Power dynamic adjusting method based on battery quality, charging pile and system thereof

Abstract

The invention relates to a power dynamic adjustment method based on battery quality, a charging pile and a system. The method comprises the following steps: acquiring power utilization information of a battery, calling test charging current and collecting test charging information, so as to realize comprehensive collection of a battery basic state and a test response state; the two types of information are substituted into a battery characteristic model, core characteristic factors such as battery polarization, internal resistance, concentration difference and an SEI film can be systematically identified, the limitation that in the prior art, only surface electric quantity information is depended on is broken through, and deep perception of the internal state of the battery is achieved; the charging strategy instruction sequence realizes ordered switching through a trigger condition, the charging process is ensured to be progressive according to the battery state, and the response sub-strategy dynamically generates an adjustment parameter through actually measured charging information, so that the charging instruction can adapt to the battery state change in real time, and the rigid defect of a fixed charging strategy is avoided; the pertinence and the adaptability of a charging strategy are remarkably improved, the charging efficiency is guaranteed, the battery performance can be effectively protected, and the safety risk is reduced.

Description

Technical Field

[0001] This invention relates to the field of charging technology, and more specifically, to a method for dynamic power adjustment based on battery quality, a charging pile and its system. Background Technology

[0002] With the rapid development of the electric vehicle industry, charging piles, as core supporting facilities, are increasingly widely used. Their charging performance and battery compatibility directly affect the user experience, battery life, and charging safety of electric vehicles. Currently, most existing charging piles adopt a fixed charging strategy or simply refer to basic information such as the remaining battery capacity, lacking a comprehensive understanding and in-depth analysis of individual battery differences and dynamic states.

[0003] Specifically, existing technologies do not fully consider the differences in core battery characteristics, such as the fundamental attribute differences caused by different battery types and usage durations, as well as the dynamic waveform changes in battery temperature, power, current, and charge during charging. Furthermore, they lack effective means of identifying key internal battery characteristics (such as polarization state, internal resistance changes, concentration distribution, and SEI film state), resulting in charging strategies that cannot accurately match the actual battery state. This lack of targeted charging methods can easily lead to problems such as low charging efficiency, abnormal battery heating, and reduced cycle life, and may even pose safety hazards due to mismatches between charging parameters and battery characteristics.

[0004] Furthermore, existing technologies lack dynamic adaptation mechanisms to environmental factors, and the models used to determine battery status are mostly fixed settings, unable to achieve self-optimization through continuous sample accumulation and training. This makes it difficult to adapt to the diverse charging needs of batteries with different aging levels and types. Therefore, there is an urgent need for a method, charging station, and system capable of dynamically adjusting charging power based on the actual battery quality state to address the technical shortcomings of existing technologies, such as insufficient targeting, poor adaptability, and limited reliability of charging strategies. Summary of the Invention

[0005] In view of this, the first objective of the present invention is to provide a method for dynamic power adjustment based on battery quality.

[0006] To solve the above-mentioned technical problems, the technical solution of the present invention is: a power dynamic adjustment method based on battery quality.

[0007] Step S1: Obtain battery power consumption information;

[0008] Step S2: Retrieve the test charging current and obtain the test charging information based on the battery power consumption information;

[0009] Step S3: Input the test charging information and battery power consumption information into the preset battery feature model to obtain battery features, which include polarization feature factor, internal resistance feature factor, concentration feature factor and SEI film feature factor.

[0010] Step S4: Retrieve the charging strategy instruction sequence and corresponding response sub-strategy from the preset charging instruction library according to the battery characteristics. The charging strategy instruction sequence includes several sequentially set charging instructions. Each charging instruction includes a configured trigger condition. When the corresponding trigger condition is met, the next sequence of charging instructions is executed. The response sub-strategy is used to obtain measured charging information during the charging process and generate adjustment parameters based on the measured charging information to adjust the currently executed charging instruction.

[0011] Step S5: Generate charging current according to the charging command to complete the charging of the battery.

[0012] The advantages of this invention are as follows: By acquiring battery power consumption information, then retrieving the test charging current and collecting test charging information, a comprehensive acquisition of the battery's basic state and test response state is achieved, providing complete data support for subsequent accurate judgment of battery characteristics; by incorporating these two types of information into the battery feature model, the core characteristic factors such as battery polarization, internal resistance, concentration, and SEI film can be systematically identified, breaking through the limitation of existing technologies that only rely on surface charge information, and achieving a deep perception of the battery's internal state; the charging strategy instruction sequence achieves orderly switching through trigger conditions, ensuring that the charging process proceeds gradually according to the battery state, and the response sub-strategy dynamically generates adjustment parameters through measured charging information, enabling the charging instruction to adapt to changes in battery state in real time, avoiding the rigidity defects of fixed charging strategies; it significantly improves the targeting and adaptability of the charging strategy, ensuring charging efficiency while effectively protecting battery performance and reducing safety risks.

[0013] Furthermore, the battery power consumption information includes basic sub-information and monitoring sub-information. The basic sub-information includes battery type and usage duration, while the monitoring sub-information includes battery temperature waveform, battery power waveform, battery current waveform, and battery charge waveform. Step S1 also includes a preprocessing strategy. The preprocessing strategy calls the corresponding waveform recognition conditions based on the basic sub-information, determines the corresponding waveform window from the monitoring sub-information based on the waveform recognition conditions, and extracts the corresponding waveform based on the waveform window. By dividing the battery power consumption information into basic sub-information and monitoring sub-information, the basic sub-information clearly defines inherent attributes such as battery type and usage duration, while the monitoring sub-information covers dynamic data such as temperature, power, current, and charge waveforms, achieving a layered and comprehensive characterization of the battery state. Accurately locating the effective waveform window in the monitoring sub-information avoids interference from invalid data, ensuring that the extracted waveform data matches the battery type and usage state, and improving data quality. This provides accurate and effective data input for subsequent testing of charging current matching and analysis of battery characteristic models, further ensuring the accuracy of subsequent charging strategy formulation.

[0014] Further, step S2 includes a test matching strategy. This strategy involves retrieving the corresponding basic data format based on the basic sub-information, analyzing the detection sub-information using a waveform analysis algorithm to fill the corresponding data into the corresponding basic data format to determine missing data in the basic data format, and retrieving the corresponding test charging command from a preset test command library based on the missing data to generate the test charging current. This ensures the standardization and adaptability of data processing, providing a unified standard for subsequent data completion and analysis; accurately identifying missing items in the basic data format to avoid test deviations caused by incomplete data; and retrieving the corresponding test charging command from the test command library for missing data, making the generation of the test charging current more targeted, accurately supplementing missing information, and ensuring the completeness of the test charging information.

