Battery health state evaluation-based battery car dynamic charging optimization management system

Abstract

The invention provides a battery car dynamic charging optimization management system based on battery health state evaluation, and belongs to the technical field of battery car charging management. The system comprises a data acquisition module which is connected with an environment sensing device and a battery management system and acquires environment and battery parameters in real time; the data storage module is used for storing an initial training data set, a charging and discharging data set and a battery aging feature library; the model building module is used for building a battery health state evaluation model through a migration strategy and building a response model through a modeling strategy; the health state evaluation module is used for extracting associated features with a critical damage threshold value as a core through a mapping algorithm based on the battery parameters, the environmental parameters and an aging feature library; and the charging strategy generation module comprises a strategy generation unit, a reference parameter generation unit and a dynamic correction unit. The system effectively improves the adaptability of battery car charging, guarantees the efficient and stable operation of the battery, and remarkably improves the safety and prolongs the service life of the battery.

Description

Technical Field

[0001] This invention relates to the field of electric vehicle charging management technology, and more specifically to a dynamic charging optimization management system for electric vehicles based on battery health status assessment. Background Technology

[0002] Electric bicycles, with their convenience and environmental friendliness, have become an important tool for short-distance urban travel. Users are increasingly demanding higher ranges, better charging efficiency, and longer battery lifespans. In practical use, electric bicycles often face diverse environmental conditions, and batteries gradually age during long-term charge-discharge cycles. These factors all affect charging performance and battery safety. Therefore, how to achieve charging management that meets the needs of fast charging while protecting the battery and extending its lifespan in diverse usage scenarios has become a pressing issue in the current use of electric bicycles.

[0003] The existing technology has the following shortcomings: First, the charging strategy relies on fixed parameters or empirical values ​​and fails to be deeply correlated with the real-time health status of the battery, resulting in the charging parameters not matching the actual tolerance of the battery, which can easily accelerate battery aging with long-term use; Second, the consideration of environmental factors is relatively simple, and charging is usually limited by simple thresholds. There is a lack of multi-dimensional correlation analysis between environmental parameters and battery performance degradation, making it difficult to adapt to the safe charging needs in complex environments. Summary of the Invention

[0004] This invention provides a dynamic charging optimization management system for electric vehicles based on battery health status assessment, which solves the problems in the prior art, such as low matching degree between charging strategy and real-time battery health status, and difficulty in adapting to safe charging needs in complex environments due to single consideration of environmental factors.

[0005] To achieve the above objectives, embodiments of the present invention provide a dynamic charging optimization management system for electric bicycles based on battery health status assessment. The system comprises: a data acquisition module, connected to an environmental sensing device and the battery management system of the electric bicycle, for collecting environmental parameters and battery parameters; a data storage module, storing a preset initial training dataset, a battery charging / discharging dataset, and a battery aging feature library; a model building module, used to construct a battery health status assessment model based on the initial training dataset using a preset transfer strategy to assess the battery's health status; and also used to construct a response model based on the collected environmental parameters and the initial training dataset using a preset modeling strategy to adjust the charging strategy; and a health status assessment module, used to assess the battery's health status based on battery parameters... The system evaluates the battery using a battery health status assessment model to obtain assessment results. It also extracts associated features based on the assessment results, environmental parameters, and a battery aging feature library using a preset mapping algorithm to characterize the critical damage thresholds of various battery parameters. The charging strategy generation module includes a strategy generation unit, a baseline parameter generation unit, and a dynamic correction unit. The strategy generation unit generates a charging strategy based on the assessment results and using preset health-power matching rules. The baseline parameter generation unit performs coupled analysis of environmental parameters and battery charge / discharge datasets using the response model to generate a baseline adjustment parameter table. The dynamic correction unit corrects the parameters in the baseline adjustment parameter table based on environmental parameters and associated features using a preset adjustment strategy and updates the charging strategy.

[0006] Optionally, the migration strategy includes several preset difference intervals, each of which corresponds to an adaptation rule. The migration strategy includes: classifying the initial training dataset according to the battery's chemical system and establishing a feature mapping table to represent the health characteristic benchmarks of different types of batteries; calculating the distribution differences of each type of initial training dataset based on the established feature mapping table using the maximum mean difference algorithm, and determining the preset adaptation rule corresponding to the model parameters based on the distribution differences; dynamically adjusting the model parameters according to the adaptation rule to generate model parameter configurations adapted to different types of batteries, thereby constructing a battery health status assessment model.

[0007] Optionally, the evaluation results include a battery health status score and a health level, with each health level corresponding to a preset score range. The battery health status evaluation model evaluates the battery by: performing feature fusion on the input battery parameters to extract multi-dimensional feature vectors characterizing battery capacity decay, internal resistance growth, and polarization characteristics; mapping the multi-dimensional feature vectors to a unified health evaluation space based on the feature mapping table and adaptation rules to generate health status feature values; and calculating the battery health status score through regression analysis based on the health status feature values, and determining the health level according to the score range.

[0008] Optionally, the modeling strategy includes: classifying and clustering the initial training dataset and environmental parameters, dividing the environmental parameters into several environmental feature groups, each environmental feature group corresponding to a set of charging and discharging scenarios; for each environmental feature group, extracting the critical threshold parameters for battery charging and discharging under that scenario from the initial training dataset; constructing an association matrix between environmental parameters and battery charging and discharging safety boundaries based on each environmental feature group and its corresponding critical threshold parameters; and iteratively calculating the association matrix using a preset rule reasoning algorithm to generate a set of charging and discharging response rules under different combinations of environmental parameters as the response model.

