Charging current adaptive curve optimization method for fast charging scene

Abstract

The invention relates to a charging current adaptive curve optimization method for a fast charging scene. The method comprises the following steps: acquiring charging current data, monomer voltage data, module voltage data and charger capability data about a battery pack operation process in a fast charging scene; according to the charger capability data, performing dual-view waveform analysis on the charging current data to obtain charging current waveform data; performing spectral response analysis on a multi-frequency detection perturbation component in the charging current waveform data according to the monomer voltage data and the module voltage data to obtain impedance related characteristic data; according to the impedance related characteristic data, performing budget distribution analysis on the lithium precipitation risk and the thermal risk in the fast charging scene to obtain charging risk budget data; and according to the charging risk budget data, performing rolling constraint optimization analysis on the charging current waveform data of the next time domain to obtain charging current adaptive curve data. By adopting the method, the stability and consistency of the charging process can be improved.

Description

Technical Field

[0001] This application relates to the field of intelligent charging technology, and in particular to a method, apparatus and computer device for optimizing the adaptive curve of charging current for fast charging scenarios. Background Technology

[0002] In traditional technologies, the method for adapting charging current in fast charging scenarios typically employs a combination of "protocol negotiation + closed-loop regulation + multi-constraint current limiting". This means that the charger and the terminal (or BMS) first exchange capability sets via fast charging protocols (such as USBPD / PPS and various proprietary protocols) to negotiate voltage / current levels or continuous PPS voltage adjustment. Then, during charging, dynamic current limiting and power allocation are performed based on real-time sampling of battery voltage, current, temperature, SOC, internal resistance estimation, port and cable voltage drops (including cable identification / impedance detection), and charger power margin. Common control strategies include CC-CV and segmented constant current (stepcharging) / pulse or ramp current, temperature-zoned derating, and automatic current reduction when approaching cutoff voltage or high SOC. Overshoot and overheating are suppressed through the charging chip / BMS feedback loop (such as current loop + voltage loop, PID / model prediction, etc.), triggering protection (over-temperature, over-voltage, over-current, connector malfunction) when necessary. However, traditional technologies do not provide adaptive functions for charging current in fast charging scenarios, resulting in insufficient stability and consistency in the charging process during fast charging. Summary of the Invention

[0003] Therefore, it is necessary to provide a method, apparatus, and computer device for optimizing the adaptive charging current curve for fast charging scenarios, which can improve the stability and consistency of the charging process, in order to address the above-mentioned technical problems.

[0004] Firstly, this application provides a method for optimizing the adaptive charging current curve for fast charging scenarios, including: Acquire charging current data, individual cell voltage data, module voltage data, and charger capability data related to the battery pack's operation in fast charging scenarios; Based on the charger capability data, a dual-view waveform analysis is performed on the charging current data to obtain the charging current waveform data. Based on the individual unit voltage data and the module voltage data, spectral response analysis is performed on the multi-frequency detection perturbation component in the charging current waveform data to obtain impedance-related characteristic data. Based on the impedance-related characteristic data, a budget allocation analysis is performed on the lithium plating risk and thermal risk in the fast charging scenario to obtain charging risk budget data. Based on the charging risk budget data, rolling constraint optimization analysis is performed on the charging current waveform data in the next time domain to obtain the charging current adaptive curve data.

[0005] Secondly, this application also provides a charging current adaptive curve optimization device for fast charging scenarios, including: The scenario data acquisition module is used to acquire charging current data, single cell voltage data, module voltage data, and charger capability data related to the battery pack operation process in fast charging scenarios. The dual-view waveform analysis module is used to perform dual-view waveform analysis on the charging current data based on the charger capability data to obtain charging current waveform data. The spectral response analysis module is used to perform spectral response analysis on the multi-frequency detection perturbation component in the charging current waveform data based on the individual cell voltage data and the module voltage data, and to obtain impedance-related characteristic data. The budget allocation analysis module is used to perform budget allocation analysis on the lithium plating risk and thermal risk in the fast charging scenario based on the impedance-related characteristic data, and obtain charging risk budget data. The rolling constraint optimization analysis module is used to perform rolling constraint optimization analysis on the charging current waveform data in the next time domain based on the charging risk budget data, so as to obtain the charging current adaptive curve data.

[0006] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement any step of a charging current adaptive curve optimization method for fast charging scenarios.

[0007] The aforementioned method, apparatus, and computer equipment for adaptive charging current curve optimization in fast charging scenarios simultaneously acquire charging current, individual cell / module voltage, and charger capability data during fast charging. Under charger capability constraints, the charging current is constructed into a dual-purpose waveform of "energy transfer base current + multi-frequency detection perturbation," enabling the system to obtain spectral response information usable for electrochemical state identification without significantly sacrificing charging efficiency. Furthermore, by utilizing individual cell and module voltage data, spectral response analysis is performed on the multi-frequency detection perturbation component to extract impedance-related characteristic data, achieving online characterization of limiting factors such as ohmic resistance, interfacial reactions, and diffusion polarization, thereby providing a basis for analysis. The budget allocation for lithium risk and thermal risk provides a real-time, quantifiable basis. Based on this, the risk budget is embedded as an explicit constraint into the rolling constraint optimization of the charging current waveform in the next time domain. This allows the adaptive charging current curve to be continuously adjusted according to changes in battery state, temperature, and charger dynamic capabilities. This avoids the overly conservative or local runaway problems caused by traditional coarse-grained current limiting strategies based on fixed thresholds or static lookup tables. Ultimately, while meeting the upper limits of single-cell voltage, temperature, and charger capabilities, it reduces the risk of lithium plating and overheating, improves the stability and consistency of the charging process, shortens the charging time to reach the target state of charge, and improves the availability and safety margin of fast charging throughout the entire lifespan. Attached Figure Description

[0008] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 This is an application environment diagram of a charging current adaptive curve optimization method for fast charging scenarios in one embodiment; Figure 2 This is a flowchart illustrating a charging current adaptive curve optimization method for fast charging scenarios in one embodiment. Figure 3 This is a structural block diagram of a charging current adaptive curve optimization device for fast charging scenarios in one embodiment. Figure 4 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0010] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0011] This application provides a charging current adaptive curve optimization method for fast charging scenarios, which can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104, or it can be located in the cloud or on other network servers. Server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0012] In one exemplary embodiment, such as Figure 2 As shown, a method for adaptive charging current curve optimization for fast charging scenarios is provided, and this method is applied to... Figure 1 Taking the server in the example, the explanation includes the following steps 202 to 210. Wherein:

[0013] Step 202: Obtain charging current data, single cell voltage data, module voltage data, and charger capability data related to the battery pack operation process in the fast charging scenario.

[0014] Step 204: Based on the charger capability data, perform dual-view waveform analysis on the charging current data to obtain the charging current waveform data.

[0015] Step 206: Based on the individual cell voltage data and module voltage data, perform spectral response analysis on the multi-frequency detection perturbation component in the charging current waveform data to obtain impedance-related characteristic data.

[0016] Step 208: Based on impedance-related characteristic data, perform budget allocation analysis on lithium plating risk and thermal risk in fast charging scenarios to obtain charging risk budget data.

[0017] Step 210: Based on the charging risk budget data, perform rolling constraint optimization analysis on the charging current waveform data in the next time domain to obtain the charging current adaptive curve data.

[0018] Fast charging scenarios refer to charging the battery pack with a current / power higher than the conventional charging rate under a limited charging time target, and are subject to constraints such as voltage, temperature, and charger capacity.

[0019] Among them, the battery pack operation process is the evolution of current, voltage, temperature and related states of the battery pack over time during charging.