[0015] Furthermore, the battery feature model includes several identification nodes and a matching database. Each identification node is configured with corresponding identification conditions, and the identification nodes are connected by preset identification relationships. When the test charging information and battery power consumption information meet an identification condition, the corresponding identification node is activated. The matching database stores several battery features using the feature identification vector as an index. The battery feature model calculates the corresponding feature identification vector based on the activated identification node to generate the battery feature. The battery feature model improves identification efficiency through the setting of identification nodes and identification conditions. The preset identification relationships between identification nodes ensure the logical coherence and rationality of feature identification, avoiding feature deviations caused by isolated judgments. The matching database stores battery features using the feature identification vector as an index. Combined with the feature identification vector calculated by the activated node, it can quickly and accurately generate the corresponding battery feature, improving the consistency and reliability of battery feature identification and providing strong support for the accurate matching of charging strategies.

[0016] Furthermore, the response sub-strategy is configured with ideal charging information, and generates corresponding adjustment parameters based on the deviation between the measured charging information and the ideal charging information. The response sub-strategy presets ideal charging information, clearly defining the target benchmark for the charging process, providing a clear reference for the generation of adjustment parameters and avoiding blind adjustments. The targeted generation of adjustment parameters further improves the stability and adaptability of the charging process, effectively reducing battery damage caused by deviations in charging parameters and ensuring charging performance.

[0017] To achieve the second objective of this invention, a charging pile with dynamic power adjustment based on battery quality is provided. This charging pile is equipped with the aforementioned dynamic power adjustment method based on battery quality, including a controller, a charging execution module, and a communication module. The controller obtains BMS information from the vehicle communication system via the communication module to generate battery power consumption information, and executes the charging command through the charging execution module. The advantages of this invention are as follows: the controller obtains BMS information from the vehicle communication system via the communication module, ensuring that the source of battery power consumption information is direct and accurate, comprehensively reflecting the actual state of the battery; the coordinated cooperation between the controller and the charging execution module achieves precise transmission and execution of charging commands, enabling the dynamic adjustment method to be applied practically, improving the intelligence and adaptability of the charging pile, and meeting the personalized charging needs of different batteries.

[0018] Furthermore, the system also includes an environmental detection module. This module detects environmental information and generates macro-environmental constraint factors and dynamic environmental constraint factors based on that information. The controller further includes parameters for correcting battery characteristic factors using macro-environmental constraint factors and for correcting generated adjustment parameters using dynamic environmental constraint factors. The addition of the environmental detection module addresses the impact of environmental factors on battery charging, filling a gap in existing technologies that neglect environmental variables. The macro-environmental constraint factors correct battery characteristic factors, enabling battery feature identification to adapt to actual environmental conditions and avoiding feature judgment biases caused by environmental interference. The dynamic environmental constraint factors correct adjustment parameters, ensuring that charging command adjustments not only adapt to the battery's own state but also respond to environmental changes in real time, further improving the safety, stability, and reliability of the charging process.

[0019] To achieve the third objective of this invention, a power dynamic adjustment system based on battery quality is provided, including the aforementioned charging pile, and further including a model training subsystem for training the battery feature model. The model training subsystem includes an aging sample module, a measured sample module, a sample expansion module, a sample processing module, a model building module, and a model training module. The aging sample module is used to generate aging samples, the measured sample module is used to generate measured samples, the sample expansion module is used to generate extended samples based on the aging samples and measured samples, the sample processing module is configured with an annotation strategy, the annotation strategy configures a corresponding reliability value for each sample, the model building module is used to split the battery features into several recognition nodes and generate corresponding recognition conditions, and the model training module is used to input the samples into the battery feature model to correct the parameters of the recognition conditions and the parameters of the recognition relationship connections.

[0020] The advantages of this invention are as follows: The model training subsystem enables continuous optimization of the battery feature model, avoiding the decline in adaptability caused by model fixation, and allowing the model to continuously adapt to new battery types and states; the aging sample module and the measured sample module provide diverse basic samples, and the sample expansion module enriches the number and coverage of samples, providing sufficient data support for model training; the annotation strategy of the sample processing module assigns reliable values ​​to samples, enabling the screening of high-quality samples and avoiding the impact of low-quality samples on model accuracy; the model construction module splits the recognition nodes and generates recognition conditions, making the model structure clearer, and the model training module corrects relevant parameters, improving the accuracy of the model in judging battery features.

[0021] Furthermore: the aging sample module acquires measured information of different types of batteries to generate battery characteristics, and generates corresponding test charging information and battery power consumption information after charging test and power consumption test respectively to generate corresponding aging samples;

[0022] The measured sample module is configured with a feature estimation strategy. The feature estimation strategy obtains the measured charging information during charging of the charging pile to generate the battery features, and combines the test charging information, battery power consumption information and battery features corresponding to the battery to generate the corresponding aging sample.

[0023] The sample expansion module is equipped with a feature association network. This network includes associated feature points for each type of battery feature. Based on aging samples and measured samples, it configures test charging information and battery power consumption information for corresponding associated feature points, and generates extended samples based on the association relationships for unconfigured associated feature points. The aging sample module generates samples using measured information and charging / power consumption tests of different battery types, comprehensively reflecting the characteristics of batteries under different aging states and improving sample representativeness. The measured sample module's feature inference strategy generates samples based on actual charging scenarios at charging piles, ensuring a high degree of fit between the samples and actual application scenarios, thus improving the model's practicality. The feature association network in the sample expansion module generates extended samples through associated feature points, effectively filling the coverage gaps of the basic samples, enriching the diversity and comprehensiveness of the samples, providing a more comprehensive data source for model training, and further improving the model's generalization ability and judgment accuracy.

[0024] Furthermore, the endorsement strategy includes calculating and weighting the baseline reliability sub-value, reliability transitivity sub-value, attenuation reliability sub-value, type reliability sub-value, and influence reliability sub-value for each sample to obtain the reliability value. The baseline reliability sub-value reflects the reliability value of the sample's baseline information, the reliability transitivity sub-value reflects the reliability value obtained by the sample from other samples, the attenuation reliability sub-value reflects the reliability value of the sample over time, the type reliability sub-value reflects the reliability value of the sample's type, and the influence reliability sub-value reflects the reliability value generated by the influence of other samples on the sample.

[0025] The adjustment parameters generated by the model training module are proportional to the reliability values ​​of the corresponding samples. The annotation strategy obtains reliability values ​​by weighting five categories of reliability sub-values: baseline, reliable transmission, decay, type, and impact. This comprehensively considers various influencing factors of the samples, ensuring that the evaluation of sample reliability values ​​is scientific and comprehensive. The proportionality between the model training adjustment parameters and the sample reliability values ​​gives higher weight to high-quality samples in correcting model parameters and reduces interference from low-quality samples, thereby improving the accuracy and efficiency of model training. Through precise sample quality assessment and targeted parameter correction, the battery feature model can continuously output accurate battery feature judgments, providing a long-term reliable guarantee for the formulation of charging strategies. Attached Figure Description

[0026] Figure 1 This is a flowchart of a power dynamic adjustment method based on battery quality according to the present invention;

[0027] Figure 2This is a schematic diagram of the charging pile architecture.

[0028] Figure 3 This is a schematic diagram of the system architecture. Detailed Implementation

[0029] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, so that the technical solution of the present invention can be more easily understood and mastered.

