Intensive power supply controller for charging pile and automobile BMS

Abstract

The invention relates to the technical field of electric vehicle power supply control, and discloses a charging pile and vehicle BMS intensive power supply controller. The system comprises an electric energy demand modeling module, a multi-source integration module, a dynamic adjustment module, a safety boundary judgment module and an optimized output module. The electric energy demand modeling module obtains battery parameters transmitted by the BMS, analyzes voltage and current thresholds, extracts characteristic quantities to generate a battery demand sequence and constructs an electric energy demand characteristic template; the multi-source integration module collects output parameters of the charging pile and environmental data, matches characteristic quantities and then fuses the parameters to generate an integrated parameter sequence; the dynamic adjustment module calculates a charging stage division threshold, identifies a power boundary condition and generates a dynamic regulation and control instruction set; the safety boundary judgment module monitors output fluctuation and a BMS feedback signal, and compares a safety threshold value to generate a boundary state identifier; and the optimization output module adjusts output waveform parameters according to the identifier, matches the battery demand change trend, generates an optimization power supply strategy, and adapts to the dynamic demand of the battery.

Description

Technical Field

[0001] This invention relates to the field of electric vehicle power supply control technology, specifically to a centralized power supply controller for charging piles and vehicle BMS. Background Technology

[0002] With the rapid development of the electric vehicle industry, charging piles, as the core infrastructure for electric vehicle energy replenishment, directly affect battery charging efficiency, lifespan, and safety during the charging process through their coordinated power supply performance with the vehicle's battery management system (BMS). Currently, most charging piles on the market have relatively simple interaction methods with the BMS, often using fixed parameter output modes, and are unable to dynamically adjust the power supply strategy according to the real-time status of the battery.

[0003] During actual charging, battery parameters change with factors such as charging time and ambient temperature. For example, when the ambient temperature is too low, battery activity decreases. If charging is still performed using the conventional current and voltage, it will not only prolong the charging time but may also lead to lithium plating inside the battery, affecting its lifespan. Conversely, when the battery is in a high-charge state, continuing to charge with a high current can easily cause the battery to overheat, posing a safety hazard. In existing technologies, some charging piles have basic parameter adjustment functions, but they can only adjust based on a single battery voltage or current signal, lacking integrated analysis of multi-source data. This makes it difficult to accurately construct a battery energy demand model, resulting in a mismatch between power supply parameters and the actual battery needs.

[0004] The existing communication protocols between charging piles and the Battery Management System (BMS) are mostly general standard protocols, lacking customized designs for centralized power supply control. This results in data transmission delays or packet loss, further affecting the timeliness and accuracy of power supply control. Simultaneously, in terms of safety boundary determination, existing technologies typically use fixed safety thresholds, failing to dynamically adjust the safety range based on battery state changes and environmental factors. When the battery exhibits abnormal conditions, it is difficult to make timely and accurate judgments and responses, increasing safety risks during charging. These problems severely restrict the performance improvement of electric vehicle charging systems, failing to meet users' demands for efficient and safe charging, and hindering the further development of the electric vehicle industry. Summary of the Invention

[0005] The purpose of this invention is to provide an integrated power supply controller for charging piles and vehicle BMS to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides a centralized power supply controller for charging piles and vehicle BMS, the system comprising: The power demand modeling module acquires battery parameter data transmitted by the vehicle's BMS, analyzes battery voltage and current thresholds, extracts multiple sets of battery cell state features, generates a battery demand sequence based on the order of appearance of the features, and constructs a power demand feature template. Based on the power demand feature template, the multi-source integration module collects charging pile output parameters and environmental monitoring data, matches feature quantities in the battery demand sequence, and fuses multi-source data to generate an integrated parameter sequence. The dynamic adjustment module calculates the charging stage division threshold based on the integrated parameter sequence, identifies the power boundary conditions of different charging sections, and generates a dynamic control instruction set. The safety boundary determination module, based on the dynamic control instruction set, monitors the output fluctuation characteristics of the charging pile and the BMS feedback signal, and generates a boundary status identifier by comparing the safety threshold range. The optimization output module adjusts the output waveform parameters of the charging pile according to the boundary state identifier, matches the characteristic quantity change trend in the battery demand sequence, and generates an optimized power supply strategy.