[0020] Among them, the charging current data is sampled data that characterizes the magnitude of the current flowing into the battery pack during the charging process and its change over time.

[0021] Among them, the individual cell voltage data is the sampled data of the voltage change of each individual cell in the battery pack over time.

[0022] Among them, the module voltage data is the sampled data of the module terminal voltage or the module aggregate voltage as a function of time, which consists of multiple individual units.

[0023] Among them, the charger capability data refers to the upper limit of voltage / current / power and dynamic limit that the charger can provide under the current operating conditions.

[0024] Among them, dual-view waveform analysis is an analysis and processing process that constructs the charging current as "energy transmission base current + information detection perturbation" and determines its parameters under the premise of meeting the charger's capacity constraints.

[0025] Among them, the charging current waveform data is the waveform parameter or timing data of the charging current changing over time, obtained after dual-view waveform analysis, and used for actual execution.

[0026] Among them, the multi-frequency detection perturbation component is a current perturbation component superimposed on the base current, containing multiple frequency points and with a small amplitude, used to excite the battery to generate observable spectral response information.

[0027] Among them, spectral response analysis is an analytical processing process that uses voltage observation data to extract amplitude and phase responses at various perturbation frequency points and form frequency domain response results.

[0028] Among them, impedance-related characteristic data are characteristic quantities obtained from spectral response analysis that characterize the battery's ohmic internal resistance, interface charge transfer, and diffusion polarization.

[0029] Among them, lithium plating risk is the risk level corresponding to the possibility or trend of lithium metal deposition on the negative electrode during fast charging.

[0030] Among them, thermal risk refers to the risk level of exceeding temperature limits, increased probability of thermal runaway triggering, or the appearance of local hot spots due to heat generation during fast charging.

[0031] Budget allocation analysis is an analytical process that, within a given risk constraint framework, allows for the allocation and dynamic adjustment of risk margins between lithium plating risk and thermal risk.

[0032] Among them, the charging risk budget data is the allowable risk amount, margin or upper limit constraint data corresponding to lithium plating risk and thermal risk, respectively, output by the budget allocation analysis.

[0033] Among them, rolling constraint optimization analysis is an analysis process that iteratively solves for the next time-domain optimal charging current waveform under constraints such as voltage, temperature, charger capability, and risk budget, guided by the objective function within the rolling time domain.

[0034] Among them, the adaptive charging current curve data is a curve data of charging current changing with time or SOC that is continuously updated through rolling constraint optimization and adaptively adjusted according to changes in battery state and risk budget.

[0035] Specifically, during fast charging, the BMS synchronously collects charging current data, individual cell voltage data, and module voltage data of the battery pack at a preset sampling period. It also obtains charger capability data through communication interfaces with the charger (such as vehicle-side charging protocols or BMS-charger communication links). This charger capability data includes at least the charger's maximum available current / power, dynamic derating limits, output voltage range, and cable voltage drop or input-side limitations. The data from each channel is then timestamped and missing points are filled to form a time-series dataset.

[0036] Based on the charger's capability data, the feasible current domain is determined (e.g., the current upper limit allowed at the current moment, the rise rate limit, and the ripple / EMI limit). Then, within the feasible current domain, waveform data containing the energy transfer base current component and the multi-frequency detection perturbation component is constructed from the charging current data. The base current component is used to maximize the energy charged per unit time, while the perturbation component is used to inject a identifiable signal into the system without significantly affecting the charging efficiency. The frequency, amplitude, and phase parameters of the perturbation component are configured according to the feasible current domain and the power spectral budget, ultimately outputting charging current waveform data that meets the charger's capability constraints.

[0037] Using individual unit voltage data and module voltage data as observations, a spectral domain correspondence between "perturbation excitation and voltage response" is established for the known multi-frequency probe perturbation component in the charging current waveform data (whose frequency, amplitude, and phase parameters are determined by the charging current waveform data). Response components that are in phase and at the same frequency as the multi-frequency probe perturbation component are extracted from the individual unit voltage data and module voltage data. Then, synchronous correlation / phase-locked demodulation operations based on each frequency point are performed on the response components in the individual unit voltage data and module voltage data that are in phase and at the same frequency as the multi-frequency probe perturbation component. The amplitude and phase of the voltage response at each frequency point are projected to obtain the complex spectral response (complex response spectrum data) for the multi-frequency probe perturbation component. Finally, based on the complex response spectrum data and under the consistency constraint between the individual unit scale and the module scale, impedance-related feature data is extracted, allowing the impedance-related feature data to characterize the spectral response features corresponding to ohmic internal resistance, interface charge transfer, and / or diffusion polarization.

[0038] Temperature-based mapping is performed on impedance-related characteristics to obtain temperature-dependent decoupled impedance characteristic data, thereby reducing the interference of characteristic drift on risk assessment under different temperature conditions. Then, based on the temperature-dependent decoupled impedance characteristic data, lithium plating margin and thermal margin are calculated to form risk-dominant sequence data (used to characterize the dominant risk type and its priority at the current stage). Cross-risk dynamic reallocation is performed around the risk-dominant sequence, prioritizing the allocation of basic budgets to dominant risks. Then, constraint propagation and closure correction are performed on the budgets of non-dominant risks based on the coupling relationship and margin change trend between the two types of risks, outputting charging risk budget data (including at least one or both of the lithium plating risk budget and thermal risk budget).

[0039] Based on charging risk budget data, an optimization problem is constructed in the rolling time domain to maximize effective charging capacity and / or minimize the estimated time to the target SOC. This problem incorporates upper limits on individual cell voltage, module / individual cell temperature, charger capability constraints, and charging risk budget constraints. The solution yields the base current component and multi-frequency probe perturbation component in the next time domain (or directly obtains the complete waveform in the next time domain). Safety clipping is applied before execution to ensure that hard constraints are not violated. Through periodic rolling updates (data is re-acquired, features and budget are updated, and output is re-optimized each control cycle), adaptive charging current curve data that adapts to changes in battery state and risk budget are finally obtained.

[0040] The aforementioned adaptive charging current curve optimization method for fast charging scenarios simultaneously acquires charging current, individual cell / module voltage, and charger capability data during fast charging. Under charger capability constraints, the charging current is constructed into a dual-purpose waveform of "energy transfer base current + multi-frequency detection perturbation," enabling the system to obtain spectral response information usable for electrochemical state identification without significantly sacrificing charging efficiency. Furthermore, the individual cell and module voltage data are used to perform spectral response analysis on the multi-frequency detection perturbation component to extract impedance-related characteristic data, achieving online characterization of limiting factors such as ohmic resistance, interface reactions, and diffusion polarization. This provides a basis for assessing lithium plating risk and... The allocation of thermal risk budget provides a real-time, quantifiable basis. Based on this, the risk budget is embedded as an explicit constraint into the rolling constraint optimization of the charging current waveform in the next time domain. This allows the adaptive charging current curve to be continuously adjusted according to changes in battery status, temperature, and charger dynamic capabilities. This avoids the overly conservative or local runaway problems caused by traditional coarse-grained current limiting strategies based on fixed thresholds or static lookup tables. Ultimately, while meeting the upper limits of single-cell voltage, temperature, and charger capabilities, it reduces the risk of lithium plating and overheating, improves the stability and consistency of the charging process, shortens the charging time to reach the target state of charge, and improves the availability and safety margin of fast charging throughout the entire lifespan.