[0009] Optionally, the mapping algorithm includes: constructing a multi-dimensional feature set based on evaluation results, environmental parameters, and a battery aging feature library; and performing coupling operations on the multi-dimensional features in the multi-dimensional feature set through a feature fusion algorithm to obtain associated features.

[0010] Optionally, the construction of the multi-dimensional feature set includes: performing feature decomposition on the battery health status parameters in the evaluation results to extract basic feature vectors characterizing the internal physicochemical properties of the battery; encoding environmental parameters by type to generate real-time environmental feature vectors; using the battery aging feature library, the current health status, and the real-time environmental feature vectors as inputs, and outputting a critical damage threshold through a preset safety boundary prediction model; and fusing the basic feature vectors, environmental feature vectors, and critical damage thresholds to form a multi-dimensional feature set.

[0011] Optionally, each health level corresponds to a baseline charging power range, and the health-power matching rule includes: for each health level, determining the adjustment step size through a preset calculation rule; when the battery health level is in a preset low-level range, adjusting the maximum charging power to a preset proportion of the upper limit of the baseline charging power range corresponding to that health level, and using the adjusted maximum charging power as a constraint condition for charging power adjustment; based on the constraint condition, combined with the baseline charging power range corresponding to the health level and the adjustment step size, determining the charging power value for each time period through a preset power allocation formula to generate a charging strategy adapted to the current battery health state.

[0012] Optionally, the adjustment strategy includes: calculating the deviation between the real-time rate of change of environmental parameters and the average rate of change within a preset time window, and determining whether to update the strategy based on the deviation; when updating the strategy, selecting target parameters that match the current environmental parameters from the benchmark adjustment parameter table, and extracting the critical damage threshold from the associated features; calculating a safety margin index based on the difference between the real-time monitored battery state data and the critical damage threshold using preset quantization rules, and determining the parameter correction coefficient; correcting the target parameters based on the parameter correction coefficient and the critical damage threshold using preset correction rules; adjusting the benchmark adjustment parameter table according to the corrected target parameters, and updating the charging strategy according to the adjusted benchmark adjustment parameter table.

[0013] Optionally, the correction rule includes: correcting the target parameter using a preset correction function; if the corrected target parameter exceeds the corresponding critical damage threshold, then repeatedly correcting the target parameter using a gradient reduction method, and interrupting charging when the number of corrections reaches a preset threshold.

[0014] Optionally, the electric vehicle dynamic charging optimization management system further includes an adaptive optimization module, which is used to: collect new data generated during system operation, compare the new data with the data in the initial training dataset to obtain the difference value; when the difference value exceeds a preset threshold, merge the new data into the initial training dataset, and use an incremental training algorithm to update the battery health status assessment model and response model.

[0015] The present invention provides a dynamic charging optimization management system for electric vehicles based on battery health status assessment. This system acquires environmental and battery parameters in real time through a data acquisition module, relies on basic data support from a data storage module, establishes an appropriate assessment and response model through a model building module, extracts relevant features through a health status assessment module, and finally generates a complete charging strategy from initial to dynamic correction through a charging strategy generation module. This forms a closed-loop management mechanism of "data acquisition - model assessment - strategy generation - strategy adjustment," enabling dynamic optimization of the charging strategy based on battery health status and environmental conditions, ensuring charging safety while improving charging adaptability. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in this invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0017] Figure 1This is a framework diagram of the electric vehicle dynamic charging optimization management system provided in an embodiment of the present invention;

[0018] Figure 2 This is the overall control flowchart of the electric vehicle dynamic charging optimization management system provided in the embodiments of the present invention;

[0019] Figure 3 This is a flowchart of the response modeling strategy provided in an embodiment of the present invention;

[0020] Figure 4 This is a flowchart of the health-power matching rules provided in an embodiment of the present invention. Detailed Implementation

[0021] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0022] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application all comply with the relevant provisions of national laws and regulations. In the embodiments of this application, certain existing industry solutions such as software, components, and models may be mentioned. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this application, and do not imply that the applicant has already used or necessarily used such solutions.

[0023] As mentioned above, with the deep integration of electric vehicles and smart charging technology, battery health management in complex environments faces numerous challenges. Traditional charging systems struggle to dynamically adjust charging parameters based on the battery's real-time health status and cannot accurately adapt charging strategies to environmental changes, leading to shortened battery life and increased safety risks. Charging efficiency and battery protection are difficult to balance. Therefore, developing a more efficient dynamic charging optimization management system for electric vehicles is extremely important.

[0024] To address this issue, this invention provides a dynamic charging optimization management system for electric vehicles based on battery health status assessment. By collecting environmental and battery parameters, a health status assessment model and a response model are constructed to analyze the battery status and environmental impact, extracting related features to determine critical damage thresholds, generating and dynamically correcting charging strategies to form a closed-loop management system throughout the entire process. This effectively improves the adaptability and safety of electric vehicle charging and ensures efficient and stable battery operation.

[0025] The following is combined with Figures 1-4 This invention is described in detail.