[0007] Preferably, the power demand feature template includes a battery cell state coding structure, a feature quantity time sequence arrangement set, and voltage and current weight parameters; the integrated parameter sequence includes multi-source data mapping relationship, feature matching result set, and time sequence alignment parameters; the dynamic control instruction set includes stage division node number, power boundary condition set, and control parameter index; the boundary state identifier includes safety threshold comparison result, fluctuation feature classification identifier, and abnormal state code; and the optimized power supply strategy includes output waveform adjustment parameters, feature quantity matching relationship table, and power supply mode switching instruction.

[0008] Preferably, the power demand modeling module includes: The feature extraction submodule acquires battery parameter data transmitted by the vehicle BMS, performs feature parsing operations on the parameter data, extracts the feature set of each battery cell, records the temporal position of each feature in the data stream, compares the relationship between the first occurrence position of key features in the parameter sequence and the data length, classifies them according to battery cell type, and obtains the feature distribution results. The sequence construction submodule extracts the data segments corresponding to the key features in the battery parameter data based on the feature distribution results, truncates the data based on the time interval of the feature in the data stream, constructs a feature fragment set based on the position of the data truncated for each feature, and reassembles it with the battery unit to which the feature belongs to obtain the feature time series set. The template generation submodule calculates the weight parameter values ​​of all features based on the time series set of features, performs time series rearrangement processing on the feature segment set based on the original order of features in the battery parameter data, and splices the time series segments of multiple features in the same battery cell according to the first occurrence position to obtain the power demand feature template.

[0009] Preferably, the multi-source integration module includes: Based on the power demand characteristic template, the data acquisition submodule combines the charging pile output parameter set and the environmental monitoring data stream to perform time-series alignment of all data sources according to their acquisition timestamps, establish a time-series alignment parameter table, and obtain multi-source data time-series index values. The fusion processing submodule obtains the aligned multi-source data set based on the multi-source data time series index value, performs feature matching operation on the multi-source data of each time series point, generates the integrated parameter value of each time series point using the feature weighted fusion method, and integrates the parameter values ​​of all time series points to generate an integrated parameter sequence. The verification feedback submodule compares the feature matching results with the feature quantity distribution in the original power demand feature template based on the integrated parameter sequence, generates data quality evaluation indicators, and feeds them back to the fusion processing submodule.

[0010] Preferably, the dynamic adjustment module includes: The stage division submodule analyzes the characteristic patterns of the battery charging stage based on the integrated parameter sequence, divides the charging stage intervals according to the characteristic pattern change nodes, marks the boundary time sequence point numbers of each stage, and generates a stage division node list. The power calculation submodule divides the node list according to the stage, collects the extreme values ​​of power demand in each stage interval, calculates the power transition threshold between adjacent stages, and generates a set of power boundary conditions in combination with the battery rated parameters. The instruction generation submodule matches the output capability parameters of the charging pile with the set of power boundary conditions, generates an index table of control parameters for each stage, and integrates the control parameters of all stages to generate a dynamic control instruction set.

[0011] Preferably, the security boundary determination module includes: Based on the dynamic control instruction set, the signal monitoring submodule collects the output waveform characteristic parameters of the charging pile and the battery status data fed back by the BMS in real time, and extracts the waveform fluctuation characteristic quantity and battery abnormal characteristic code. The threshold comparison submodule, based on the waveform fluctuation feature quantity and the battery abnormal feature code, calls the preset safety threshold parameter table, performs a step-by-step comparison operation between the feature quantity and the threshold parameter, and generates a feature state comparison result set. The status identifier generation submodule classifies the abnormality level of the feature status based on the feature status comparison result set, and generates boundary status identifiers according to the abnormality level encoding.