[0041] In an exemplary embodiment, based on charger capability data, dual-view waveform analysis is performed on the charging current data to obtain charging current waveform data, including steps 302 to 308. Wherein:

[0042] Step 302: Based on the charger capability data, perform a capability mapping analysis between the charger output capability and the upper limit of the battery charging current in the fast charging scenario to obtain the current feasible domain data.

[0043] Step 304: Based on the current feasible domain data, perform information gain configuration analysis on the multi-frequency detection perturbation component in the charging current data to obtain the information-optimal multi-frequency detection perturbation component.

[0044] Step 306: Based on the optimal multi-frequency detection perturbation component, perform basic current allocation processing on the residual current margin in the current feasible domain data to obtain the energy transfer basic current component data.

[0045] Step 308: The energy transfer base current component data is superimposed with the information optimal multi-frequency detection perturbation component to obtain the charging current waveform data.

[0046] Among them, the charger output capability is the upper limit of the output voltage, output current and output power that the charger can continuously or instantaneously provide under the current operating conditions, as well as its dynamic derating characteristics.

[0047] Among them, the upper limit of battery charging current is the maximum allowable charging current limit determined by the safety boundaries of the battery pack (such as the upper limit of single cell voltage, the upper limit of temperature, the SOC range limit and lifespan constraints, etc.).

[0048] Capability mapping analysis is an analytical process that involves constraining and converting the charger's output capability with the upper limit of the battery's charging current, and then merging their intersections to obtain the range of executable current on the battery side.

[0049] Among them, the current feasible domain data is the set of charging current upper and lower bounds, slope constraints, and envelope data that satisfy the charger capability and battery safety constraints at a given time or in the time domain.

[0050] Information gain configuration analysis is a parameter configuration process that, under the constraint of the current feasible domain, selects frequency points and assigns amplitude and phase to multi-frequency probe perturbation components to maximize the amount of observable information or the degree of recognition.

[0051] Among them, the information-optimal multi-frequency detection perturbation component is the multi-frequency current perturbation signal component that is most favorable for extracting the battery spectrum response under constraints, obtained after information gain configuration.

[0052] The residual current margin is the current margin that can still be used for base current energy transmission after deducting the current space occupied by the multi-frequency detection perturbation component (including the retention of peak and ripple constraints) within the current feasible domain.

[0053] Among them, the base current allocation process is the calculation process of determining the base current amplitude and its time-varying trajectory under the conditions of residual current margin and dynamic constraints, so as to realize the allocation of energy transfer target.

[0054] Among them, the energy transfer base current component data is the base current command or its parameterized timing data output by the base current allocation processing, which is used to undertake the main charging energy input.

[0055] Specifically, based on the charger's capability data (including maximum output current / power, output voltage range, dynamic derating limit, allowable current change rate, and cable voltage drop or input-side constraint information), the output capability that the charger can provide at the current moment is mapped to the upper limit of the charging current allowed by the battery side. Then, the upper limit of the charging current on the battery side (such as the current limiting boundary formed by the upper limit of the single cell voltage, the upper limit of the temperature, or the upper limit of the BMS allowable current) is combined to perform an intersection operation to form the current feasible domain data that changes over time. The current feasible domain data can be used as the upper and lower limit curves of the current, the allowable current envelope, and the set of current slope constraints.

[0056] Based on the current feasible domain data, the ripple capacity or allocable current margin in the feasible domain is converted into the power spectral budget of the multi-frequency probe perturbation component (i.e., the upper limit of the allowable amplitude or the allowable energy share of each candidate frequency band). Then, under the constraint of the power spectral budget, a structured complementary sensitivity arrangement is performed on the candidate perturbation frequency set. Frequency combination with complementary sensitivity to target characteristics such as ohmic internal resistance, interface charge transfer and diffusion polarization is selected first. Furthermore, information gain configuration is implemented on the amplitude and phase parameters of the selected frequency points (for example, based on the criteria of maximizing parameter identifiability, improving the spectral response signal-to-noise ratio and suppressing inter-frequency crosstalk, it is achieved through hierarchical amplitude allocation and phase coding / orthogonal phase design) to obtain the information-optimal multi-frequency probe perturbation component.

[0057] After determining the optimal multi-frequency detection perturbation component, the current margin occupied by this perturbation component in the time domain (including peak occupancy, mean square occupancy, and the retention margin introduced by ripple constraints) is calculated. Based on this, the corresponding occupancy is subtracted from the current feasible domain data to obtain the remaining current margin. Base current allocation processing is performed on the remaining current margin. That is, under the premise of satisfying the upper limit of current, slope limit, and charger dynamic capability constraints, the amplitude and variation trajectory of the energy transfer base current component are determined so that the base current component is as close as possible to the upper boundary of the remaining margin to increase the energy charged per unit time. The base current can be smoothed and slope limited according to the SOC range or individual unit constraint changes, and the energy transfer base current component data is output.

[0058] The energy transfer base current component data and the information-optimal multi-frequency detection perturbation component data are superimposed and synthesized in the time domain to form preliminary charging current waveform data. A consistency check is then performed on the synthesized waveform to ensure that it meets the current feasible domain data constraints (including upper limit of current amplitude, slope constraint, and ripple limit) at any given time, resulting in the final charging current waveform data. The charging current waveform data can be represented as a discrete-time current command or a set of parameterized waveforms.

[0059] In this embodiment, the charger output capability and the upper limit of the battery charging current are unified into the same constraint space and a current feasible region is formed by capability mapping analysis. This avoids overcurrent, undercurrent, or frequent current limiting jitter caused by inconsistencies between the charger's dynamic limit or the battery's safety boundary from the source. Within the current feasible region, information gain is configured to obtain the optimal multi-frequency detection perturbation component, so that the charging current carries observable spectral excitation information while undertaking energy transmission, thereby improving the reliability of subsequent state identification and risk assessment. Then, based on the "remaining current margin", the base current is allocated so that the energy transmission base current component is as close as possible to the upper boundary of the feasible region without crowding out the margin required for detection perturbation. Finally, the charging current waveform formed by superposition balances charging speed and observability while meeting capability and safety constraints, reducing the conservatism of the strategy and improving the stability and robustness of the fast charging process.

[0060] In an exemplary embodiment, based on current feasible domain data, information gain configuration analysis is performed on the multi-frequency detection perturbation component in the charging current data to obtain the information-optimal multi-frequency detection perturbation component, including steps 402 to 406. Wherein:

[0061] Step 402: Perform amplitude limiting and ripple constraint mapping on the current feasible domain data to obtain the power spectrum budget data of the multi-frequency detection perturbation component.

[0062] Step 404: Based on the power spectrum budget data of the multi-frequency detection perturbation components, perform structured complementary sensitivity arrangement analysis on the candidate perturbation frequency points in the charging current data to obtain multi-frequency complementary frequency point set data.

[0063] Step 406: Perform coded information gain allocation on the amplitude and phase parameters in the multi-frequency complementary frequency point set data to obtain the information-optimal multi-frequency detection perturbation component.

[0064] Among them, the amplitude limiting and ripple constraint mapping is the process of converting the upper limit of current, slope limit and allowable ripple boundary in the current feasible domain into frequency domain constraints on the perturbation amplitude or energy that can be assigned to each candidate frequency band / frequency point.

[0065] Among them, the multi-frequency detection perturbation component power spectrum budget data is a set of budget data obtained after amplitude limiting and ripple constraint mapping, which is the allowable perturbation energy share, amplitude limit or power limit for each candidate frequency band / frequency point.

[0066] Among them, the structured complementary sensitivity arrangement analysis is a process of evaluating the sensitivity of candidate frequency points to different electrochemical mechanism characteristics under the power spectrum budget constraint and selecting and grouping frequency points according to the complementary coverage principle.