[0026] like Figure 1 and Figure 2As shown, this embodiment of the invention provides a dynamic charging optimization management system for electric bicycles based on battery health status assessment. The system comprises: a data acquisition module, connected to an environmental sensing device and the battery management system of the electric bicycle, to collect environmental parameters and battery parameters; a data storage module, storing a preset initial training dataset, a battery charging / discharging dataset, and a battery aging feature library; a model building module, used to construct a battery health status assessment model based on the initial training dataset using a preset transfer strategy to assess the battery's health status; and also used to construct a response model based on the collected environmental parameters and the initial training dataset using a preset modeling strategy to adjust the charging strategy; and a health status assessment module, used to assess the battery's health status based on the battery parameters... The battery health status assessment model evaluates the battery to obtain assessment results. It also extracts related features based on the assessment results, environmental parameters, and a battery aging feature library using a preset mapping algorithm to characterize the critical damage thresholds of various battery parameters. The charging strategy generation module includes a strategy generation unit, a baseline parameter generation unit, and a dynamic correction unit. The strategy generation unit generates a charging strategy based on the assessment results and a preset health-power matching rule. The baseline parameter generation unit performs coupled analysis of environmental parameters and battery charge / discharge datasets using the response model to generate a baseline adjustment parameter table. The dynamic correction unit corrects the parameters in the baseline adjustment parameter table based on environmental parameters and related features using a preset adjustment strategy and updates the charging strategy.

[0027] Specifically, the initial training dataset contains charge-discharge cycle data for different types of batteries, such as lead-acid and lithium-ion batteries, under standard environmental conditions. Each data record includes a timestamp, charge / discharge current, voltage, initial and final SOC, and health status labels. The battery charge-discharge dataset covers charge-discharge curves, capacity decay rates, and safety threshold parameters of batteries under different combinations of environmental parameters. The battery aging feature library stores data covering aging-related features throughout the battery's entire lifecycle, including: basic battery parameters, cycle data, performance decay curves, charge-discharge characteristics, environmental impact data, fault correlation data, and model calibration data. This data provides underlying feature support for battery aging trend prediction, health level classification, and charging strategy adaptation. Correlation features refer to safety boundary features of parameters such as voltage, current, and temperature that reflect the battery's current health status and environmental conditions, such as maximum allowable charging current and maximum allowable temperature. These features together constitute the critical damage threshold; that is, when the parameters exceed this threshold, the risk of battery damage increases significantly. The benchmark adjustment parameter table refers to a table of benchmark charging parameter values ​​obtained through response model analysis under different environmental parameters, such as benchmark charging current and benchmark charging voltage at different temperatures. The correction process of the dynamic correction unit refers to adjusting the parameters in the baseline adjustment parameter table based on changes in real-time environmental parameters and critical damage thresholds in the associated characteristics. This ensures that the final charging strategy adapts to environmental changes while preventing battery damage. The charging strategy includes a charging power curve, which specifies the power values ​​and durations of different charging stages.

[0028] Specifically, during the system deployment phase, the data storage module calls upon a pre-set initial training dataset, and the model building module constructs a battery health status assessment model using a transfer strategy. Simultaneously, a dynamic response model is generated based on the environmental parameter set and modeling strategy. Finally, the trained battery health status assessment model is transferred to the health status assessment module, and the response model is transferred to the charging strategy generation module, completing the initialization of the system's core capabilities. During the system operation phase, the data acquisition module collects battery and environmental parameters in real time and transmits them to the health status assessment module. This module calculates the assessment results based on the battery health status assessment model and dynamically generates critical damage thresholds using a battery aging feature library and mapping algorithm. The charging strategy generation module's strategy generation unit generates a charging strategy using health-power matching rules, while the benchmark parameter generation unit analyzes environmental parameters and charge / discharge datasets using the response model and outputs a benchmark adjustment parameter table. The dynamic correction unit dynamically adjusts the charging strategy based on the environmental change rate and critical damage threshold through adjustment strategies.

[0029] The electric vehicle dynamic charging optimization management system provided in this invention achieves dynamic optimization of charging safety through a multi-module collaborative closed-loop architecture: the data acquisition module directly connects to the environmental sensing device and the battery management system, synchronously capturing environmental parameters and battery parameters, eliminating data fragmentation in traditional solutions; the data storage module integrates the initial training dataset, charge / discharge dataset, and aging feature library, constructing a multi-source knowledge foundation; the model building module generates a health assessment model and a response model in parallel based on transfer and modeling strategies, laying the foundation for dual capabilities of state perception and strategy control; the health status assessment module integrates real-time battery parameters, environmental parameters, and the aging feature library, extracting related features with critical damage thresholds as the core through mapping algorithms, realizing quantitative early warning of damage risks; the charging strategy generation module innovatively designs a three-level architecture of strategy generation unit, benchmark parameter generation unit, and dynamic correction unit—the strategy unit generates a basic strategy according to the power rules matched to the health status, the benchmark unit couples environmental and charge / discharge data to output an adjustment parameter table, and the correction unit adjusts parameters and updates the strategy in real time based on environmental mutations and critical thresholds, forming a full-process adaptive control of "data-driven - health assessment - benchmark generation - dynamic correction", which greatly improves the safety response speed compared to traditional static charging solutions.

[0030] Preferably, the migration strategy includes: classifying the initial training dataset according to the battery's chemical system and establishing a feature mapping table to represent the health characteristic benchmarks of different types of batteries; calculating the distribution differences of each type of initial training dataset using the maximum mean difference algorithm based on the established feature mapping table, and determining the preset adaptation rules corresponding to the model parameters based on the distribution differences; dynamically adjusting the model parameters according to the adaptation rules to generate model parameter configurations adapted to different types of batteries, so as to construct a battery health status assessment model.

[0031] More preferably, the migration strategy has a number of pre-defined difference intervals, and each difference interval corresponds to an adaptation rule.