[0012] Preferably, the optimized output module includes: The strategy generation submodule analyzes the abnormal state distribution pattern based on the boundary state identifier and generates an output waveform adjustment parameter table by combining the characteristic quantity change trend in the power demand characteristic template. The output control submodule adjusts the driving parameters of the charging pile power output device based on the output waveform adjustment parameter table to match the power boundary conditions corresponding to the current charging stage. The adaptive adjustment submodule monitors the matching degree between the output waveform and the BMS feedback signal, and dynamically updates the output waveform adjustment parameters based on the matching degree deviation.

[0013] Preferably, the strategy generation submodule includes: The pattern analysis unit identifies the time interval and spatial distribution characteristics of the abnormal state distribution based on the abnormal state code in the boundary state identifier, and establishes an abnormal state distribution map. Based on the abnormal state distribution map, the parameter mapping unit queries the set of feature quantities for the corresponding time interval in the power demand feature template and establishes a mapping relationship table between abnormal states and feature quantities. The adjustment strategy unit generates a combination of output waveform adjustment parameters for different abnormal states based on the mapping table.

[0014] Preferably, the output control submodule includes: The drive parameter calculation unit converts the adjustment parameters into drive signal parameters of the power device based on the output waveform adjustment parameter table, including frequency modulation parameters and duty cycle adjustment parameters; The real-time adjustment unit dynamically adjusts the output value of the drive signal parameters according to the power boundary conditions corresponding to the current charging stage; The protection execution unit monitors the deviation between the actual output of the power device and the drive signal parameters, and initiates protective adjustment when the deviation exceeds the tolerance value.

[0015] Preferably, the adaptive adjustment submodule includes: The matching degree calculation unit collects the output waveform characteristics of the charging pile and the battery response data fed back by the BMS in real time, and calculates the matching degree index between the waveform characteristics and the battery response. The deviation analysis unit identifies the time interval and characteristic pattern of the deviation based on the deviation between the matching index and the preset target value. Based on the deviation analysis results, the parameter update unit dynamically updates the parameter values ​​in the output waveform adjustment parameter table.

[0016] Compared with the prior art, the beneficial effects of the present invention are: By acquiring battery parameter data transmitted by the vehicle's BMS through the power demand modeling module, analyzing battery voltage and current thresholds, extracting multiple sets of battery cell state feature quantities and generating a battery demand sequence, and constructing a power demand feature template, the system can accurately capture the power demand of the battery under different states, providing a precise basis for subsequent power supply parameter adjustments and avoiding the problem of mismatch between power supply and battery demand in the traditional fixed parameter output mode.

[0017] The multi-source integration module, based on an energy demand characteristic template, collects charging pile output parameters and environmental monitoring data, matches characteristic quantities in the battery demand sequence, and fuses multi-source data to generate an integrated parameter sequence. This achieves comprehensive integration of charging pile output information, battery status information, and environmental information. This multi-source data fusion approach breaks through the limitations of relying on single data in traditional technologies, and can more comprehensively and accurately reflect the overall status of the charging system, making subsequent dynamic adjustments more targeted and effectively improving the accuracy of power supply control.

[0018] The dynamic adjustment module calculates the charging stage division threshold based on the integrated parameter sequence, identifies the power boundary conditions of different charging segments, and generates a dynamic control instruction set. It can divide the charging stage and adjust the power boundary in real time according to changes in battery status and environmental factors. During charging, as the battery capacity increases and environmental conditions change, the charging stage and corresponding power demand will continuously change. This module can promptly identify these changes and generate corresponding control instructions, ensuring that suitable power supply parameters are provided to the battery in each charging stage. This optimizes the charging process and avoids the problems of low charging efficiency or battery damage caused by fixed charging stage divisions.

[0019] The safety boundary determination module, based on a dynamic control command set, monitors the charging pile's output fluctuation characteristics and BMS feedback signals, and generates boundary state identifiers by comparing them with safety threshold ranges. This enables real-time monitoring and determination of the safety status during the charging process. Compared to traditional fixed safety thresholds, this module can flexibly adjust safety judgment standards by combining dynamic control commands and real-time feedback signals. When fluctuations occur in the charging pile's output or abnormal BMS feedback signals occur, it can quickly identify and generate accurate boundary state identifiers, promptly detecting potential safety hazards during the charging process and providing support for subsequent safety responses.