[0030] Reference Figure 1 As shown, the execution flow of the dynamic charging method based on battery state is as follows:

[0031] Step S1, acquiring battery power consumption information, specifically includes three parts: basic sub-information collection, monitoring sub-information collection, and preprocessing strategy execution. Basic sub-information is directly read from the vehicle battery management system through a data interaction channel established by the charging pile's communication module. Battery type refers to the battery's chemical system category and specifications, and usage time refers to the cumulative operating time from the battery's initial use to the current charging operation. Monitoring sub-information is collected in real-time by dedicated sensing components configured in the charging pile. Battery temperature waveform refers to the continuous data curve showing the temperature change of the battery surface and internal core area over time within a preset sampling period; battery power waveform refers to the continuous data curve showing the battery's input or output power change over time; battery current waveform refers to the continuous data curve showing the magnitude and direction of the current flowing through the battery change over time; and battery charge waveform refers to the continuous data curve showing the percentage of remaining battery charge change over time. The collection of the above monitoring sub-information must be performed at a uniform sampling frequency to ensure data time synchronization. The preprocessing strategy execution process is as follows:

[0032] First, based on the battery type and usage duration in the collected basic sub-information, the preset waveform recognition condition mapping library is queried. This mapping library pre-stores exclusive waveform validity judgment standards and interference data filtering rules corresponding to different combinations of battery types and different usage durations. The waveform validity judgment standards are used to define the data range that can truly reflect the battery's working state, and the interference data filtering rules are used to eliminate abnormal data caused by sensor errors, external electromagnetic interference, and other factors.

[0033] Subsequently, the collected waveform data of each monitoring sub-information are compared point by point with the corresponding waveform recognition conditions to filter out continuous time segments that meet the valid judgment criteria and have had interference data removed. This continuous time segment is the waveform window. Finally, the waveform data within the waveform window are standardized by using a data smoothing algorithm to remove residual minor noise and converting the data into a unified data format, ultimately forming standardized waveform data for subsequent analysis and processing.

[0034] The data smoothing algorithm uses the moving average method, and its calculation formula is as follows:

[0035]

[0036] In the formula Indicates the number after standardization. Waveform data values ​​of each sampling point This indicates the number of sampling points in the moving average window. Indicates the first element in the original waveform data. The core of this formula is to eliminate the random error of a single sampling point by calculating the average of the data from adjacent sampling points, thus ensuring that the extracted waveform data can accurately reflect the true state of the battery.

[0037] Step S2, which retrieves the test charging current and obtains test charging information based on battery power consumption information, revolves around a test matching strategy. This strategy ensures that the test charging current effectively supplements battery status information through standardized data processing, accurate identification of missing data, and targeted test command invocation. Firstly, the basic data format is a standardized data storage framework pre-defined based on the electrochemical characteristics of different battery types and battery degradation patterns under different usage durations. Its field settings correspond one-to-one with the waveform characteristics of the monitoring sub-information, covering the storage locations of key characteristic parameters such as the average amplitude change rate, peak value, valley value, and duration of the waveform. The controller retrieves the matching basic data format from the pre-defined format database based on the battery type and usage duration obtained in step S1, ensuring the targeted and standardized nature of data processing. Next, a waveform analysis algorithm is executed to parse the monitoring sub-information preprocessed in step S1. The core steps of this algorithm include feature extraction, data matching, and validity verification. First, the key parameters of each waveform are calculated using the feature extraction formula, where the formula for calculating the waveform change rate is...

[0038] In the formula This represents the rate of change of the waveform over a certain time period. This indicates the change in waveform amplitude within that time period. This represents the corresponding time change; the formula for calculating the average amplitude of the waveform is:

[0039]

[0040] In the formula Indicates the average amplitude of the waveform. This indicates the total number of sampling points within the waveform window. Indicates the first The waveform amplitude of each sampling point is then calculated. Next, the calculated key feature parameters are matched one by one according to the field definitions of the basic data format, and the corresponding parameter values ​​are filled into the corresponding fields in the format to complete the data filling operation. Afterwards, data validity is validated using a preset validity judgment formula.

[0041] In the formula This indicates the efficiency of filling fields with basic data formats. This indicates the number of fields that have been populated and conform to the field data specification. Indicates the total number of fields in the basic data format, when When the preset effective threshold is not reached, the information corresponding to fields that are not filled or do not conform to the specifications is considered missing data. The controller uses this formula to accurately determine the type and quantity of missing data in the basic data format. The test instruction library is a preset set of instructions that stores test charging instructions corresponding one-to-one with various types of missing data. Each instruction includes key parameters such as the charging current change mode, application duration, and amplitude range. These parameters are optimized based on test data of different battery types and usage durations, enabling efficient acquisition of missing information without damaging the battery. Based on the determined type and quantity of missing data, the controller retrieves one or more corresponding test charging instructions from the test instruction library. The charging execution module combines these instructions into an actual test charging current and applies it to the battery. Simultaneously, the charging pile's sensing components collect real-time response data of the battery during the test charging process. This response data constitutes the test charging information, which includes dynamic changes in parameters such as temperature, power, current, and charge during the test charging process. Together with the battery power consumption information in step S1, this forms the complete data foundation for subsequent battery feature analysis.

[0042] Step S3 is the process of inputting test charging information and battery power consumption information into a preset battery feature model to obtain battery features.

[0043] The polarization characteristic factor is a parameter reflecting the strength of polarization caused by charge separation during battery charging, and it is directly related to the voltage response characteristics of the battery during charging. The internal resistance characteristic factor is a parameter related to the magnitude of the impedance that hinders the flow of current inside the battery, and it directly affects charging energy loss and heat generation. The concentration characteristic factor is a parameter reflecting the uniformity of electrolyte concentration distribution inside the battery, and it is related to ion migration efficiency and charging rate. The SEI film characteristic factor is a parameter reflecting the integrity and stability of the solid electrolyte interface film on the battery surface, and its state directly affects the cycle life and safety of the battery. The battery characteristic model is a model for identifying internal battery characteristics based on battery electrochemical principles and a large amount of sample data. It consists of several identification nodes and a matching database. The identification nodes are logical units in the model used to determine the correlation dimension of specific features. Each identification node corresponds to a clear feature correlation direction. For example, some nodes are specifically associated with the determination of polarization-related information, some nodes are specifically associated with the determination of internal resistance-related information, some nodes are associated with the determination of concentration distribution-related information, and some nodes are associated with the determination of SEI film state-related information. At the same time, some nodes are used for cross-validation of different types of information. The identification relationship connection is a pre-defined rule used to define the logical association between each identification node, including parallel, progressive, and causal relationships. Parallel relationships indicate that multiple identification nodes must simultaneously meet their respective identification conditions to participate in subsequent feature calculations. Progressive relationships indicate that the identification condition verification of the next identification node can only be triggered after the previous identification node is activated. Causal relationships indicate that the activation result of the previous identification node will dynamically adjust the identification condition parameter values ​​of the next identification node. The matching database is a dataset that pre-stores standard feature identification vectors and corresponding complete battery features. Each stored entry contains a standard feature identification vector and the corresponding standard description or value range of polarization feature factor, internal resistance feature factor, concentration feature factor, and SEI film feature factor. The specific execution steps are as follows: First, the battery power consumption information preprocessed in step S1 and the test charging information obtained in step S2 are fused. The time-synchronized parameters in the two types of information are correlated to form a fused dataset covering the battery's basic state test response state, ensuring data integrity and correlation.

