control constraints, and can serve as a basis for selecting equilibration actions and verifying the control barrier function.

[0049] Table 1 Data table of characteristic quantities of a single interface

[0050] like Figure 3The image shows a thermal visualization of the set of interface bearing characteristics for individual cells. Combined with Table 1, it visually demonstrates the changes and differences in these characteristics among different cells during the equilibrium process. Cell 1 has an energy shift of 0.06 and a control barrier function value of 0.61, indicating relatively low overall interface bearing characteristics. Cell 5 has an energy shift of 0.29 and a control barrier function value of -0.05, indicating relatively high overall interface bearing characteristics that are close to the constraint boundary. Cell 6 has an energy shift of -0.14 and a control barrier function value of 0.37, showing that its energy direction differs from the other cells. Overall, the thermal map, through color gradients and numerical annotations, reflects the distribution characteristics of individual cell energy shifts, irreversible contact losses, interface polarization memory, thermal sensitivity, and pressure hysteresis, providing a visual reference for equilibrium control decisions.

[0051] like Figure 4 The figure shows a comparison between the energy offset of individual cells and the control barrier function value. The black stepped line represents the energy offset of individual cells, the blue and red filled areas represent positive and negative energy offsets respectively, and the green dashed line represents the control barrier function value. The trend in the figure shows that as the energy offset of an individual cell increases, the control barrier function value gradually decreases. For some cells, such as cell 5, the control barrier function value is already negative, indicating that its candidate equalization action is restricted. Cell 6, with a different energy offset direction, shows a negative value range, but its control barrier function remains positive. Figure 4 It intuitively demonstrates the correlation between the energy offset of a single cell and the value of the control barrier function, which can be used to assess the constraint risk and strategy adjustment of a single cell in equilibrium control.

[0052] In this implementation scheme, this step achieves the quantification of safety margin for candidate equalization actions, boundary admission screening, hierarchical generation of equalization waveforms, and dynamic optimization of the equalization path. The control barrier function incorporates equalization current amplitude, equalization duration, polarization residue, thermal load, pressure disturbance, and equalization energy direction into the constraint judgment, enabling it to identify and eliminate inviolable actions that exceed the interface bearing capacity of individual cells, preventing high-impedance cells from continuing to bear excessively strong equalization current or excessively long equalization duration when the interface bearing capacity is insufficient. For candidate equalization actions with low safety margins but not exceeding the limits, staggered intermittent equalization waveforms are generated to reduce the risk of thermal load accumulation and polarization residue; for candidate equalization actions with sufficient safety margins, continuous equalization waveforms are generated to improve equalization efficiency. The mixed integer rolling optimization algorithm jointly determines the target equalization path, target equalization current waveform, target equalization duty cycle waveform, and target equalization recovery interval from the candidate equalization action set, enabling equalization control to take into account both energy offset correction efficiency and solid-state interface protection requirements, forming an equalization execution strategy that can be updated in real time and corrected in a closed loop.

[0053] Specifically, the process of writing the battery pack historical operation file and correcting the model calibration parameters based on the battery pack historical operation file is as follows: After equalization, a battery pack historical operation file is created, and the individual cell interface response tensor, reversible contact loss, irreversible contact loss, interface polarization memory, thermal sensitivity, pressure hysteresis, individual cell interface bearing domain, control barrier function trigger record, target equalization path, target equalization current waveform, target equalization duty cycle waveform, target equalization recovery interval, and equalization branch switch drive and equalization current control log are written into the battery pack historical operation file. The battery pack historical operation file uses equalization events as the recording unit, and each record is accompanied by a complete traceability identifier and timestamp index. Based on the battery pack historical operation file, the expectation-maximization algorithm is used to update the model calibration parameters of the solid-state interface composite state-space model. The model calibration parameters include state transition parameters and observation mapping. The parameter and noise covariance parameters, and the model calibration parameters are updated using a batch offline learning method. The accumulated equalization event records are used as the training sample set. The process noise covariance and observation noise covariance of the state-space model are re-estimated using the expectation-maximization algorithm. The updated model calibration parameters are then used to estimate the interface acceptance feature set in the next running window. The equalization branch switch drive and equalization current control logs include equalization branch number, equalization branch control command, equalization start timestamp, equalization end timestamp, actual equalization current sequence, actual equalization duty cycle sequence, individual energy offset change record, and control barrier function value change record. The state prediction error is obtained by the difference between the predicted interface acceptance feature set and the corrected interface acceptance feature set. The observation residual is obtained by the difference between the terminal voltage estimate output by the state observation equation and the individual terminal voltage sequence.