[0067] Among them, the multi-frequency complementary frequency point set data is frequency point set data obtained after complementary sensitivity arrangement, which is divided into at least two subsets according to complementary relationship and forms complementary coverage of target features.

[0068] Among them, the coded information gain allocation is a parameter configuration process that allocates the amplitude and phase of the selected frequency points in a hierarchical manner under the power spectrum budget constraint and introduces phase orthogonal or pseudo-random coding in order to improve the amount of observable information and suppress inter-frequency crosstalk.

[0069] Specifically, based on the given upper limit of current, slope constraints, and allowable ripple / voltage ripple boundaries in the feasible current domain data, the "time-domain available perturbation current space" is transformed into a "frequency-domain allocable perturbation energy space." This involves calculating, for each candidate frequency band, the upper limit of perturbation amplitude, upper limit of mean square amplitude, or upper limit of energy share allowed for superposition without exceeding the feasible domain boundaries and ripple limits, thus forming multi-frequency probe perturbation component power spectrum budget data. The power spectrum budget data can be stored segmented by frequency band (e.g., low / medium / high frequency bands) or stored point-by-point by candidate frequency point to provide hard constraint boundaries for subsequent frequency point selection and amplitude / phase allocation.

[0070] In the candidate frequency set, frequency points that meet the feasibility requirements for multi-frequency detection of perturbation component power spectral budget data are prioritized. A "sensitivity fingerprint" is established for each frequency point, characterizing its sensitivity and discriminative ability to target mechanistic features (e.g., ohmic resistance correlation, interface charge transfer correlation, diffusion polarization correlation). Using "complementary coverage" as the arrangement criterion, frequency point combinations with complementary sensitivities on different mechanistic features and low redundancy on the same mechanistic feature are selected. The selected frequency points are then divided into at least two frequency subsets (e.g., ohmic / interface / diffusion subsets or low / medium / high frequency subsets) according to their complementary relationships, thereby outputting multi-frequency complementary frequency set data.

[0071] Under the power spectral budget constraint, the amplitude allocation weights of each frequency point or subset of frequency points in the multi-frequency complementary frequency point set data are first determined, so that the frequency points or subsets that contribute more to the observable information receive a higher amplitude share, while limiting the total perturbation energy and peak occupancy to avoid crowding out the base current margin. The phase parameters of each frequency point are encoded using a design (e.g., orthogonal phase sets or pseudo-random phase codebooks) to improve the separability of different frequency points during demodulation and suppress inter-frequency crosstalk. The amplitude allocation results and phase encoding results are assembled into a multi-frequency probed perturbation component parameter set containing frequency point parameters, amplitude parameters, and phase parameters, thereby obtaining the information-optimal multi-frequency probed perturbation component.

[0072] In this embodiment, an executable power spectral budget is formed by limiting and mapping the current feasible domain data to obtain controllable and verifiable frequency domain resource constraints for the multi-frequency detection perturbation components within the peak occupancy, ripple level, and charger capability boundaries. Under this power spectral budget constraint, the candidate perturbation frequencies are further structured and complementary in sensitivity arrangement, so that the selected frequencies complement each other to achieve complementary coverage of target mechanisms such as ohmic internal resistance, interface charge transfer, and diffusion polarization, and reduce frequency redundancy, thereby obtaining a higher amount of identifiable information with less perturbation resources. Finally, the amplitude and phase of the selected frequencies are allocated using coded information gain. Through hierarchical amplitude allocation and phase coding, the frequency separability is improved and inter-frequency crosstalk is suppressed, so that the obtained optimal multi-frequency detection perturbation components improve the signal-to-noise ratio and robustness of spectral response extraction without significantly sacrificing charging efficiency.

[0073] In an exemplary embodiment, based on the power spectrum budget data of the multi-frequency detected perturbation components, a structured complementary sensitivity arrangement analysis is performed on the candidate perturbation frequency points in the charging current data to obtain a set of multi-frequency complementary frequency points, including steps 502 to 508. Wherein:

[0074] Step 502: Based on the power spectrum budget data of the multi-frequency detection perturbation component, pre-screen the candidate perturbation frequency points in the charging current data to obtain candidate frequency point pool data.

[0075] Step 504: Based on the candidate frequency pool data, perform target feature decoupling analysis on each candidate perturbation frequency to obtain frequency decoupling score data.

[0076] Step 506: Based on the frequency decoupling score data, perform complementary coverage arrangement analysis on the candidate frequency pool data to obtain complementary frequency skeleton set data.

[0077] Step 508: Based on the complementary frequency skeleton set data, perform cross-frequency segmentation mapping on the power spectrum budget data to obtain multi-frequency complementary frequency set data.

[0078] Frequency pre-screening is a process of screening candidate perturbation frequencies based on power spectrum budget and feasibility constraints, eliminating unfeasible frequencies and retaining usable frequencies.

[0079] Among them, the candidate frequency pool data is a set of available perturbation frequency points obtained after frequency point pre-screening, along with their corresponding budget upper limit, allowable amplitude range, or bandwidth conditions.

[0080] Among them, the target feature decoupling analysis is an analytical process that evaluates the separability and aliasing degree of target features such as ohmic internal resistance, interface charge transfer and diffusion polarization at each candidate frequency point.

[0081] Among them, the frequency point decoupling score data is a score or weight data output from the target feature decoupling analysis, used to quantify the separability of each frequency point feature.

[0082] Among them, complementary coverage orchestration analysis is an orchestration process based on decoupling score selection and combination of frequency points, so that different frequency points form complementary coverage for different target features and suppress redundancy.

[0083] Among them, the complementary frequency skeleton set data is the minimum sufficient frequency combination obtained by complementary coverage arrangement or the skeleton frequency subset data divided according to the complementary relationship.

[0084] Cross-frequency segmentation mapping is a mapping process that allocates the power spectrum budget to the skeleton frequency points and their subsets according to frequency bands or frequency point upper limits to form the final multi-frequency point set.

[0085] Specifically, each candidate perturbation frequency point is matched with the power spectrum budget data of the multi-frequency detection perturbation components for feasibility. Frequency points that are insufficient to generate detectable perturbation amplitudes in the corresponding frequency band, cannot be superimposed under ripple / EMI constraints, or cannot be stably generated and demodulated under the sampling frequency and control bandwidth conditions are eliminated. Frequency points that meet the budget constraints and are feasible are retained as candidate frequency point pool data. The candidate frequency point pool data can simultaneously record the budget upper limit, allowable amplitude range, and available bandwidth information for each frequency point.

[0086] Based on the candidate frequency pool data, a "sensitivity vector" or "contribution vector" to the target features is constructed for each candidate perturbation frequency. The target features include at least ohmic internal resistance related features, interface charge transfer related features, and diffusion polarization related features. The decoupling score is calculated based on the discriminative power of the sensitivity vector on different target feature dimensions (e.g., the concentration of the target feature contribution, the degree of cross-correlation / aliasing between frequency points, or the feature separability index is used as the scoring basis), thereby forming frequency decoupling score data.

[0087] With the goal of "covering all target features and minimizing redundancy", frequency points with high decoupling scores and outstanding sensitivity in a certain target feature dimension are selected as the starting point of the skeleton. Then, while ensuring that the covered target features do not degrade, frequency points that can enhance the uncovered or weakly covered target features are added iteratively. Frequency points that contribute highly to the same target feature are suppressed or replaced. Finally, complementary frequency point skeleton set data is obtained. The complementary frequency point skeleton set data is reflected as a set of minimal sufficient frequency point combinations or a skeleton subset divided according to complementary relationships.