[0044] Next, for each identification node in the battery feature model, data items related to the identification conditions of that node are extracted from the fusion dataset. The identification conditions of each identification node are set based on its associated feature dimensions. For example, the identification condition of the polarization-related identification node is whether the ratio of the current change to the corresponding voltage change in the test charging information conforms to the preset rule. The identification condition of the internal resistance-related identification node is whether the voltage-current ratio at different stages during the test charging process is within the preset range. The identification condition of the concentration-related identification node is whether the synergistic relationship between the battery charge change rate and the temperature change rate meets the preset requirements. The identification condition of the SEI film-related identification node is whether the fluctuation amplitude of the voltage stabilization stage during the test charging process conforms to the preset standard.

[0045] Following the connection order of the identification relationships between the identification nodes, each identification node undergoes identification condition verification. If the verification result meets the identification conditions of the node, the node is determined to be activated; otherwise, it remains inactive. For identification nodes with progressive relationships, their own identification conditions must be verified only after the preceding node is activated. For identification nodes with causal relationships, the parameters of their own identification conditions must be adjusted based on the activation results of the preceding node before verification. Then, the feature identification vector is calculated. The feature identification vector is a multi-dimensional vector formed by arranging the corrected weight values ​​of all identification nodes in a preset order. First, a basic weight value is preset for each identification node. The basic weight value is set according to the importance of the battery feature determination based on the feature association dimension corresponding to the node; the higher the importance, the larger the basic weight value.

[0046] The basic weight value is corrected by the node activation state coefficient. A node activation state coefficient of 1 indicates that the identification node is activated, and 0 indicates that the identification node is not activated. The formula for calculating the corrected weight value of a single identification node is as follows: ,in This represents the corrected weight value of the i-th identification node. This represents the preset basic weight value of the i-th identification node. This represents the activation state coefficient of the i-th recognition node. The corrected weight values ​​of all recognition nodes are combined sequentially according to a preset node arrangement to form a feature recognition vector. ,in This represents the total number of identification nodes in the battery feature model. Finally, corresponding battery features are matched from the matching database. A vector similarity calculation algorithm is used to find the standard feature recognition vector in the matching database that is most similar to the currently generated feature recognition vector. The vector similarity calculation formula is as follows:

[0047] ,in Indicates the currently generated feature recognition vector Matching the first in the database Standard feature recognition vectors similarity, Indicates the first Standard feature recognition vectors The Middle The formula quantifies the matching degree by calculating the cosine similarity between two vectors, with a similarity value closer to 1 indicating a higher degree of matching. Once the standard feature recognition vector with the highest similarity value is found, the complete battery feature corresponding to that vector is used as the battery feature of the current battery. If multiple standard feature recognition vectors have the same and highest similarity value, the common range of the battery features corresponding to these vectors is taken as the battery feature of the current battery. This process accurately yields the polarization characteristic factor, internal resistance characteristic factor, concentration characteristic factor, and SEI film characteristic factor of the current battery.

[0048] Step S4 retrieves the charging strategy instruction sequence and the corresponding response sub-strategy from a pre-set charging instruction library according to the battery characteristics and performs a dynamic adjustment process. The charging instruction library is a pre-constructed structured database that stores various charging strategy instruction sequences and corresponding response sub-strategies adapted to different battery characteristics. Each stored entry forms an associated mapping with a specific combination of battery characteristics, and its data is optimized based on a large number of battery charging experiments and electrochemical characteristic analyses. The charging strategy instruction sequence is an ordered set of charging operations designed to adapt to a specific battery state, consisting of several charging instructions arranged in a progressive relationship according to the charging stages. Each charging instruction contains core operation parameters such as the charging current adjustment mode, voltage control range, and charging duration, which are used to guide the specific output of the charging execution module. The trigger condition is a quantitative determination criterion for judging whether the current charging instruction has completed its mission and can be switched to the next instruction. Its setting is closely related to the battery characteristics and charging stages, and directly determines the progress rhythm of the charging instruction sequence. The response sub-strategy is a dynamic correction mechanism that supports the charging strategy instruction sequence, used to calibrate the parameter deviation in real-time during the charging process to ensure that the charging state always conforms to the actual needs of the battery. The ideal charging information is a set of target parameters preset based on the battery characteristics and industry charging technical specifications, covering the ideal temperature change range, ideal power fluctuation range, ideal current amplitude standard, ideal power increase rate, etc. at different charging stages, providing a clear benchmark for judging the charging state. The measured charging information is the dynamic parameters collected in real-time real-time through the charging pile sensing components during the charging process, including the measured battery temperature, measured charging power, measured charging current, measured power increase rate, etc., which correspond one-to-one with the parameter types of the ideal charging information and the collection frequency is consistent with the collection frequency of the monitoring sub-information in step S1 to ensure the data time synchronization. The deviation information is the quantification result of the difference between the measured charging information and the corresponding parameters of the ideal charging information, reflecting the deviation degree of the current charging state from the ideal state. The adjustment parameter is a parameter generated based on the deviation information for correcting the current charging instruction, which can directly act on the charging execution module to adjust the core charging output.

[0049] The specific execution steps are as follows: First, connect the polarization characteristic factor, internal resistance characteristic factor, concentration difference characteristic factor, and SEI film characteristic factor obtained in step S3, and perform normalization processing on these four types of battery characteristics to eliminate the influence of the dimension difference of different characteristic factors on the matching accuracy. The normalization calculation formula is

[0050] , where represents the normalization result of the j-th characteristic factor, represents the original value of the j-th characteristic factor, represents the minimum value of this characteristic factor in the associated data of the charging instruction library, This represents the maximum value of the feature factor in the associated data of the charging instruction database. This formula maps all feature factors to the same numerical range. Then, the normalized battery features are used as the retrieval basis for a matching query in the charging instruction database. The degree of fit between the normalized battery features and the feature ranges associated with each charging strategy instruction sequence in the database is calculated. The formula for calculating the degree of fit is...

[0051]

[0052] ,in This indicates the degree of fit between normalized battery characteristics and a given charging strategy instruction sequence. This represents the weight coefficient of the j-th feature factor. The weight coefficient is set according to the degree of influence of each feature factor on the adaptability of the charging strategy and satisfies the following conditions: This represents the center value of the characteristic interval associated with the charging strategy instruction sequence. This formula quantifies the proximity of each characteristic factor to the target interval and performs a weighted summation to accurately select the charging strategy instruction sequence with the highest fit and its corresponding response sub-strategy, which serves as the charging execution plan for the current battery. Next, the execution process of the charging strategy instruction sequence is initiated, executing the charging instructions sequentially according to their order. During the execution of each charging instruction, the key state parameters corresponding to that instruction are monitored in real time. These key state parameters are determined based on the core objectives of the charging instruction and may include the cumulative increase in battery capacity, peak battery temperature, and stable charging power duration. Simultaneously, the key state parameters are evaluated to determine whether they meet the trigger conditions configured for the current charging instruction. The trigger condition determination logic is as follows: when the key state parameter reaches a preset target threshold or meets a preset continuous state requirement, the trigger condition is considered met, the current charging instruction is automatically terminated, and the process switches to the next sequence of charging instructions. If the trigger condition is not met, the current charging instruction continues to be executed until the trigger condition is met or the preset maximum execution time of the instruction is reached, ensuring that the charging process proceeds in an orderly manner in stages and closely follows changes in battery state.