[0032] Specifically, the "battery chemical system" includes lead-acid battery system and lithium-ion battery system. Batteries with different chemical systems exhibit significant differences in charge-discharge characteristics and health degradation patterns. A feature mapping table records core health characteristic parameters of batteries with different chemical systems, such as capacity degradation coefficient, internal resistance growth rate, and charge-discharge efficiency benchmark values, along with their corresponding weights. The maximum mean difference algorithm quantifies the data distribution distance by calculating the mean difference of different initial training datasets in the regenerating kernel Hilbert space; the calculation result is expressed numerically, ranging from 0 to 1. The adaptation rule refers to the preset model parameter adjustment formula for different distribution difference values. The difference interval is a numerical range divided based on the calculation results of the maximum mean difference algorithm. Each interval corresponds to a unique adaptation rule. The system automatically calls the corresponding adaptation rule by identifying the interval to which the difference value belongs, achieving rapid adjustment of model parameters and ensuring consistent evaluation accuracy of the health status assessment models for batteries with different chemical systems.

[0033] In a preferred embodiment of the present invention, the migration strategy achieves the adaptability of the battery health status assessment model to batteries with different chemical systems through classification mapping, difference calculation, and rule adaptation. First, the initial training dataset is classified according to the battery chemical system, and a feature mapping table is established. Then, the maximum mean difference algorithm is used to calculate the distribution difference to determine the adaptation rules, and finally, the model parameters are dynamically adjusted. Furthermore, multiple preset difference intervals are defined, each interval corresponding to a specific adaptation rule. The entire technical solution forms a complete logic of "classification-difference quantification-interval matching-rule invocation," enabling the battery health status assessment model to accurately adjust parameter configurations according to the differences in chemical characteristics of different batteries, thereby improving the generalization ability of the battery health status assessment model to diverse battery types.

[0034] Preferably, the evaluation results include a battery health status score and a health level, with each health level corresponding to a preset score range. The battery health status evaluation model evaluates the battery by: performing feature fusion on the input battery parameters to extract multi-dimensional feature vectors characterizing battery capacity decay, internal resistance growth, and polarization characteristics; mapping the multi-dimensional feature vectors to a unified health evaluation space based on the feature mapping table and adaptation rules to generate health status feature values; and calculating the battery health status score through regression analysis based on the health status feature values, and determining the health level according to the score range.

[0035] Specifically, feature fusion can employ an autoencoder neural network to compress the input multidimensional battery parameters into 5-7 dimensional feature vectors, with each dimension corresponding to a core feature in battery capacity decay, internal resistance growth, or polarization characteristics. A unified health assessment space refers to a common feature space to which feature vectors from batteries with different chemical systems are mapped using a feature mapping table. Each coordinate point in this space corresponds to a unique health status feature value, eliminating the influence of chemical system differences on the assessment results. Regression analysis can use a support vector regression algorithm, taking the health status feature value as input and outputting a health status score of 0%-100%. The health level system adopts a three-level hierarchical architecture: Level 1, Level 2, and Level 3. Each level corresponds to a preset scoring range, where Level 1 represents the optimal health status (SOH ≥ 85%), Level 2 represents moderate decay (70% ≤ SOH < 85%), and Level 3 indicates severe aging (SOH < 70%).

[0036] The battery health status assessment model provided by the preferred embodiment of the present invention reduces redundant parameter interference and improves assessment accuracy through feature fusion; the unified health assessment space solves the problem of inconsistent assessment standards for different battery types; and the hierarchical scoring mechanism provides an intuitive decision basis for charging strategy generation, enabling the system to respond quickly to changes in battery health status.

[0037] like Figure 3 As shown, preferably, the modeling strategy includes: classifying and clustering the initial training dataset and environmental parameters, dividing the environmental parameters into several environmental feature groups, each environmental feature group corresponding to a set of charging and discharging scenarios; for each environmental feature group, extracting the critical threshold parameters for battery charging and discharging under that scenario from the initial training dataset; constructing an association matrix between environmental parameters and battery charging and discharging safety boundaries based on each environmental feature group and its corresponding critical threshold parameters; and iteratively calculating the association matrix using a preset rule reasoning algorithm to generate a set of charging and discharging response rules under different combinations of environmental parameters, which serves as the response model.

[0038] Specifically, the classification and clustering uses the K-means clustering algorithm to divide environmental parameters into multiple environmental feature groups based on similarity, such as "high temperature and high humidity group" and "normal temperature and normal humidity group." Each group corresponds to a typical charging and discharging scenario, such as outdoor charging in summer and indoor charging in spring and autumn. The advantage of this classification method is that it can accurately capture the common features of different environments, avoid isolated analysis of environmental parameters, and make subsequent scenario matching more targeted. Critical threshold parameters include the maximum allowable charging current, minimum discharge cutoff voltage, and temperature protection trigger value of the battery in the scenario, which are extracted from the initial training dataset using the percentile method. The matrix elements in the association matrix represent the safety threshold of the corresponding parameter under a certain environmental group. The rule reasoning algorithm uses a generative rule engine. Through iterative reasoning of "if the environmental feature group is A, then the charging and discharging parameters must satisfy B", it generates a response rule set containing 20-50 rules. Each rule clearly defines the correspondence between the combination of environmental parameters and the charging and discharging operation. This rule form is simple and easy to understand, which facilitates the system to quickly execute reasoning and can output the corresponding charging and discharging adjustment rules in a timely manner according to environmental changes, improving the practicality and efficiency of the response model.