[0020] The optimized output module adjusts the charging pile's output waveform parameters based on boundary state indicators, matches the changing trends of characteristic quantities in the battery demand sequence, and generates an optimized power supply strategy, further refining the adjustment of power supply parameters. By adjusting the output waveform parameters, the electrical energy output by the charging pile can better meet the battery's demand characteristics under different states, reducing energy loss during battery charging. Simultaneously, by matching the changing trends of the battery demand sequence, it ensures that the power supply strategy can be optimized in real time as the battery state changes. This improves charging efficiency, effectively protects the battery, extends battery life, and reduces safety risks during charging, providing a more reliable and efficient solution for electric vehicle charging. Attached Figure Description

[0021] Figure 1 This is a schematic diagram illustrating the working principle of the integrated power supply controller for charging piles and vehicle BMS described in this invention. Figure 2 A flowchart for dynamically adjusting the module's operation; Figure 3 A flowchart for optimizing the operation of the output module. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Please see Figure 1This invention provides an integrated power supply controller for charging piles and vehicle BMS. The system includes: an energy demand modeling module that acquires battery parameter data transmitted by the vehicle BMS, including real-time information such as battery voltage and current thresholds; after parsing these parameters, the module extracts multiple sets of battery cell state feature quantities, including physical quantities such as voltage fluctuation characteristics and current change rate; and generates a battery demand sequence based on the order of appearance of the feature quantities in the data stream, reflecting the changes in battery energy demand at different time points; and constructs an energy demand feature template based on this sequence, which standardizes the representation of battery demand. A multi-source integration module, based on the energy demand feature template, simultaneously collects charging pile output parameters such as output voltage and current, and environmental monitoring data such as temperature and humidity; matches these multi-source data with the feature quantities in the battery demand sequence; the matching process is completed through time alignment and feature weighting; and after fusion, an integrated parameter sequence is generated, which unifies the data formats from different sources. The dynamic adjustment module analyzes the stage changes during the charging process based on the integrated parameter sequence, calculates the threshold for dividing the charging stage (the threshold is determined based on battery characteristic curves and historical data), identifies the power boundary conditions for different charging segments, defines the allowable power output range for each stage, and generates a dynamic control instruction set containing specific control parameters. The safety boundary determination module, based on the dynamic control instruction set, monitors the charging pile's output fluctuation characteristics in real time, such as voltage ripple and current spikes, as well as BMS feedback signals such as battery temperature and SOC value. It compares this monitoring data with preset safety threshold ranges, set according to battery type and charging standard, and generates boundary state identifiers to indicate whether the system is in a safe operating state. The output optimization module, based on the boundary state identifiers, adjusts the charging pile's output waveform parameters, such as frequency and duty cycle, to match the characteristic quantity change trends in the battery demand sequence. These characteristic quantity change trends reflect the real-time demand fluctuations of the battery, generating an optimized power supply strategy. This strategy optimizes the charging process by adjusting power output devices.

[0024] Example 1: The construction of the energy demand feature template begins with the detailed definition of the battery cell state coding structure. The battery cell state coding structure receives the raw battery parameter data stream from the vehicle's BMS. The data stream contains multi-dimensional information such as real-time voltage sampling values, current measurements, and temperature readings for each battery cell. The coding process first assigns a unique digital identifier to each battery cell, which is associated with the physical location or logical address within the battery pack. Continuous analog voltage and current signals are then subjected to high-speed analog-to-digital conversion to quantize them into discrete digital values ​​with a specific precision. The coding structure organizes these data in a hierarchical format. The top layer is the battery pack identification code, the middle layer is the battery cell index sequence, and the bottom layer is the set of specific parameter values ​​corresponding to that cell. This structure ensures that each data point can be accurately traced back to a specific battery cell, establishing a clear data attribution relationship for subsequent feature processing. In the encoding process, a data validity check bit is also introduced to mark abnormal data points that may occur during transmission. The encoder will monitor the continuity of the data stream in real time. Once a timestamp interruption or a value jump exceeding the reasonable range is detected, a suspicious flag is marked in the check bit of that data point. This encoding mechanism not only preserves the original numerical characteristics of the parameters, but also embeds data quality information, providing a basis for reliability judgment for subsequent processing modules.