[0053] Throughout the execution of the charging strategy command sequence, a response sub-strategy is run synchronously for dynamic adjustments. First, the ideal charging information corresponding to the currently executed charging command is retrieved from the response sub-strategy. This ideal charging information is updated synchronously with the switching of charging commands, always adapting to the current charging stage. Then, the actual charging information is collected in real time through the charging pile's sensing components, ensuring that the collected parameter types completely correspond to the parameter types of the ideal charging information.

[0054] Next, the deviation information is calculated. Deviation quantification is performed for each pair of corresponding parameters. When the ideal parameter value is not zero, the deviation calculation formula is as follows: ,in Indicates the first Deviation information of each parameter, Indicates the first Measured values ​​of each parameter Indicates the first The ideal values ​​of each parameter; when the ideal parameter is zero, the deviation calculation formula is: To avoid calculation errors, the overall deviation is then calculated using the following formula: ,in This represents the combined deviation of all parameters. Indicates the first The deviation weights of each parameter are set according to the degree of influence of the parameter on charging safety and battery life, and satisfy the following: , This indicates the total number of parameters involved in the deviation calculation. Finally, adjustment parameters are generated based on the overall deviation. The formula for generating the adjustment parameters is as follows: ,in This indicates that the parameters are being adjusted. This represents the preset adjustment coefficient, which is set based on the battery type and the current charging stage. It is used to control the adjustment range and prevent sudden parameter changes. Represents a sign function, when When the value is 1, The value is -1, used to determine the adjustment direction. The generated adjustment parameters directly affect the currently executed charging command, correcting core operating parameters such as the charging current amplitude in real time. This ensures that the charging process always closely approximates the ideal charging state, effectively avoiding deviations in charging parameters caused by battery state fluctuations or external interference, and guaranteeing the safety, stability, and efficiency of charging.

[0055] Step S5: Generate charging current according to the charging command to complete the charging of the battery.

[0056] Reference Figure 2 As shown, in another embodiment, the charging pile is a charging device with battery status sensing, adaptive matching of charging strategies, and real-time power adjustment capabilities. This enables personalized dynamic adjustment of charging power based on the actual battery status, solving the problems of poor adaptability, battery damage, and high safety risks associated with traditional charging piles with fixed power charging. The hardware architecture of this charging pile mainly includes a controller communication module, a charging execution module, a sensing component environmental detection module, and a power supply module. Each module interacts with data and transmits commands through a preset communication protocol, forming a complete closed-loop control system.

[0057] The controller is the core decision-making unit of the charging pile. It adopts a high-performance microprocessor chip and has built-in storage unit, arithmetic unit and interface unit. The storage unit pre-stores core data and related algorithm programs such as waveform recognition condition mapping library, format database, test instruction library, battery feature model and charging instruction library. The arithmetic unit is responsible for executing various algorithms such as data preprocessing, waveform analysis, feature recognition, strategy matching and deviation calculation. The interface unit provides a standardized connection channel with other modules to ensure efficient transmission of instructions and data.

[0058] The communication module adopts a dual-mode communication architecture of vehicle Ethernet and CAN bus to establish a stable data link between the charging pile and the vehicle battery management system. Its communication protocol complies with the ISO 15118 standard and supports bidirectional transmission of basic sub-information, test charging command feedback data, etc. It also has a communication anomaly detection and reconnection mechanism to ensure the reliability and real-time performance of data transmission. Communication delay is controlled within a preset threshold to ensure timely response of charging strategies.

[0059] The charging execution module is the core execution unit for power output and regulation. It consists of a power conversion circuit, a current and voltage regulation unit, and a drive circuit. The power conversion circuit is responsible for converting the AC power input from the grid into the DC power required for battery charging. The conversion efficiency is optimized through a preset power factor correction algorithm. The current and voltage regulation unit uses PWM pulse width modulation technology to achieve precise regulation of charging current and voltage. Its regulation accuracy meets the requirements of the electrochemical characteristics of battery charging. The drive circuit receives the control signal output from the controller and drives the power devices to complete power output and dynamic adjustment according to the instructions. It also has fault protection functions such as overcurrent, overvoltage, and overheating. When an abnormal state is detected, it can quickly cut off the power output and feed back fault information to the controller. The sensing components are the core of state perception, consisting of temperature sensors, current sensors, voltage sensors, and charge sensors. The temperature sensors are distributed and installed on the battery surface, the core area of ​​the battery, and key parts of the charging execution module to collect multi-dimensional temperature data. Their measurement range and accuracy match the temperature change range during battery charging. The current and voltage sensors are connected in series or parallel in the charging circuit to collect dynamic data of charging current and voltage in real time. The sampling frequency is consistent with the acquisition frequency of the monitoring sub-information in step S1. The charge sensor communicates and coordinates with the vehicle battery management system to accurately obtain data on the remaining battery charge and the rate of charge change. All data collected by the sensing components are preprocessed through signal conditioning circuits, including filtering and amplification, to ensure the accuracy and stability of the data.

[0060] The environmental monitoring module is used to collect key parameters of the charging pile's operating environment, including ambient temperature, humidity, atmospheric pressure, and dust concentration. Its core components are an environmental parameter acquisition unit and a constraint factor generation unit. The environmental parameter acquisition unit acquires environmental data in real time through corresponding environmental sensors. The constraint factor generation unit generates macro-environmental constraint factors and dynamic environmental constraint factors based on the environmental data. The macro-environmental constraint factors are used to correct battery characteristic factors, while the dynamic environmental constraint factors are used to correct charging adjustment parameters. The formula for generating the macro-environmental constraint factors is as follows:

[0061] In the formula Indicates macro-environmental constraint factors. Indicates the first Measured values ​​of environmental parameters. Indicates the first Baseline values ​​for class environment parameters, Indicates the first Maximum allowed values ​​for class environment parameters Indicates the first Minimum allowed values ​​for class environment parameters, Indicates the first The weighting coefficients of the class environment parameters satisfy:

[0062] This formula quantifies the comprehensive impact of environmental factors on battery characteristics by standardizing and weighting various environmental parameters; the formula for generating the dynamic environmental constraint factor is as follows: In the formula Indicates dynamic environmental constraint factors. Indicates the first Rate of change of environmental parameters The weighting coefficient representing the rate of environmental change is used to quantify the impact of real-time changes in environmental parameters on charging adjustment parameters. The power supply module provides a stable operating power source for each module of the charging pile, including an AC-to-DC power unit and a backup power unit. The AC-to-DC power unit converts the AC power from the grid into the DC voltage required by each module. The backup power unit uses a lithium battery pack and automatically switches power when the grid power supply is interrupted, ensuring the normal operation of critical data storage and fault alarm functions of core modules such as the controller communication module. The working logic of the power dynamic adjustment charging pile is deeply integrated with the dynamic charging strategy process in steps S1 to S4. The communication module and sensing components complete the collection of battery power consumption information in step S1. The controller executes a test matching strategy to complete the retrieval of test charging current and the collection of test charging information in step S2. The controller's built-in battery feature model completes the identification of battery features in step S3. The charging execution module and controller work together to complete the execution and dynamic adjustment of the charging strategy command sequence in step S4. Simultaneously, the constraint factors generated by the environmental detection module provide real-time correction for battery feature identification and charging parameter adjustment, ensuring that the charging power adjustment always adapts to the battery state and environmental conditions, achieving a safe, efficient, and accurate charging process.