[0039] In a preferred embodiment of the present invention, the modeling strategy divides environmental parameters into multiple feature groups and corresponds them to charging and discharging scenarios through classification and clustering. Critical threshold parameters under each scenario are extracted to construct an association matrix. Then, a rule-based reasoning algorithm generates a set of charging and discharging response rules as the response model, forming a complete modeling logic of "environmental grouping - threshold extraction - matrix association - rule generation". This technical solution enables the response model to accurately associate environmental parameters with the battery charging and discharging safety boundary. The technical effect is that it breaks through the limitations of traditional single-threshold judgment of environmental factors, achieving targeted adaptation of charging and discharging strategies under different environmental combinations. This provides a more realistic decision-making basis for subsequent charging parameter adjustment and improves the response accuracy of the charging strategy to complex environmental changes.

[0040] Preferably, the mapping algorithm includes: constructing a multi-dimensional feature set based on evaluation results, environmental parameters, and a battery aging feature library; and performing coupling operations on the multi-dimensional features in the multi-dimensional feature set through a feature fusion algorithm to obtain associated features.

[0041] Further preferably, the construction of the multi-dimensional feature set includes: performing feature decomposition on the battery health status parameters in the evaluation results to extract basic feature vectors characterizing the internal physicochemical properties of the battery; encoding environmental parameters by type to generate real-time environmental feature vectors; using the battery aging feature library, the current health status, and the real-time environmental feature vectors as inputs, and outputting a critical damage threshold through a preset safety boundary prediction model; and fusing the basic feature vectors, environmental feature vectors, and critical damage thresholds to form a multi-dimensional feature set.

[0042] Specifically, the core of the mapping algorithm is to establish a quantitative relationship between the evaluation results, environmental parameters and critical damage threshold through feature correlation analysis. The coupling operation adopts a feature fusion algorithm based on the attention mechanism, which assigns weights to each feature in the multi-dimensional feature set, and then generates associated features through weighted summation. The advantage is that it highlights the influence of key features on the critical damage threshold and improves the accuracy of feature representation.

[0043] Specifically, the basic feature vector includes five core dimensions: remaining electrode active material, electrolyte conductivity, membrane porosity, electrode expansion rate, and interfacial impedance. These features are extracted from battery parameters through electrochemical impedance spectroscopy analysis. Environmental parameters are converted into binary vectors through unique thermal encoding to ensure that environmental features can be directly identified by the model. The safety boundary prediction model takes the current health status data of the battery, the battery cycle data in the battery aging feature library, and the real-time environmental feature vector as input. It adopts an LSTM time-series prediction algorithm with an attention mechanism to dynamically output the safety boundary thresholds of various parameters in the future, such as temperature critical values ​​and voltage upper limits, by extracting key patterns such as "temperature-voltage coupling features" and "cycle aging cumulative effects". This provides a threshold benchmark for safety margin calculation and avoids safety risks caused by insufficient adaptability of static thresholds from the source.

[0044] The mapping algorithm provided by the preferred embodiment of the present invention exhibits significant advantages through multi-dimensional feature fusion and precise quantitative association: the coupling operation based on the attention mechanism can dynamically allocate weights according to the importance of features, strengthen the influence of key factors such as health status and environment on the critical damage threshold, and avoid interference from irrelevant features; the basic feature vector extracted by electrochemical impedance spectroscopy accurately captures the internal physicochemical characteristics of the battery, and the multi-dimensional feature set constructed by combining the environmental features of one-thermal encoding and the safety boundary predicted by deep learning is comprehensive and detailed, laying a high-quality foundation for the generation of associated features, and ultimately enabling the associated features to accurately characterize the complex relationship between battery parameters, environment and critical damage threshold, thereby improving the scientificity and safety of charging strategy correction.

[0045] like Figure 4 As shown, preferably, each health level corresponds to a baseline charging power range, and the health-power matching rule includes: for each health level, determining the adjustment step size through a preset calculation rule; when the battery health level is in a preset low-level range, adjusting the maximum charging power to a preset proportion of the upper limit of the baseline charging power range corresponding to that health level, and using the adjusted maximum charging power as a constraint condition for charging power adjustment; based on the constraint condition, combined with the baseline charging power range corresponding to the health level and the adjustment step size, determining the charging power value for each time period through a preset power allocation formula to generate a charging strategy adapted to the current battery health state.

[0046] Specifically, the adjustment step size refers to the fixed adjustment range of the charging power each time, and the calculation rule can be expressed by the following formula:

[0047]

[0048] Where, ΔP step To adjust the step size, α is the baseline step size coefficient, and L mid L is the median value of the reference charging power range corresponding to the current health level. min L represents the upper limit of the power range for the lowest health level. max The maximum power range is defined as the highest health level, where β is the non-linear degradation exponent (β≥1), and K is the environmental impact coefficient, dynamically obtained from a preset temperature / humidity-degradation rate mapping table. This calculation rule uses a dual adjustment of health status score and environmental coefficient to increase the step size when the battery is healthy to improve charging efficiency, and decrease it when the battery is aging or in harsh environments to reduce damage risk. The lower-level range refers to the range of health levels 1-2. When the battery health level is in this range, the maximum charging power is set to a preset proportion of 70% of the upper limit. This design actively limits power to reduce the aging rate of the battery in vulnerable conditions. The power allocation formula is expressed as:

[0049]

[0050] Where P(t) represents the charging power value at time t, P max Indicates the adjusted maximum charging power; P base T represents the lower limit of the reference power range corresponding to the health level. cc λ represents the duration of the constant current charging phase, λ represents the decay coefficient of the constant voltage phase (obtained based on the performance decay curves in the battery aging feature library), and SOH represents the real-time health status value.