[0088] The power spectrum budget data is allocated to the complementary frequency skeleton set data by frequency band or frequency point budget upper limit, forming a "skeleton frequency point - budget segment" correspondence. Within each budget segment, auxiliary frequencies that satisfy the budget surplus and do not disrupt the complementary structure are added (e.g., to improve demodulation robustness or enhance the signal-to-noise ratio of a certain frequency band), thereby outputting multi-frequency complementary frequency set data. The multi-frequency complementary frequency set data includes at least two frequency subsets divided according to their complementary sensitivity relationship.

[0089] In this embodiment, candidate perturbation frequencies are pre-screened based on the power spectrum budget of multi-frequency probe perturbation components. This eliminates frequencies that are not feasible or have insufficient budget during the configuration phase, avoiding subsequent demodulation failures and resource waste. Secondly, by performing target feature decoupling analysis on candidate frequencies and generating scores, frequencies with higher separability to target mechanisms such as ohmic internal resistance, interface charge transfer, and diffusion polarization can be prioritized, thereby reducing feature aliasing and inter-frequency redundancy. Furthermore, complementary coverage arrangement is used to obtain a complementary frequency skeleton set, enabling the selected frequency combination to form complementary coverage of key mechanisms and obtain higher information density with fewer frequencies. Finally, the power spectrum budget is segmented across frequencies and mapped to the skeleton set to generate a multi-frequency complementary frequency set. This allows for coordinated matching of frequency structure and budget resources while satisfying amplitude limiting and ripple constraints, thereby improving the executability of perturbation configuration and the robustness of spectral response extraction.

[0090] In an exemplary embodiment, based on the individual cell voltage data and the module voltage data, spectral response analysis is performed on the multi-frequency detection perturbation component in the charging current waveform data to obtain impedance-related characteristic data, including steps 602 to 606. Wherein:

[0091] Step 602: Perform noise reduction preprocessing on the individual unit voltage data and module voltage data to obtain voltage response data; Step 604: Based on the frequency parameters, amplitude parameters and phase parameters in the multi-frequency detection perturbation component, the voltage response data is phase-encoded and demodulated to obtain complex response spectrum data; Step 606: Based on the complex response spectrum data, perform cross-scale consistency constraint fusion on the individual unit voltage data and the module voltage data to obtain impedance-related characteristic data.

[0092] Among them, noise reduction preprocessing is a process of performing detrending, filtering, outlier handling and synchronization alignment on the voltage sampling sequences of individual units and modules to suppress noise and interference.

[0093] Among them, the voltage response data is the timing data of individual unit voltage and module voltage that retains the effective response components related to multi-frequency detection perturbations after noise reduction preprocessing.

[0094] Phase-coded demodulation is a demodulation process that uses the frequency, amplitude, and phase parameters of the perturbation components to perform synchronous correlation / phase-locked loop operations on the voltage response and introduces phase coding to improve the frequency separability.

[0095] Among them, the complex response spectrum data is a set of complex amplitude and phase results of the voltage response at each perturbation frequency point obtained by phase encoding demodulation, which is used to characterize the frequency domain response characteristics.

[0096] Among them, cross-scale consistency constraint fusion is a fusion process that verifies, weights, aggregates, or selects the weakest constraint on the complex response spectra of the individual scale and the module scale at the same frequency point using amplitude and phase consistency constraints to output more robust features.

[0097] Specifically, based on the sampling timestamp alignment, detrending and baseline component suppression (e.g., removing DC / low-frequency components that change slowly with SOC) are performed on the individual unit voltage data and module voltage data respectively. This is combined with bandpass or notch filtering to suppress noise and power frequency / switching ripple interference unrelated to the perturbation frequency. Simultaneously, outliers are removed or interpolated to improve timing continuity, thereby outputting voltage response data for subsequent demodulation. This voltage response data retains effective response components that match the frequency bands of the multi-frequency probed perturbation components and can be used to form voltage response sequences for individual unit channels and module channels respectively.

[0098] A reference quadrature demodulation base (in-phase / quadrature) is constructed for each frequency point using the frequency parameters of the multi-frequency probed perturbation components. The phase parameters of the multi-frequency probed perturbation components are then used as a codebook to perform phase encoding on the reference base, enabling the separability of different frequency points or subsets during demodulation. Subsequently, synchronous correlation or phase-locked loop (PLL) operations are performed on the voltage response data and the reference base to extract the in-phase and quadrature components at each frequency point and perform amplitude normalization (corresponding to the amplitude parameters), thereby obtaining the complex amplitude-phase response corresponding to each frequency point and collecting them to form complex response spectrum data.

[0099] Consistency checks are performed on individual cell voltage data and module voltage data using complex response spectrum data at the same frequency point. Amplitude consistency residuals and phase consistency residuals are constructed as constraints. Under the condition of satisfying the consistency constraints, the complex responses of individual cell channels are aggregated (e.g., weighted by consistency weights, or using the weakest cell / largest residual cell as constraint boundaries) to obtain a representative complex response spectrum of the battery pack. Subsequently, the frequency domain ratio of voltage to current or equivalent parameter fitting is calculated based on the representative complex response spectrum, and impedance-related characteristic data is output. The impedance-related characteristic data includes at least ohmic internal resistance related characteristics, interface charge transfer related characteristics, and / or diffusion polarization related characteristics, and cross-scale consistency indices can be output simultaneously to reflect the reliability of the characteristics.

[0100] In this embodiment, high signal-to-noise ratio voltage response data is obtained by performing noise-resistant preprocessing on the individual cell voltage data and module voltage data. This effectively suppresses interference introduced by switching ripple, sampling jitter, and slow SOC trend under fast charging conditions, thereby improving the stability of subsequent spectral analysis. Furthermore, phase encoding demodulation of the voltage response data is performed using the frequency, amplitude, and phase parameters of the multi-frequency probe perturbation component. This improves frequency point separability and suppresses inter-frequency crosstalk under multi-frequency parallel excitation conditions, thus obtaining more reliable complex response spectrum data. Finally, cross-scale consistency constraint fusion is performed on the individual cell voltage data and module voltage data based on the complex response spectrum data. This eliminates the amplification effect of individual cell measurement noise and local anomalies on feature extraction at the battery pack level and enhances the sensitivity to the "weakest cell" or consistency deviation, thereby obtaining more accurate and robust impedance-related feature data.

[0101] In an exemplary embodiment, the voltage response data is phase-encoded and demodulated based on the frequency parameters, amplitude parameters, and phase parameters in the multi-frequency probe perturbation component to obtain complex response spectrum data, including steps 702 to 708. Wherein:

[0102] Step 702: Baseline component removal and bandwidth constraint filtering are performed on the voltage response data to obtain narrowband response data.

[0103] Step 704: Based on the phase parameters, perform phase codebook matching correlation analysis on the narrowband response data to obtain code correlation output data.

[0104] Step 706: Based on the amplitude parameter, perform frequency point orthogonality correction on the code correlation output data to obtain orthogonal complex response data.

[0105] Step 708: Based on the frequency parameters, perform complex domain coherent accumulation and phase drift compensation on the orthogonal complex response data to obtain complex response spectrum data.

[0106] Baseline component removal is a process that removes the DC component and low-frequency trend terms caused by slow changes in SOC and polarization drift from the voltage response in order to highlight the AC response caused by perturbations.

[0107] Among them, bandwidth-constrained filtering is a process that applies bandpass / multi-bandpass filtering to the voltage response based on the candidate perturbation frequency and its allowable bandwidth to suppress noise and interference in non-target frequency bands.