[0063] Reference Figure 3 As shown, in another embodiment, the power dynamic adjustment system is an integrated system that realizes accurate perception of battery status and adaptive regulation of charging power. Its core architecture consists of a power dynamic adjustment charging pile cluster model training subsystem, a host management subsystem, and a communication network. Each part realizes data interaction and command coordination through standardized communication protocols, which can adapt to the personalized charging needs of various types of electric vehicle batteries and improve the overall safety, efficiency, and intelligence level of the charging network.

[0064] The communication network adopts a dual-mode architecture that integrates 5G and industrial Ethernet. The 5G network is used to realize long-distance wireless data transmission between the charging pile and the upper management subsystem, ensuring the real-time data in mobile scenarios. The industrial Ethernet is used for high-speed wired data interaction between the model training subsystem and the upper management subsystem, ensuring the efficient transmission of a large number of training samples and model parameters. It also has data encryption and authentication functions to prevent data leakage or tampering. The upper-level management subsystem is the core of the system's overall scheduling. It consists of a data storage unit, a status monitoring unit, a strategy scheduling unit, and a fault early warning unit. The data storage unit adopts a distributed database architecture to store battery power consumption information, test charging information, charging execution data, and environmental parameters uploaded by all charging piles, supporting long-term storage and rapid retrieval of massive amounts of data. The status monitoring unit receives real-time operating status data from each charging pile, including charging power, battery temperature, charging progress, and module operating status, and displays this data in real-time through a visual interface. When abnormal data is detected, an early warning is triggered. The strategy scheduling unit dynamically allocates grid power resources based on the battery characteristics and charging needs of each charging pile to avoid power instability caused by local charging pile overload. The fault early warning unit classifies and identifies fault information uploaded by charging piles according to preset fault judgment rules, generates corresponding fault handling suggestions, and pushes them to the operation and maintenance terminal. It also records information such as the fault occurrence time, type, and handling process.

[0065] The model training subsystem is the core unit that ensures the continuous optimization of the battery feature model. It is used to solve problems such as the inability of a fixed model to adapt to differences in battery aging types. Its core components include an aging sample module, a measured sample module, a sample expansion module, a sample processing module, a model building module, and a model training module.

[0066] The aging sample module is used to generate battery sample data at different aging levels. The core implementation method combines accelerated aging testing with segmented data acquisition. The specific steps are as follows: First, select several groups of mainstream battery types, such as ternary lithium batteries and lithium iron phosphate batteries. Each group of batteries needs to complete initial performance calibration in advance, recording initial polarization characteristics, internal resistance characteristics, concentration characteristics, SEI film characteristics, and standard charge-discharge curves as baseline data. Then, build an accelerated aging test platform, which consists of a charge-discharge device, an environmental chamber, and a data acquisition device. The charge-discharge device must support switching between constant current, constant voltage, and constant power modes, with an output current range covering 0.1C to 2C and a voltage regulation accuracy of no less than 0.01V. The environmental chamber is used to control the test environment temperature to be stable at 45℃, simulating a high-temperature accelerated aging scenario. The data acquisition device is synchronized with the charge-discharge device, with a sampling frequency set to 10Hz. Next, accelerated aging cycles are performed. Each aging cycle includes three stages: charging, discharging, and resting. The charging stage uses a constant current / constant voltage mode, with the current set at 1C and the voltage set to the battery's rated voltage, until the charging current drops to 0.05C and charging stops. The discharging stage uses a constant current mode, with the discharging current set at 1C, until the battery voltage drops to the cutoff voltage. The resting stage lasts 30 minutes to ensure the battery's internal state stabilizes. Every 50 aging cycles constitute one aging stage. After each aging stage, the cycle is paused, and the battery undergoes full performance testing. Tests include polarization voltage testing at different currents, AC impedance testing, ion diffusion rate testing, and SEI film impedance testing. The polarization characteristic factor, internal resistance characteristic factor, concentration characteristic factor, and SEI film characteristic factor for that stage are calculated from the test data. Simultaneously, complete voltage, current, and temperature data for the last 10 charge-discharge cycles within that stage are collected to form sample data for that aging stage, and basic information such as battery type, number of aging cycles, and remaining capacity decay rate corresponding to the aging stage are labeled. Repeat the above cycle until the remaining battery capacity decays to 60% of the initial capacity, completing the sample collection for the entire aging cycle, and finally forming a complete aging sample set covering the four stages of initial state, mild aging, moderate aging, and severe aging.

[0067] The actual test sample module is used to collect sample data in actual charging scenarios. The core implementation method adopts a multi-dimensional data filtering and cleaning combined with precise labeling. The specific operation steps are as follows: First, an actual test data acquisition channel is established through the upper-level management subsystem to receive raw data uploaded by each charging pile in real time. The raw data must include three core data types: battery basic information, charging environment information, and charging process data. The battery basic information includes battery type, battery number, usage time, and cumulative number of charging times. The charging environment information includes ambient temperature, ambient humidity, and charging period. The charging process data includes real-time data of charging current, voltage, temperature, and power. The sampling frequency is consistent with the charging pile's sensing components at 10Hz. Then, data filtering is performed. The filtering criteria are set as follows: data integrity is not less than 95%, that is, the number of sampling points with missing data in a single charging record does not exceed 5%; the charging time is not less than 30 minutes to ensure coverage of the complete charging stage; and there are no abnormal shutdown or fault records during the charging process. Next, data cleaning was performed, using the Z-score method to remove outliers. The mean μ and standard deviation σ of each parameter sequence were calculated, and data points exceeding the range of μ±3σ were identified as outliers. Missing data corresponding to outliers were supplemented using linear interpolation. Simultaneously, timestamps were calibrated to ensure complete synchronization of time series for different parameters. Finally, sample annotation was performed. Based on the cleaned charging process data, polarization feature factors, internal resistance feature factors, concentration feature factors, and SEI film feature factors were automatically extracted using a feature calculation model. The input to the feature calculation model was the cleaned voltage, current, temperature, and charge data sequences, and the output was the quantified values ​​of the four feature factors. Charging effect parameters, including charging efficiency and maximum temperature rise during the charging process, were also annotated. This resulted in a complete measured sample for each sample, including input data, feature factors, and charging effect annotations. The samples were categorized into three scenarios based on the charging environment: room temperature charging, low temperature charging, and high temperature charging, and into constant current charging stage and constant voltage charging stage, achieving full coverage of scenarios and stages.