[0051] For example, during the operation of the electric vehicle dynamic charging optimization management system, when the health status assessment module determines that the current battery health level is level 3, corresponding to SOH = 60% and a reference charging power range of [0.3C, 0.5C], the strategy generation unit first calculates the adjustment step size based on the health-power matching rule: taking the median value L of the power range for health level 4. mid =0.4C, K=0.7, minimum health level 3, maximum health level L min =0.5C, maximum level 5, upper limit L max =1C, the adjustment step size ΔP is calculated using the formula of the calculation rule. step ≈0.014C, where α=0.2C, β=1.2; subsequently, the maximum charging power is adjusted to 70% of the upper limit of the benchmark for this level, 0.5C, i.e., P. max=0.35C, and finally the charging curve is generated through the power distribution formula. Constant current stage: P(t)=min(0.35,0.3+0.014×t / 20)(T cc =20min), if t=10min, P=0.307C, constant pressure stage: λ=0.08, P(t)=0.35×e -0.08(t-20) ×0.7, when t=30min, P=0.12C, when t=40min, P=0.05C.

[0052] In a preferred embodiment of the present invention, the health-power matching rule quantifies the battery health state into discrete health levels and binds them to a benchmark charging power range, achieving for the first time a precise match between health state and charging strategy. Its calculation rule introduces a nonlinear decay exponent β and an environmental correction factor K, causing the adjustment step size to shrink exponentially as the health level decreases and automatically reduce in high temperature / high humidity environments, effectively suppressing the charging risk of aging batteries. It innovatively forces the maximum power derating to be triggered at low health levels, and combines the segmented power allocation formula to form a four-fold protection mechanism of "health grading - step size adaptation - low-level hard constraint - dual-stage optimization", which greatly improves the battery cycle life compared to traditional solutions.

[0053] Preferably, the adjustment strategy includes: calculating the deviation between the real-time rate of change of environmental parameters and the average rate of change within a preset time window, and determining whether to update the strategy based on the deviation; when updating the strategy, selecting target parameters that match the current environmental parameters from the benchmark adjustment parameter table, and extracting the critical damage threshold from the associated features; calculating a safety margin index based on the difference between the real-time monitored battery state data and the critical damage threshold using preset quantization rules, and determining the parameter correction coefficient; correcting the target parameters based on the parameter correction coefficient and the critical damage threshold using preset correction rules; adjusting the benchmark adjustment parameter table according to the corrected target parameters, and updating the charging strategy according to the adjusted benchmark adjustment parameter table.

[0054] More preferably, the correction rule includes: correcting the target parameter using a preset correction function; if the corrected target parameter exceeds the corresponding critical damage threshold, then repeatedly correcting the target parameter using a gradient reduction method, and interrupting charging when the number of corrections reaches a preset threshold.

[0055] Specifically, the target parameters refer to the core charging parameters in the benchmark adjustment parameter table that match the current environmental parameters; the quantification rule is: calculate the safety margin index for each parameter separately and select the smaller index as the final safety margin index. The calculation of the safety margin index can be expressed as:

[0056]

[0057] Where M represents the safety margin index, T limit T represents the critical damage threshold. real This represents real-time battery status data. The parameter correction coefficient is set according to the safety margin classification: 1.1 when the safety margin index is ≥0.6, 1.0 when it is 0.3-0.6, and 0.8 when it is <0.3. The rate of change of environmental parameters refers to the amount of change of environmental parameters per unit time. The preset time window is 10 minutes; correction is initiated when the deviation value is >0.3. The correction function is a piecewise function, which can be expressed by the following formula:

[0058]

[0059] Among them, Q C Q represents the corrected target parameter. t Q represents the original value of the reference adjustment parameter table. S This represents the critical damage threshold, and the k-th generation parameter correction coefficient. Repeated corrections use a gradient reduction method, decreasing the parameter by 5% each time until the parameter is ≤ the critical damage threshold. For example, after the first correction, the target parameter is 4.3V, and the critical threshold is 4.2V. The second correction is 4.3 × 0.95 = 4.085V, which meets the requirements. The advantage of this adjustment strategy is that through multi-dimensional parameter verification and dynamic correction, it ensures that the charging parameters remain within a safe range under environmental changes and battery state fluctuations, thus improving the reliability of the strategy.

[0060] For example, when an electric vehicle is charging in a 42℃ high-temperature environment, the data acquisition module detects that the ambient temperature rises sharply from 25℃ to 45℃ within 10 minutes, at a rate of 2℃ / min. The system calculates the deviation between its real-time rate of change and the preset 10-minute window average rate of change of 1.0℃ / min, and the deviation is |2.0-1.0|=1.0>0.3, thus triggering a strategy update. The dynamic correction unit immediately selects the target parameter charging voltage threshold of 4.25V corresponding to the 45℃ environment from the benchmark adjustment parameter table and extracts the critical damage threshold of 4.2V from the weighted 7 associated feature output. Based on the difference between the real-time battery voltage of 4.15V and the critical threshold of 4.2V, the safety margin index is calculated using the formula, which is 0.119, less than the 0.3 threshold. Therefore, the parameter correction coefficient k=0.8. At the same time, the deviation value of the ambient temperature rise rate is detected as |1.7. - 1.0 | = 0.7 > 0.3, Q is calculated using a correction function based on k and the critical threshold of 4.2V. C =min(0.8×4.25,4.2)=3.4V. Since 3.4V is less than the critical damage threshold of 4.2V, the correction is stopped. The correction value of 3.4V is then updated to the voltage threshold item in the reference adjustment parameter table, and the charging strategy is updated according to the adjusted reference adjustment parameter table.