[0108] Among them, the narrowband response data is voltage time series data that mainly contains effective response components near candidate perturbation frequency points after baseline component removal and bandwidth constraint filtering.

[0109] Among them, phase codebook matched correlation analysis is a demodulation analysis that uses a preset phase codebook to construct a reference sequence and performs matched correlation with the narrowband response to extract the in-phase / quadrature response at each frequency point.

[0110] Among them, the code-related output data is the correlation complex output or correlation peak sequence data corresponding to each candidate frequency point obtained from phase codebook matching correlation analysis.

[0111] Among them, frequency orthogonality correction is to decouple and compensate for the coupling and crosstalk between frequency points in the code correlation output, so that the responses of different frequency points are as orthogonal and independent as possible in mathematics.

[0112] Among them, the orthogonal complex response data are complex response data obtained after frequency orthogonality correction, in which crosstalk between frequencies is suppressed and amplitude and phase are comparable.

[0113] Among them, coherent accumulation in the complex domain is an accumulation process that superimposes the complex responses at the same frequency point in the complex domain in phase to improve the signal-to-noise ratio and enhance stability.

[0114] Phase drift compensation is a compensation process that estimates and corrects phase errors caused by sampling jitter, frequency shift, or reference phase change to ensure effective coherent accumulation.

[0115] Specifically, the DC / low-frequency baseline components introduced by slow changes in SOC, slow polarization drift, or sampling bias are removed from the voltage response data (e.g., by using moving average, low-order polynomial fitting, or high-pass detrending). Then, the filter bandwidth is set according to the candidate frequency set of the multi-frequency probe perturbation components. Bandpass filtering or multi-bandpass filtering matching the frequency set is applied to the voltage response data to suppress non-target frequency noise and switching ripple interference, resulting in narrowband response data that mainly contains effective response components near the candidate frequency points.

[0116] A phase codebook is constructed using the phase parameters corresponding to each candidate frequency point in the multi-frequency probe perturbation component. Based on this codebook, a reference sequence with the same frequency and a predetermined phase code is generated. The narrowband response data is then subjected to synchronous correlation or matched filtering operations with each reference sequence. The correlation peak or correlation complex component is output at each candidate frequency point, thus obtaining the code correlation output data. The code correlation output data reflects the in-phase / orthogonal projection result of the voltage response under the corresponding phase coding conditions. It can be used to distinguish the responses of different frequency points or different subsets, thereby suppressing inter-frequency crosstalk.

[0117] The amplitude of the code correlation output at each frequency point is normalized using the amplitude parameter to make the demodulated outputs at different frequencies comparable. The coupling relationship between frequency points is constructed based on the cross-correlation term or the estimated crosstalk term between the code correlation outputs. Orthogonalization correction is performed on the code correlation output (e.g., through decoupling matrix compensation, Gram-Schmidt orthogonalization, or least squares decoupling) to obtain the orthogonal complex response data after crosstalk suppression between frequency points.

[0118] The accumulation window length and accumulation step are determined by the parameters of each frequency point. Coherent accumulation of the orthogonal complex response data is performed in the complex domain to improve the effective signal-to-noise ratio. At the same time, the possible sampling clock deviation, frequency offset or reference phase drift is estimated by the frequency point parameters. The phase is compensated and corrected during the coherent accumulation process to keep the accumulation process in phase superposition rather than mutual cancellation. Finally, the stable complex amplitude and phase results at each candidate frequency point are output to form complex response spectrum data.

[0119] In this embodiment, by removing the baseline component and performing bandwidth-constrained filtering on the voltage response data to form narrowband response data, the slow-changing trend term and non-target frequency band noise interference under fast charging conditions can be effectively suppressed, thus providing a high signal-to-noise ratio input for perturbation frequency response extraction. Further, phase codebook matching correlation analysis is performed using phase parameters to obtain code correlation output data, enabling the phase-encoded, separable extraction of multi-frequency parallel perturbations at the demodulation end, reducing inter-frequency aliasing and enhancing the detection capability for weak response frequencies. Subsequently, frequency orthogonality correction is performed on the code correlation output based on amplitude parameters to obtain orthogonal complex response data, which can decouple and compensate for crosstalk caused by non-ideal orthogonality of the codebook and system coupling, making the complex responses at each frequency more independent and scale-consistent. Finally, complex domain coherent accumulation is performed based on frequency parameters, and phase drift compensation is performed to obtain complex response spectrum data. This can achieve in-phase superposition gain even in the presence of sampling jitter, frequency shift, or phase drift, significantly improving the stability and accuracy of the complex response spectrum.

[0120] In an exemplary embodiment, based on impedance-related characteristic data, a budget allocation analysis is performed on the lithium plating risk and thermal risk in a fast charging scenario to obtain charging risk budget data, including steps 802 to 806. Wherein:

[0121] Step 802: Perform temperature-based mapping on the impedance-related characteristic data to obtain temperature-affected decoupling impedance characteristic data; Step 804: Based on the temperature-affected decoupling impedance characteristic data, a margin-first analysis is performed on the lithium plating risk and thermal risk to obtain the risk-dominant sequence data. Step 806: Perform cross-risk dynamic redistribution analysis on the risk-dominant sequence data to obtain charging risk budget data.

[0122] Among them, temperature reference mapping is a process of converting the impedance-related features extracted at the current temperature to a preset reference temperature for temperature compensation, so as to eliminate the systematic influence of temperature on the feature values.

[0123] Among them, the temperature-affected decoupling impedance characteristic data is the impedance characteristic data obtained after temperature reference mapping, which characterizes the battery's ohmic internal resistance, interface processes and diffusion polarization under a unified reference temperature condition.

[0124] Among them, margin priority analysis is an analytical process that calculates lithium plating margin and thermal margin separately and determines the risk control priority according to the urgency of the margin.

[0125] Among them, the risk-dominant sequence data is a sequence or weighted data output by margin-first analysis, used to characterize the current dominant risk type and the priority order of lithium plating risk and thermal risk.

[0126] Among them, the cross-risk dynamic redistribution analysis is an analytical process that recursively updates and closes the budget for lithium plating and thermal risks based on the risk-dominant sequence, so that the budget is dynamically adjusted between the two types of risks as the operating conditions change.

[0127] Specifically, using a preset reference temperature as a unified reference, the impedance-related characteristics extracted under the current temperature conditions (such as ohmic internal resistance characteristics, interface charge transfer characteristics, diffusion polarization characteristics, and their derived indices) are converted using a temperature compensation model or calibration mapping relationship to make them equivalent to the characteristic expressions at the reference temperature. The temperature compensation model can use lookup table interpolation, piecewise linear mapping, or parameterized models to correct the temperature coefficient and output temperature-affected decoupling impedance characteristic data.

[0128] Lithium plating margin and thermal margin are separately constructed as metrics. The lithium plating margin can be derived from the decoupling impedance characteristic data, characteristic rate of change, or a combination thereof related to the temperature effect on negative electrode polarization / diffusion limitation. The thermal margin can be derived from the decoupling impedance characteristic data related to internal resistance heating, combined with the current temperature and temperature rise trend. The two types of margins are then compared to form risk-dominant sequence data. The risk-dominant sequence data is used to characterize which type of risk is dominant in the current fast charging stage and the priority order of the two types of risks, and can simultaneously output the corresponding margin difference or confidence weight.