[0054] In this implementation plan, this step achieves complete archiving of equalization process data, interface state data, control constraint data, and execution result data, forming a battery pack historical operation archive based on equalization events. This ensures that each equalization action has a clear state source, control basis, execution path, waveform parameters, and result record. The battery pack historical operation archive provides data support for subsequent operation tracing, interface degradation trend analysis, verification of control barrier function triggering causes, and evaluation of equalization strategy effectiveness. Updating the parameters of the solid-state interface composite state-space model based on the accumulated equalization event records can improve the model's adaptability to different cycle stages, different interface degradation degrees, and different thermo-coupling conditions, making subsequent estimation of interface bearing characteristic sets, generation of individual interface bearing domains, and equalization control decisions more accurate and stable.

[0055] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0056] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A high-energy-density battery equalization control method based on solid-state electrolytes, characterized in that, Includes the following steps: S1, collect solid-state battery balancing correlation data and preprocess the solid-state battery balancing correlation data; S2, divide the running window based on the solid-state battery equalization correlation data, generate a grouped orthogonal coded duty cycle sequence based on the running window, and perform multi-layer frequency band analysis and tensor splicing to obtain the single-unit interface response tensor; S3, based on the single interface response tensor, constructs a solid-state interface composite state space model and generates a set of interface-bearing feature quantities; A single interface acceptance domain is formed based on the set of interface acceptance feature quantities; S4. Construct a high-energy-density solid-state electrolyte battery pack balance energy topology table based on solid-state battery balance correlation data and single-cell interface domain; generate balance branch control commands based on control barrier functions and perform balance branch switching transistor drive and balance current control. S5 writes the battery pack's historical operation records and corrects the model's calibration parameters based on the battery pack's historical operation records.

2. The high-energy-density battery equalization control method based on solid-state electrolyte according to claim 1, characterized in that: The specific process for collecting solid-state battery equalization correlation data is as follows: Solid-state battery balancing correlation data is synchronously collected from the battery management system, balancing branch sampling circuit, temperature sampling circuit, and stacking pressure sampling circuit of the high-energy-density solid-state electrolyte battery pack. The solid-state battery balancing correlation data includes the cell location number, balancing branch number, balancing branch connection relationship, sampling timestamp, cell terminal voltage sequence, total battery pack current sequence, cell temperature sequence, balancing branch current sequence, balancing branch voltage sequence, balancing switch drive pulse sequence, stacking pressure sequence, and cycle number.

3. The high-energy-density battery equalization control method based on solid-state electrolyte according to claim 1, characterized in that: The specific process for preprocessing the solid-state battery equalization correlation data is as follows: The individual cell terminal voltage sequence, battery pack total current sequence, individual cell temperature sequence, equalization branch current sequence, equalization branch voltage sequence, equalization switch drive pulse sequence, and stacking pressure sequence are time-aligned according to the sampling timestamp; isolated outliers in the individual cell terminal voltage sequence, individual cell temperature sequence, and stacking pressure sequence are removed using the Hampel filtering algorithm; and solid-state battery equalization correlation data corresponding to the sampling timestamp are completed using the cubic spline interpolation algorithm. The on-time duration and driving cycle of the equalization switch drive pulse sequence are statistically analyzed to obtain the equalization switch duty cycle sequence; the individual cell terminal voltage sequence, battery pack total current sequence, individual cell temperature sequence, equalization branch current sequence, equalization branch voltage sequence, and stacking pressure sequence are standardized using the Z-score normalization algorithm. The duty cycle sequence and number of cycles of the equalization switch are normalized using the max-min normalization algorithm.