[0068] The sample expansion module employs a feature association network to increase the number and coverage of samples. The core implementation method combines feature mapping network construction based on battery electrochemical characteristics with interpolation expansion. The specific steps are as follows: First, a feature association network is constructed. The network uses a three-layer fully connected structure. The input layer contains the quantized values ​​corresponding to four feature factors, the hidden layer has 64 nodes, and the output layer contains the waveform features of key parameters during the charging process, including voltage change rate, current fluctuation amplitude, and temperature rise rate. Network training uses valid data from existing aging samples and measured samples as training samples. The input is the feature factors, and the output is the corresponding key parameter waveform features. Mean squared error is used as the loss function, and iterative training is performed using a gradient descent algorithm until the loss value is below 0.001. After training, the intrinsic correlation between feature factors is mined based on this network, determining core correlations such as a positive correlation between polarization feature factors and internal resistance feature factors, and a negative correlation between concentration feature factors and SEI film feature factors. Next, sample expansion is performed. For feature combination intervals that are insufficiently covered in the aging sample set and the measured sample set, expanded samples are generated using a combination of linear interpolation and random perturbation. Specifically, feature interpolation points are selected between two adjacent valid samples. Adjacent valid samples must have the same battery type and consistent trend in feature factor changes. The feature factor values ​​of the interpolation points are calculated using a linear interpolation formula, which is: In the formula This represents the value of a characteristic factor at the interpolation point. and These are the feature factor values ​​for two adjacent valid samples, respectively. and These represent the number of aging cycles or the charging time for two adjacent valid samples, respectively. The interpolation point corresponds to the number of aging cycles or the charging duration. To avoid excessive homogenization of the extended samples, a random perturbation within ±5% is applied to the interpolated feature factor values. The perturbation range is determined based on the feature fluctuation range of the actual samples. Finally, waveform data of key charging process parameters corresponding to the extended samples are generated through a feature association network to verify the effectiveness of the extended samples. The verification criterion is that the matching degree between the feature factors of the extended samples and the waveform data is not lower than the average matching degree of the effective samples. The matching degree is calculated using cosine similarity to ensure that the extended samples conform to the electrochemical characteristics of the battery, ultimately forming an expanded sample set covering more feature combination scenarios.

[0069] The sample processing module assigns a reliability value to each sample through an annotation strategy to screen for high-quality samples. The core of the annotation strategy is to obtain the reliability value by weighting the impact of the baseline reliability propagation attenuation type on the five types of reliability sub-values. The specific calculation formula is as follows: In the formula Indicates the reliable value of the sample. This represents the baseline reliability sub-value, which is set based on the accuracy level of the sample acquisition equipment. The higher the equipment accuracy, the larger the baseline reliability sub-value. This represents the reliable transmission sub-value, set based on the transmission stability of the sample data. It takes the highest value when there is no packet loss or error during transmission. This represents the reliability sub-value for battery degradation, which is set based on the battery aging level corresponding to the sample. The lower the aging level, the larger the reliability sub-value for battery degradation. The type reliability sub-value is set based on the completeness of the battery type corresponding to the sample. The highest value is taken when the battery type information is complete and the labeling is accurate. The reliability sub-value is determined based on the level of environmental interference during sample collection; the lower the environmental interference, the greater the impact on the reliability sub-value. These are the weighting coefficients for each reliable sub-value, satisfying... The weighting coefficients were obtained through calibration using a large amount of experimental data.

[0070] The model building module is used to split the identification nodes and generate identification conditions. Based on the correlation dimension of battery features, the model is split into polarization identification nodes, internal resistance identification nodes, concentration identification nodes, SEI film identification nodes, and cross-validation nodes. Initial identification conditions are set for each node, based on battery electrochemical theory and basic sample data. The model training module uses the gradient descent algorithm to correct the parameters of the battery feature model. The training process uses the reliability value of the samples as weights; samples with higher reliability values ​​have a greater impact on the correction of model parameters. The specific training steps are as follows: First, the battery power consumption information and test charging information from the sample data are used as input, and the labeled battery features are used as output to initialize the parameters of the battery feature model; then, the loss value between the predicted features output by the model and the true features labeled by the samples is calculated. The loss function calculation formula is as follows:

[0071]

[0072] In the formula This represents the total loss value. Represents the total number of samples. This represents the reliable value of the s-th sample. This represents the model prediction value of the j-th feature factor for the s-th sample. This represents the true labeled value of the j-th feature factor of the s-th sample; then, the model parameters are iteratively updated using the gradient descent algorithm until the loss value is lower than the preset threshold or the maximum number of training iterations is reached; finally, the trained and optimized model parameters are synchronized to the controllers of all charging piles through the upper-level management subsystem to achieve global update of the battery feature model.

[0073] The specific operation steps are as follows: First, the battery characteristics are decomposed into five core dimensions based on the core electrochemical reaction mechanism during battery charging: polarization characteristics, internal resistance characteristics, concentration characteristics, SEI film characteristics, and cross-validation. Each dimension corresponds to a type of identification node. Then, the identification nodes are further subdivided: the polarization characteristics dimension is subdivided into two identification nodes: current-dependent polarization nodes and voltage-response polarization nodes; the internal resistance characteristics dimension is subdivided into two identification nodes: ohmic internal resistance nodes and polarization internal resistance nodes; the concentration characteristics dimension is subdivided into two identification nodes: ion diffusion concentration nodes and concentration gradient change nodes; the SEI film characteristics dimension is subdivided into two identification nodes: SEI film integrity nodes and SEI film ion conduction nodes; and the cross-validation dimension is subdivided into two identification nodes: feature consistency verification nodes and environmental adaptability verification nodes. A total of 10 identification node types are obtained, and each node type can be further subdivided. Each node corresponds to a unique feature association direction. Next, initial identification conditions for each identification node are generated. The identification conditions are set using quantitative thresholds or trend judgment methods. Specifically, the setting method is based on the effective data of the aging sample set and the measured sample set, and the normal range of the corresponding parameters of each node is determined through statistical analysis. For example, the identification condition for the current-dependent polarization node is set as the polarization voltage change is between 0.05V and 0.15V when the test charging current jumps from 0.5C to 1C. This threshold is determined by statistically analyzing the polarization voltage change of all effective samples under the same current step and taking the range of the 95% confidence interval. The identification condition for the ion diffusion concentration node is set as the ratio of the battery charge growth rate to the temperature growth rate is stable between 0.8 and 1.2. This ratio range is determined by statistically analyzing the normal charging stage data of the effective samples. Finally, the identification relationship connections between identification nodes are defined. Parallel relationships apply to two nodes with the same feature dimension. For example, an ohmic internal resistance node and a polarization internal resistance node must simultaneously meet their respective identification conditions to activate subsequent calculations of the internal resistance feature dimension. Progressive relationships apply to related nodes with different dimensions. For example, a current-dependent polarization node and a voltage-response polarization node must be activated first before triggering the identification condition verification of the SEI film integrity node. Causal relationships apply to nodes affected by the environment. For example, the identification condition threshold of the environmental adaptability verification node is dynamically adjusted according to changes in ambient temperature. For every 10°C increase in temperature, the threshold range expands by 10%, ensuring that the identification conditions are adapted to the actual environment. The working collaborative logic of the power dynamic adjustment system is as follows: Each power dynamic adjustment charging pile completes battery status acquisition and preliminary charging strategy execution through its own module, and uploads the relevant data to the upper-level management subsystem in real time.