[0129] Prioritizing the dominant risks identified by the risk-dominant sequence, a basic budget is allocated to these risks to ensure their risk growth is controlled. Then, based on the margin of another risk, risk coupling trends (e.g., increased internal resistance leading to increased heat generation and induced polarization), and total budget constraints, the budgets for non-dominant risks are dynamically revised and closed. This redistribution can employ recursive update rules or rolling window integral constraints, allowing the budget to adaptively adjust with margin changes in each control cycle. The final output is charging risk budget data, which includes at least one or both of the lithium plating risk budget data and the thermal risk budget data.

[0130] In this embodiment, temperature-dependent impedance characteristic data is obtained by mapping impedance-related characteristic data to a temperature reference. This effectively reduces the interference of temperature changes on apparent impedance drift, allowing subsequent risk assessment to focus more on the electrochemical state itself. Furthermore, based on the decoupled impedance characteristics, margin-priority analysis is performed on lithium plating risk and thermal risk to generate risk-dominant sequence data. This can automatically identify the current dominant risk in different SOC and temperature ranges and determine the risk control priority according to the margin urgency, avoiding overly conservative or localized loss of control caused by fixed weights. Finally, charging risk budget data is obtained by dynamically reallocating across risks based on the risk-dominant sequence. This allows the limited risk margin to be adaptively allocated according to the dominant risk priority and the non-dominant risk restriction, and updated in real time with the operating conditions. This improves the constraint efficiency and robustness of the two key failure modes, lithium plating and overheating, during fast charging.

[0131] In an exemplary embodiment, a cross-risk dynamic reallocation analysis is performed on the risk-dominant sequence data to obtain charging risk budget data, including steps 902 to 908. Wherein:

[0132] Step 902: Lock the dominant risk in the risk-dominant sequence data to obtain dominant risk type data and basic budget data corresponding to the dominant risk type; Step 904: Based on the dominant risk type data, perform cross-coupling analysis on the risk sensitivity data in the temperature influence decoupling impedance characteristic data to obtain cross-risk coupling coefficient data; Step 906: Based on the cross-risk coupling coefficient data, perform reverse constraint propagation analysis on the basic budget data to obtain the budget ceiling data for non-dominant risks; Step 908: Based on the budget ceiling data of non-dominant risks, perform budget closure processing on the charging risk budget data to obtain the charging risk budget data.

[0133] Among them, the dominant risk lock-in is the process of determining the risk categories that should be prioritized for control based on the risk dominance sequence and using them as the main objects of budget allocation.

[0134] Among them, the dominant risk type data is the category identification data that represents the current dominant risk, which is usually one of lithium plating risk or thermal risk.

[0135] Among them, the basic budget data is the minimum guaranteed budget amount or initial budget parameter data that is pre-allocated to the dominant risk at the beginning of the budget redistribution phase.

[0136] Among them, risk sensitivity data is a quantitative data describing the degree of influence of changes in impedance characteristics on changes in the intensity of lithium plating risk or thermal risk.

[0137] Among them, cross-coupling analysis is an analytical process based on risk sensitivity assessment of the mutual influence path between two types of risks and calculation of their coupling relationship.

[0138] Among them, the cross-risk coupling coefficient data is coefficient or matrix data obtained from cross-coupling analysis, used to quantify the degree to which lithium plating risk and thermal risk amplify or inhibit each other.

[0139] Among them, reverse constraint propagation analysis is an analytical process that propagates the budget constraints of the dominant risk back to the non-dominant risk side through cross-risk coupling relationships to derive its allowable boundary.

[0140] Among them, the budget ceiling data for non-dominant risks is the maximum budget amount that can be allocated to non-dominant risks in the current period without weakening the control effect of dominant risks.

[0141] Among them, budget closure processing is a process of consistency correction and pairing output of two types of risk budgets under the total budget, upper limit constraint and non-negativity constraint.

[0142] Specifically, in each control cycle, the priority ranking or weighting results of the risk-dominant sequence data are read, and the highest priority data is determined as the dominant risk type (lithium plating risk or thermal risk). Basic budget data corresponding to this dominant risk type is then generated according to a preset basic quota rule for the dominant risk. The basic quota rule can adopt a structure of "minimum guaranteed budget for dominant risk + remaining budget to be allocated," and the size of the basic budget can be updated iteratively based on the SOC range, temperature range, or the urgency of the dominant risk margin.

[0143] Based on the dominant risk type data, sensitivity indicators related to the two types of risks are extracted from the temperature-affected decoupling impedance characteristic data (e.g., sensitivity reflecting increased heating due to increased internal resistance, sensitivity reflecting increased polarization due to limited diffusion, and the slope or elasticity coefficient of both as they change with operating conditions). Using the dominant risk type data as a reference, a "mapping of the impact of dominant risk changes on non-dominant risks" and a "feedback mapping of non-dominant risk changes on dominant risks" are established, and cross-risk coupling coefficient data are calculated. These cross-risk coupling coefficients can be represented as scalar coefficients, matrix coefficients, or piecewise coefficients, used to characterize the mutual amplification, mutual inhibition, or weak coupling relationship between the two types of risks.

[0144] Using the baseline budget of the dominant risk type data as a priori constraint input, the non-dominant risk increment that may be caused by changes in the dominant risk budget is propagated to the non-dominant risk side as a reverse constraint using the cross-risk coupling coefficient data. During the calculation, the reverse constraint propagation can employ inequality constraint derivation, linear approximation propagation, or robust boundary contraction strategies. This ensures that while the dominant risk budget is not weakened, the non-dominant risk budget is limited to a safe range that will not reversely increase the dominant risk or cause the system to exceed its limits. This allows for the calculation of the maximum allowable budget boundary for non-dominant risks in the current period, i.e., the upper limit data for the non-dominant risk budget.

[0145] Under the total budget constraint, the dominant risk budget is fixed as the basic budget data, and the non-dominant risk budget is restricted to the budget upper limit data range. The remaining available budget is allocated or recovered, so that the two types of risk budgets meet the three constraints of "non-negativity, upper limit constraint and total closure". At the same time, smoothing or rate of change constraint can be applied to the budget results to avoid drastic changes in the budget between adjacent periods. Finally, the closed charging risk budget data is output, which includes the paired budget results of lithium plating risk budget data and thermal risk budget data.

[0146] In this embodiment, by locking the dominant risk in the risk-dominant sequence data and prioritizing the allocation of the basic budget, constraint resources can be concentrated on the most pressing risk channels in a timely manner during fast charging, avoiding the loss of control of dominant risks due to fixed weight allocation. Furthermore, based on the risk sensitivity of the temperature-dependent decoupling impedance characteristics of the dominant risk type, cross-risk coupling coefficients are obtained through cross-coupling analysis. This explicitly characterizes the mutual amplification or suppression relationship between lithium plating risk and thermal risk, thereby upgrading budget allocation from "independent control" to "coupled constraint control". Then, the coupling coefficients are used to perform reverse constraint propagation on the basic budget to obtain the upper limit of the non-dominant risk budget. This can suppress the reverse push-up effect of non-dominant risks on dominant risks while ensuring the effectiveness of the dominant risk budget, reducing the blind spot of budget allocation. Finally, through budget closure processing, consistent and executable charging risk budget data is output within the boundaries of the total budget and the upper limit, which can improve the stability and feasibility of the risk budget and provide more accurate and robust constraint inputs for subsequent rolling constraint optimization, thereby improving the efficiency and stability of the fast charging strategy while meeting the safety boundary.