4. The high-energy-density battery equalization control method based on solid-state electrolyte according to claim 1, characterized in that: The specific process of dividing the running window based on the solid-state battery equalization correlation data and generating a grouped orthogonal coded duty cycle sequence based on the running window is as follows: The total current sequence of the battery pack and the temperature sequence of each cell are divided into operating windows using a Bayesian online change point detection algorithm based on joint observation of current change rate and temperature rise slope. These windows are divided into load smoothing window, load transition window, static recovery window, and heat release window. Within these windows, the cells to be excited and their corresponding equalization branches are determined based on the cell location number and equalization branch number. Orthogonal coding modulation is then performed based on the equalization switch duty cycle sequence to generate a grouped orthogonal coding duty cycle sequence. This grouped orthogonal coding duty cycle sequence is then sent to the corresponding equalization branch, enabling the equalization branch to form an orthogonal coding excitation signal on the solid electrolyte cell. During the execution of the orthogonal coding excitation signal, the individual terminal voltage sequence, equalization branch current sequence, equalization branch voltage sequence, individual temperature sequence, and stacking pressure sequence are retrieved. A digital phase-locked loop amplifier demodulation algorithm is used to demodulate the phase of the grouped orthogonal coding duty cycle sequence and the individual terminal voltage sequence to obtain the individual voltage excitation response component. A least squares coding inversion algorithm with Tikhonov regularization is used to encode and invert the equalization branch current sequence, equalization branch voltage sequence, and individual voltage excitation response component to obtain the individual interface electrical response spectrum.

5. The high-energy-density battery equalization control method based on solid-state electrolyte according to claim 1, characterized in that: The specific process of obtaining the single interface response tensor by performing multi-layer frequency band analysis and tensor concatenation is as follows: Multi-band analysis of the individual interface electrical response spectrum was performed using Daubechies fourth-order wavelet packet decomposition to obtain the ohmic response component in the first band, the interface contact response component in the second band, and the polarization recovery response component in the third band. The thermal dynamic parameters of the individual temperature sequence and the equalization branch current sequence are identified using an autoregressive exogenous input structure to extract the thermal response components; the pressure response components of the stacked pressure sequence and the individual voltage excitation response components are extracted using the Preisach hysteresis operator identification algorithm. The ohmic response component of the first frequency band, the interface contact response component of the second frequency band, the polarization recovery response component of the third frequency band, the thermal response component, and the pressure response component are spliced ​​together according to the location number of the individual cells to obtain the individual cell interface response tensor. The monomer interface response tensor characterizes the electrical, thermal, and pressure responses of the solid electrolyte-electrode interface under equal branch spectrum excitation.

6. The high-energy-density battery equalization control method based on solid-state electrolyte according to claim 1, characterized in that: The specific process of constructing a solid-state interface composite state-space model based on the single interface response tensor and generating the set of interface-bearing feature quantities is as follows: A solid-state interface composite state-space model is constructed based on an electrochemical response layer, a solid-state interface contact layer, a thermal response layer, a pressure hysteresis layer, and an observation mapping layer. The individual cell interface response tensor, the individual cell terminal voltage sequence, the total current sequence of the battery pack, the individual cell temperature sequence, the stacking pressure sequence, and the number of cycles are input into the solid-state interface composite state-space model. The solid-state interface composite state-space model consists of the individual cell state vector with reversible contact loss, irreversible contact loss, interface polarization memory, thermal sensitivity, and pressure hysteresis. The solid-state interface composite state-space model is recursively corrected online using unscented Kalman filtering, outputting a set of interface bearing characteristics for each individual component. These characteristics include: reversible contact loss, irreversible contact loss, interface polarization memory, thermal sensitivity, and pressure hysteresis. The second-band interface contact response component participates in the partial mapping correction of reversible and irreversible contact losses; the third-band polarization recovery response component participates in the correction of interface polarization memory; the thermal response component participates in the correction of thermal sensitivity; the pressure response component and the stacked pressure sequence participate in the correction of pressure hysteresis; and the number of cycles participates in the cumulative update of irreversible contact loss. The state observation equation is formed by the formation mechanism of the single-cell terminal voltage and the mapping relationship of the response components in the single-cell interface response tensor. Based on the state observation equation of the solid-state interface composite state-space model, the source decomposition of the single-cell terminal voltage sequence is performed to obtain the energy effective voltage component, ohmic voltage drop component, interface contact voltage drop component, polarization memory voltage component, and thermo-coupling voltage offset component. The single-cell energy offset is obtained by analyzing the energy effective voltage component and the total current sequence of the battery pack through differential incremental capacity curve.