[0074] The upper-level management subsystem aggregates, stores, and monitors the data, while pushing massive amounts of charging data to the model training subsystem. The model training subsystem generates augmented processing samples based on this data, completes the training and optimization of the battery feature model, and feeds back the optimized model to the upper-level management subsystem.

[0075] The upper-level management subsystem synchronizes the optimized model to each charging pile, guiding the charging piles to update their charging strategies. At the same time, the upper-level management subsystem dynamically schedules power supply resources according to the charging needs of each charging pile, and the fault early warning unit promptly handles various faults to ensure the stable and efficient operation of the entire system. All parts work together to achieve precise dynamic adjustment of charging power, ensuring the charging safety and efficiency of batteries in different states.

[0076] Of course, the above are just typical examples of the present invention. In addition, the present invention may have many other specific embodiments. All technical solutions formed by equivalent substitution or equivalent transformation fall within the scope of protection claimed by the present invention.

Claims

1. A battery quality-based power dynamic adjustment method, characterized in that: Step S1, obtaining battery power information; Step S2, obtaining test charging current and test charging information according to battery power information; Step S3, inputting test charging information and battery power information into a preset battery characteristic model to obtain battery characteristics, the battery characteristics including polarization characteristic factors, internal resistance characteristic factors, concentration difference characteristic factors, and SEI film characteristic factors; Step S4, obtaining charging strategy instruction sequences and corresponding response sub-strategies from a preset charging instruction library according to battery characteristics, the charging strategy instruction sequences including a plurality of sequentially arranged charging instructions, each charging instruction including a trigger condition, when the corresponding trigger condition is met, the next sequence of charging instructions is executed, and the response sub-strategy is used to obtain measured charging information during the charging process and generate adjustment parameters according to the measured charging information to adjust the currently executed charging instruction; Step S5, generating charging current according to the charging instruction to complete the charging of the battery.

2. The method of claim 1, wherein: The battery power information includes basic sub-information and monitoring sub-information, the basic sub-information including battery type and use time, and the monitoring sub-information including battery temperature waveform, battery power waveform, battery current waveform, and battery power waveform; the preprocessing strategy in step S1 includes calling corresponding waveform recognition conditions according to the basic sub-information, determining corresponding waveform windows from the monitoring sub-information according to the waveform recognition conditions, and extracting corresponding waveforms according to the waveform windows.

3. The method of claim 2, wherein: Step S2 includes a test matching strategy, the test matching strategy including calling corresponding basic data formats according to the basic sub-information, analyzing and detecting the monitoring sub-information by a waveform analysis algorithm to fill corresponding data into the corresponding basic data formats to determine missing data in the basic data formats, and calling corresponding test charging instructions from a preset test instruction library according to the missing data to generate the test charging current.

4. The method of claim 1, wherein: The battery characteristic model includes a plurality of recognition nodes and a matching database, each recognition node being configured with a corresponding recognition condition, and the recognition nodes being configured with a preset recognition relationship connection line, when the test charging information and the battery power information meet a recognition condition, the corresponding recognition node is activated; the matching database stores a plurality of battery characteristics with the characteristic recognition vector as an index, and the battery characteristic model calculates a corresponding characteristic recognition vector according to the activated recognition node to generate the battery characteristics.

5. The power dynamic adjustment method based on battery quality as described in claim 1, characterized in that: The response sub-strategy is configured with ideal charging information, and the response sub-strategy generates corresponding adjustment parameters according to the deviation information of the measured charging information and the ideal charging information.

6. A power dynamic adjustment charging pile based on battery quality, configured with a power dynamic adjustment method based on battery quality according to any one of claims 1-5, characterized in that: The controller, the charging execution module, and the communication module are included, the controller obtains BMS information from the vehicle-mounted communication system through the communication module to generate the battery power information, and executes the charging instruction through the charging execution module.

7. The power dynamic regulation charging pile based on battery quality according to claim 6, characterized in that: The environment detection module is configured to detect environment information and generate a macro-environment constraint factor and a dynamic environment constraint factor according to the environment information, and the controller is further configured to correct the battery characteristic factor by the macro-environment constraint factor and correct the generated adjustment parameter by the dynamic environment constraint factor.

8. A battery quality based power dynamic adjustment system comprising the charging station as claimed in any one of claims 5-6, characterized in that: The model training subsystem is further configured to train the battery characteristic model, and the model training subsystem comprises an aging sample module, a measured sample module, a sample expansion module, a sample processing module, a model construction module and a model training module. The aging sample module is configured to generate aging samples, the measured sample module is configured to generate measured samples, the sample expansion module is configured to generate expanded samples according to the aging samples and the measured samples, the sample processing module is configured with a signing strategy, the signing strategy is configured to assign a corresponding reliability value to each sample, the model construction module is configured to split the battery characteristic into a plurality of identification nodes and generate corresponding identification conditions, and the model training module is configured to bring the samples into the battery characteristic model to correct parameters of the identification conditions and parameters of identification relationship connections.

9. A battery quality based power dynamic adjustment system as claimed in claim 8, wherein: The aging sample module acquires different types of battery measured information to generate battery characteristics, and generates corresponding test charging information and battery power consumption information after a charging test and a power consumption test, respectively, to generate corresponding aging samples. The measured sample module is configured with a characteristic inference strategy, which acquires measured charging information when a charging pile is charging to generate the battery characteristic, and generates corresponding aging samples by using the battery corresponding test charging information and battery power consumption information and the battery characteristic. The sample expansion module is configured with a feature correlation network, which includes relevant correlation feature points for each type of battery characteristic. The test charging information and the battery power consumption information of the corresponding correlation feature points are configured according to the aging samples and the measured samples, and the expanded samples are generated according to the correlation relationship at the unconfigured correlation feature points.

10. A battery quality based power dynamic adjustment system as claimed in claim 9, wherein: The signing strategy comprises calculating and weighting a baseline reliability sub-value, a reliable transmission sub-value, a decay reliability sub-value, a type reliability sub-value and an influence reliability sub-value of each sample to obtain the reliability value, respectively. The baseline reliability sub-value reflects the reliability value of the baseline information of the sample, the reliable transmission sub-value reflects the reliability value obtained from other samples, the decay reliability sub-value reflects the reliability value over time, the type reliability sub-value reflects the reliability value of the sample type, and the influence reliability sub-value reflects the reliability value generated by the influence of other samples on the sample. The adjustment parameter generated by the model training module is proportional to the reliability value of the corresponding sample.

Images