[0061] The preferred embodiment of this invention provides an adjustment strategy that dynamically senses sudden environmental changes and battery state variations to construct an adaptive safety protection system: when environmental parameters fluctuate drastically, the system immediately triggers a strategy update, accurately locking the target parameter from the benchmark parameter table and extracting the real-time safety threshold; based on the dynamic difference between battery operating data and the safety threshold, the risk margin is calculated, and the derating coefficient is automatically matched; a unique segmented correction mechanism is used to suppress parameters exceeding the standard and to preventively drate parameters that do not exceed the standard, while simultaneously updating the parameter library and optimizing the charging strategy; if the real-time state exceeds a new safety boundary, instantaneous fuse protection is applied. This achieves second-level response to drastic environmental changes, proactive battery risk interception, and a dynamic closed-loop safety control throughout the entire process, significantly improving response speed and thermal runaway protection capabilities compared to traditional solutions.

[0062] Preferably, the electric vehicle dynamic charging optimization management system further includes an adaptive optimization module, which is used to: collect new data generated during system operation, compare the new data with the data in the initial training dataset to obtain the difference value; when the difference value exceeds a preset threshold, merge the new data into the initial training dataset, and use an incremental training algorithm to update the battery health status assessment model and response model.

[0063] Specifically, the new data refers to the time-series dataset of environmental parameters and battery dynamic parameters collected in real time during system operation. The difference comparison is performed by calculating the Euclidean distance in the feature vector space and measuring the similarity between the new data and the historical data in the initial training dataset. When the feature vector of the new data deviates from the cluster center of the historical data by more than a preset threshold, it is judged as a significant difference. The incremental training algorithm refers to the online optimization of model parameters using the mini-batch gradient descent method while retaining the existing model knowledge. This includes: adding a transfer learning layer to the battery health status assessment model to prevent catastrophic forgetting; adaptively updating the covariance matrix of the safety boundary prediction model in the mapping algorithm; and dynamically tuning the correction rules in the adjustment strategy using a reinforcement learning reward function. Ultimately, this enables the system to continuously evolve in the process of battery aging and environmental adaptation.

[0064] In a preferred embodiment of the present invention, the adaptive optimization module forms a new dataset by collecting environmental and battery dynamic parameters during system operation in real time. It quantifies the degree of deviation from historical data based on the Euclidean distance in the feature vector space. When the deviation exceeds a preset threshold, an incremental training mechanism is triggered. Under the premise of ensuring that historical experience is not lost, the health assessment model is accurately adapted to the battery degradation curve, the mapping algorithm responds to environmental distribution drift, and the adjustment strategy fits the real-time operating conditions. Moreover, after each optimization, it is forced to pass the critical threshold verification and derating rule verification to ensure that the evolution process is absolutely safe. Ultimately, the charging management system can self-evolve throughout its entire life cycle.

[0065] The electric vehicle dynamic charging optimization management system provided by this invention breaks through the bottleneck of traditional static charging strategies by constructing a full-chain management system of "environmental perception - health assessment - dynamic correction - self-evolution": The data acquisition module captures environmental and battery parameters in real time, drives the migration strategy to classify training data according to battery chemical system and establish a feature mapping table, so that the health status assessment model can dynamically adapt to different battery types; The modeling strategy integrates environmental clustering and charge and discharge safety boundary to generate response models, supporting the accurate matching of benchmark parameters with operating conditions; The health assessment module, based on aging feature library and real-time data, extracts multi-dimensional correlation features containing critical damage thresholds through mapping algorithms to achieve quantitative early warning of battery damage risk; The charging strategy generation module pioneers a dual-track mechanism of "benchmark parameters + dynamic correction" - the strategy generation unit matches the power range according to the health level, and the dynamic correction unit uses the environmental change rate deviation as a trigger signal, combines safety margin indicators and critical damage thresholds to grade and correct parameters, and intercepts runaway risks through a gradient fuse mechanism; The adaptive optimization module continuously compares the differences between new and old data, incrementally trains and updates model parameters, so that the system can continuously evolve throughout the battery's entire life cycle. This solution achieves millisecond-level risk response, significantly reduces the probability of thermal runaway, and significantly extends battery cycle life.

[0066] The above description is merely a preferred embodiment of the technical solution of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A dynamic charging optimization management system for electric vehicles based on battery health status assessment, characterized in that, The electric vehicle dynamic charging optimization management system includes: The data acquisition module connects the environmental sensor and the battery management system of the electric vehicle to collect environmental and battery parameters. The data storage module stores a preset initial training dataset, a battery charge / discharge dataset, and a battery aging feature library. The model building module is used to build a battery health status assessment model based on the initial training dataset and a preset transfer strategy to assess the battery's health status; it is also used to build a response model based on the collected environmental parameters and the initial training dataset and a preset modeling strategy to adjust the charging strategy. The health status assessment module is used to assess the battery based on battery parameters using a battery health status assessment model and obtain assessment results. It is also used to extract related features based on the assessment results, environmental parameters, and battery aging feature library using a preset mapping algorithm to characterize the critical damage threshold of each battery parameter. The charging strategy generation module includes a strategy generation unit, a baseline parameter generation unit, and a dynamic correction unit. The strategy generation unit generates a charging strategy based on the evaluation results and through preset health-power matching rules. The baseline parameter generation unit performs coupled analysis of environmental parameters and battery charge / discharge datasets through the response model to generate a baseline adjustment parameter table. The dynamic correction unit is used to correct the parameters in the baseline adjustment parameter table according to environmental parameters and related characteristics, through a preset adjustment strategy, and update the charging strategy.