[0147] Based on the same inventive concept, this application also provides a charging current adaptive curve optimization device for fast charging scenarios, used to implement the aforementioned charging current adaptive curve optimization method for fast charging scenarios. For example... Figure 3 As shown, it includes: a scene data acquisition module, a binocular waveform analysis module, a spectral response analysis module, a budget allocation analysis module, and a rolling constraint optimization analysis module. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the charging current adaptive curve optimization device for fast charging scenarios provided below can be found in the limitations of the charging current adaptive curve optimization method for fast charging scenarios described above, and will not be repeated here.

[0148] The modules in the aforementioned charging current adaptive curve optimization device for fast charging scenarios can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0149] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. This computer device includes a processor, memory, input / output interfaces (I / O), and communication interfaces. Those skilled in the art will understand that... Figure 4The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0150] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0151] In one embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0152] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and executes the computer instructions, causing the computer device to perform the steps in the above method embodiments.

[0153] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0154] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0155] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0156] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for adaptive charging current curve optimization in fast charging scenarios, characterized in that, The method includes: Acquire charging current data, individual cell voltage data, module voltage data, and charger capability data related to the battery pack's operation in fast charging scenarios; Based on the charger capability data, a dual-view waveform analysis is performed on the charging current data to obtain the charging current waveform data. Based on the individual unit voltage data and the module voltage data, spectral response analysis is performed on the multi-frequency detection perturbation component in the charging current waveform data to obtain impedance-related characteristic data. Based on the impedance-related characteristic data, a budget allocation analysis is performed on the lithium plating risk and thermal risk in the fast charging scenario to obtain charging risk budget data. Based on the charging risk budget data, rolling constraint optimization analysis is performed on the charging current waveform data in the next time domain to obtain the charging current adaptive curve data.

2. The method according to claim 1, characterized in that, The step of performing dual-view waveform analysis on the charging current data based on the charger capability data to obtain charging current waveform data includes: Based on the charger capability data, a capability mapping analysis is performed on the charger output capability and the upper limit of battery charging current in the fast charging scenario to obtain current feasible domain data. Based on the current feasible domain data, information gain configuration analysis is performed on the multi-frequency detection perturbation component in the charging current data to obtain the information-optimal multi-frequency detection perturbation component. Based on the information, the optimal multi-frequency detection perturbation component is used to perform base current allocation processing on the remaining current margin in the current feasible domain data to obtain energy transfer base current component data. The charging current waveform data is obtained by superimposing the energy transfer base current component data with the information optimal multi-frequency detection perturbation component.

3. The method according to claim 2, characterized in that, The step of performing information gain configuration analysis on the multi-frequency detection perturbation component in the charging current data based on the current feasible domain data to obtain the information-optimal multi-frequency detection perturbation component includes: Amplitude limiting and ripple constraint mapping are performed on the current feasible domain data to obtain the power spectrum budget data of the multi-frequency detection perturbation component; Based on the power spectrum budget data of the multi-frequency detected perturbation components, a structured complementary sensitivity arrangement analysis is performed on the candidate perturbation frequency points in the charging current data to obtain multi-frequency complementary frequency point set data; Encoded information gain allocation is performed on the amplitude and phase parameters within the multi-frequency complementary frequency point set data to obtain the optimal multi-frequency detection perturbation component.

4. The method according to claim 3, characterized in that, The step involves performing structured complementary sensitivity arrangement analysis on candidate perturbation frequency points in the charging current data based on the multi-frequency detected perturbation component power spectrum budget data, to obtain a set of multi-frequency complementary frequency points, including: Based on the power spectrum budget data of the multi-frequency detected perturbation component, the candidate perturbation frequency points in the charging current data are pre-screened to obtain candidate frequency point pool data. Based on the candidate frequency pool data, target feature decoupling degree analysis is performed on each candidate perturbation frequency to obtain frequency decoupling degree score data. Based on the frequency point decoupling score data, complementary coverage arrangement analysis is performed on the candidate frequency point pool data to obtain complementary frequency point skeleton set data. Based on the complementary frequency skeleton set data, the power spectrum budget data is segmented and mapped across frequency points to obtain the multi-frequency complementary frequency set data.

5. The method according to claim 1, characterized in that, The step involves performing spectral response analysis on the multi-frequency detection perturbation component in the charging current waveform data based on the individual cell voltage data and the module voltage data to obtain impedance-related characteristic data, including: The voltage data of the individual units and the voltage data of the modules are subjected to noise reduction preprocessing to obtain voltage response data; Based on the frequency point parameters, amplitude parameters, and phase parameters in the multi-frequency detected perturbation components, the voltage response data is phase-encoded and demodulated to obtain complex response spectrum data; Based on the complex response spectrum data, the individual unit voltage data and the module voltage data are fused under cross-scale consistency constraints to obtain the impedance-related feature data.

6. The method according to claim 5, characterized in that, The step of performing phase encoding demodulation on the voltage response data based on the frequency parameters, amplitude parameters, and phase parameters of the multi-frequency detected perturbation components to obtain complex response spectrum data includes: Baseline component removal and bandwidth constraint filtering are performed on the voltage response data to obtain narrowband response data; Based on the phase parameters, phase codebook matching correlation analysis is performed on the narrowband response data to obtain code correlation output data; Based on the amplitude parameter, frequency point orthogonality correction is performed on the code correlation output data to obtain orthogonal complex response data; Based on the frequency parameters, the orthogonal complex response data is subjected to complex domain coherent accumulation and phase drift compensation to obtain the complex response spectrum data.

7. The method according to claim 1, characterized in that, The step involves performing a budget allocation analysis on the lithium plating risk and thermal risk in the fast charging scenario based on the impedance-related characteristic data, to obtain charging risk budget data, including: Temperature-referenced mapping is performed on the impedance-related characteristic data to obtain temperature-affected decoupled impedance characteristic data; Based on the temperature-affected decoupling impedance characteristic data, a margin-first analysis is performed on the lithium plating risk and thermal risk to obtain risk-dominant sequence data. The charging risk budget data is obtained by performing cross-risk dynamic redistribution analysis on the risk-dominant sequence data.

8. The method according to claim 7, characterized in that, The process of performing cross-risk dynamic reallocation analysis on the risk-dominant sequence data to obtain the charging risk budget data includes: The dominant risk sequence data is used to lock the dominant risk, resulting in dominant risk type data and basic budget data corresponding to the dominant risk type. Based on the dominant risk type data, cross-coupling analysis is performed on the risk sensitivity data in the temperature effect decoupling impedance characteristic data to obtain cross-risk coupling coefficient data. Based on the cross-risk coupling coefficient data, a reverse constraint propagation analysis is performed on the basic budget data to obtain the budget ceiling data for non-dominant risks; Based on the budget ceiling data of the non-dominant risk, the charging risk budget data is subjected to budget closure processing to obtain the charging risk budget data.

9. A charging current adaptive curve optimization device for fast charging scenarios, characterized in that, The device includes: The scenario data acquisition module is used to acquire charging current data, single cell voltage data, module voltage data, and charger capability data related to the battery pack operation process in fast charging scenarios. The dual-view waveform analysis module is used to perform dual-view waveform analysis on the charging current data based on the charger capability data to obtain charging current waveform data. The spectral response analysis module is used to perform spectral response analysis on the multi-frequency detection perturbation component in the charging current waveform data based on the individual cell voltage data and the module voltage data, and to obtain impedance-related characteristic data. The budget allocation analysis module is used to perform budget allocation analysis on the lithium plating risk and thermal risk in the fast charging scenario based on the impedance-related characteristic data, and obtain charging risk budget data. The rolling constraint optimization analysis module is used to perform rolling constraint optimization analysis on the charging current waveform data in the next time domain based on the charging risk budget data, so as to obtain the charging current adaptive curve data.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 8.

Images