7. The high-energy-density battery equalization control method based on solid-state electrolyte according to claim 1, characterized in that: The specific process of forming a single interface acceptance domain based on the set of interface acceptance feature quantities is as follows: The reversible contact loss, irreversible contact loss, interface polarization memory, thermal sensitivity, and pressure hysteresis are obtained by interval Taylor reachable set propagation algorithm to obtain the single-unit interface bearing domain. The single-unit interface bearing domain includes the allowable equalization current amplitude range, the allowable equalization duration range, the allowable equalization energy direction, the allowable polarization residual boundary, the allowable heat load boundary, and the allowable pressure disturbance boundary.

8. The high-energy-density battery equalization control method based on solid-state electrolyte according to claim 1, characterized in that: The specific process of constructing a high-energy-density solid-state electrolyte battery pack equilibrium energy topology table based on solid-state battery equilibrium correlation data and single-cell interface domain is as follows: The individual cell location numbers are constructed as equalization energy nodes according to the series connection relationship of the battery pack, and the equalization branch numbers are constructed as equalization energy edges according to the connection relationship of the equalization branches. The individual cell energy offset, individual cell interface domain, equalization branch current sequence, equalization branch voltage sequence and equalization switch duty cycle sequence are written into the equalization energy topology table to obtain the equalization energy topology table of high energy density solid electrolyte battery pack with interface constraints. Based on the equalization energy topology table of high energy density solid electrolyte battery pack with interface constraints, a candidate equalization action set is established according to the energy offset direction, equalization branch connectivity, allowable equalization current amplitude range, allowable equalization duration range, and allowable equalization energy direction. The candidate equalization action set includes energy output cell, energy receiving cell, equalization branch number, equalization current waveform, equalization duty cycle waveform, equalization duration, and equalization recovery interval.

9. The high-energy-density battery equalization control method based on solid-state electrolyte according to claim 1, characterized in that: The specific process of generating equalization branch control commands based on the control barrier function and driving equalization branch switching transistors and controlling equalization current is as follows: A control barrier function is established by normalizing the residual margin between the allowable boundary in the individual interface domain and the actual effect of the candidate equalization action. The set of candidate equalization actions is input into the control barrier function for boundary verification. The control barrier function imposes boundary constraints on the allowable equalization current amplitude range, allowable equalization duration range, allowable polarization residual boundary, allowable heat load boundary, and allowable pressure disturbance boundary. When a candidate equalization action makes the control barrier function value less than zero, the corresponding candidate equalization action is excluded. When a candidate equalization action makes the control barrier function value greater than or equal to zero and less than the barrier holding threshold, a staggered intermittent equalization waveform is generated. When a candidate equalization action makes the control barrier function value greater than or equal to the barrier holding threshold, a continuous equalization waveform is generated. By using a mixed integer rolling optimization algorithm, the equalization energy path and equalization current waveform are jointly solved in the candidate equalization action set to obtain the target equalization path, target equalization current waveform, target equalization duty cycle waveform and target equalization recovery interval. Based on the target equalization path, target equalization current waveform, target equalization duty cycle waveform, and target equalization recovery interval, a finite state machine control method and a pulse width modulation method are used to generate equalization branch control commands. The equalization branch control commands are then sent to the corresponding equalization branches to drive the equalization branch switching transistors and control the equalization current. The switching transistors of the corresponding equalization branches are controlled to turn on and off according to the target equalization duty cycle waveform. The equalization current between the energy output unit and the energy receiving unit is adjusted according to the target equalization path and the target equalization current waveform. During the equalization branch switching transistor driving and equalization current control process, the control barrier function value is updated in real time according to the unit terminal voltage sequence, unit temperature sequence, and stacking pressure sequence. If the updated control barrier function value is less than zero, re-execute the mixed integer rolling optimization algorithm and update the target equilibrium path.

10. The high-energy-density battery equalization control method based on solid-state electrolyte according to claim 1, characterized in that: The specific process of writing the battery pack's historical operating records and correcting the model parameters based on these records is as follows: After equalization is completed, a battery pack historical operation file is created, which includes the individual cell interface response tensor, reversible contact loss, irreversible contact loss, interface polarization memory, thermal sensitivity, pressure hysteresis, individual cell interface bearing domain, control barrier function trigger record, target equalization path, target equalization current waveform, target equalization duty cycle waveform, target equalization recovery interval, and equalization branch switch drive and equalization current control logs. Based on the battery pack historical operation file, the model correction parameters of the solid-state interface composite state-space model are updated using the expectation-maximization algorithm.

Images