2. The electric vehicle dynamic charging optimization management system according to claim 1, characterized in that, The migration strategy has several preset difference intervals, and each difference interval corresponds to an adaptation rule. The migration strategy includes: Based on the chemical system of the battery, the initial training dataset is classified and a feature mapping table is established to represent the health characteristic benchmarks of different types of batteries. Based on the established feature mapping table, the distribution differences of various initial training datasets are calculated using the maximum mean difference algorithm, and the preset adaptation rules corresponding to the model parameters are determined based on the distribution differences. The model parameters are dynamically adjusted using the adaptation rules to generate model parameter configurations that are adapted to different types of batteries, thereby constructing a battery health status assessment model.

3. The electric vehicle dynamic charging optimization management system according to claim 2, characterized in that, The evaluation results include a battery health status score and a health level. Each health level corresponds to a preset score range. The battery health status evaluation model evaluates the battery, including: Feature fusion is performed on the input battery parameters to extract multi-dimensional feature vectors characterizing battery capacity decay, internal resistance growth, and polarization characteristics. Based on the feature mapping table and adaptation rules, multi-dimensional feature vectors are mapped to a unified health assessment space to generate health status feature values. Based on the health status characteristic values, the battery health status score is calculated through regression analysis, and the health level is determined according to the score range.

4. The electric vehicle dynamic charging optimization management system according to claim 1, characterized in that, The modeling strategy includes: The initial training dataset and environmental parameters are classified and clustered, and the environmental parameters are divided into several environmental feature groups, with each environmental feature group corresponding to a set of charging and discharging scenarios; For each environmental feature group, the critical threshold parameters for battery charging and discharging under that scenario are extracted from the initial training dataset; Based on each environmental feature group and its corresponding critical threshold parameters, a correlation matrix between environmental parameters and battery charge / discharge safety boundaries is constructed. The correlation matrix is ​​iteratively calculated using a pre-defined rule-based reasoning algorithm to generate a set of charging and discharging response rules under different combinations of environmental parameters, which serves as the response model.

5. The electric vehicle dynamic charging optimization management system according to claim 1, characterized in that, The mapping algorithm includes: Based on the evaluation results, environmental parameters, and battery aging feature library, a multi-dimensional feature set is constructed. By using a feature fusion algorithm, the multi-dimensional features in the multi-dimensional feature set are coupled to obtain associated features.

6. The electric vehicle dynamic charging optimization management system according to claim 5, characterized in that, The construction of the multi-dimensional feature set includes: The battery health status parameters in the evaluation results are decomposed to extract the basic feature vectors that characterize the internal physicochemical properties of the battery. Environmental parameters are encoded by type to generate real-time environmental feature vectors; Using the battery aging feature library, current health status, and real-time environmental feature vectors as input, the system outputs the critical damage threshold through a preset safety boundary prediction model. The basic feature vector, environmental feature vector, and critical damage threshold are fused to form a multi-dimensional feature set.

7. The electric vehicle dynamic charging optimization management system according to claim 3, characterized in that, Each health level corresponds to a baseline charging power range, and the health-power matching rules include: For each health level, the adjustment step size is determined through preset calculation rules; When the battery's health level is in the preset low level range, the maximum charging power is adjusted to a preset ratio of the upper limit of the benchmark charging power range corresponding to that health level, and the adjusted maximum charging power is used as a constraint condition for charging power adjustment. Based on the constraints, combined with the baseline charging power range and adjustment step size corresponding to the health level, the charging power value for each time period is determined through a preset power allocation formula to generate a charging strategy that adapts to the current battery health status.

8. The electric vehicle dynamic charging optimization management system according to claim 1, characterized in that, The adjustment strategy includes: Calculate the deviation between the real-time rate of change of environmental parameters and the average rate of change within a preset time window, and determine whether to update the strategy based on the deviation. When updating the strategy, target parameters that match the current environmental parameters are selected from the baseline adjustment parameter table, and critical damage thresholds are extracted from the associated features. Based on the difference between real-time monitored battery status data and critical damage threshold, a safety margin index is calculated using preset quantification rules, and parameter correction coefficients are determined. Based on the parameter correction coefficient and the critical damage threshold, the target parameters are corrected according to the preset correction rules. The reference adjustment parameter table is adjusted based on the revised target parameters, and the charging strategy is updated based on the adjusted reference adjustment parameter table.

9. The electric vehicle dynamic charging optimization management system according to claim 8, characterized in that, The correction rules include: The target parameters are corrected using a preset correction function; If the corrected target parameter exceeds the corresponding critical damage threshold, the target parameter is repeatedly corrected using the gradient reduction method, and charging is interrupted when the number of corrections reaches a preset threshold.

10. The electric vehicle dynamic charging optimization management system according to claim 1, characterized in that, The electric vehicle dynamic charging optimization management system also includes an adaptive optimization module, which is used for: Collect new data generated during system operation and compare the new data with the data in the initial training dataset to obtain the difference value; When the difference exceeds the preset threshold, the new data is incorporated into the initial training dataset, and the battery health status assessment model and response model are updated using an incremental training algorithm.

Images