[0025] The generation of the feature quantity time-series arrangement set relies on deep time-series analysis of the encoded data stream. The system extracts a series of predefined features from the encoded data stream. These features include, but are not limited to, local extrema of the voltage curve, sign transition times of the current rate of change, and fluctuation periods of the internal resistance calculation. Each feature quantity is recorded with its absolute timestamp under a high-precision timer upon extraction, and its sequence position number in the complete data stream is also marked. The time-series arrangement algorithm scans the entire data stream, identifies the temporal relationships between features, and constructs a time-labeled feature quantity chain structure. In this chain structure, features are arranged in chronological order of appearance, and the time intervals between adjacent features are precisely recorded, forming a time-series relationship graph of features. For periodically occurring features, the system specifically records their occurrence period and phase information. The time-series arrangement set also considers the causal relationships between different features. For example, a voltage drop feature may appear before a temperature rise feature. This temporal dependency is quantified as a time delay parameter between features, which together constitute a time-series correlation network of features. This network can reflect the temporal regularity in the dynamic characteristics of the battery.

[0026] The determination of voltage and current weight parameters is a dynamic and adaptive process. The system assigns appropriate weight values ​​to each voltage and current feature based on the battery type, usage history, and current health status. The initial weights are set based on the specifications provided by the battery manufacturer. For example, for power batteries, the weight of current features may be higher than that of voltage features, while for energy batteries, the weight of voltage features may be more important. The weight parameters are adjusted in real time as the charging process progresses. The adjustment algorithm analyzes the contribution of each feature to battery state prediction in historical data. The contribution is measured by the correlation coefficient between the feature and key battery parameters such as SOC. Features with higher correlation coefficients will receive higher weights. The system also considers the stability of the features. Features with smaller fluctuations are usually assigned higher weights because they represent more reliable information. The weight calculation module periodically re-evaluates the weight allocation of all features to ensure that the weight values ​​reflect the current actual operating characteristics of the battery. The weight parameters are stored in vector form for easy retrieval during subsequent matching and fusion processes.

[0027] Establishing multi-source data mapping relationships requires addressing the heterogeneity of different data sources. Charging pile output parameters are typically provided in Modbus or CAN bus protocol formats, including electrical parameters such as output voltage, output current, and power factor. Environmental monitoring data comes from IoT devices such as temperature and humidity sensors, employing different communication protocols and data formats. The mapping relationship construction module first converts all data into a standardized data model, which defines the data type, unit, and valid range of each parameter. Subsequently, based on a time synchronization mechanism, a unified time tag is assigned to each data point, establishing a time alignment relationship across data sources. The mapping relationship table is implemented using a relational database structure, where each record contains fields such as timestamp, data source identifier, parameter type, and parameter value. The system creates corresponding index links in the mapping table for the feature quantities in each battery demand sequence. These links point to related parameter items in different data sources. For example, the battery voltage feature quantity may be mapped to both the charging pile output voltage and the BMS acquisition voltage. The mapping relationship also defines the correlation strength between data sources, with data with strong correlation having higher priority in fusion processing.

[0028] The formation of the feature matching result set is a multi-level filtering process. The matching engine first searches for corresponding parameter items in multi-source data for each feature quantity in the energy demand feature template. Matching criteria include similarity of parameter types, overlap of numerical ranges, and correlation of time series. For numerical features, a sliding window correlation algorithm is used to calculate the similarity score between the template feature quantity and the real-time data sequence. For state-related features, a pattern matching algorithm is used to identify whether a corresponding state pattern exists in the data. Each matching result is accompanied by a confidence score, reflecting the reliability of the match. The matching result set is organized by feature quantity, with each feature quantity corresponding to a matching result entry. Each entry records the list of matched data sources, the matching score, and the time offset of the best match. The system retains multiple possible matching candidates and sorts them according to matching quality. The matching result set also records various metadata from the matching process, such as matching calculation time and data quality indicators, providing rich contextual information for subsequent fusion processing. The calculation of time alignment parameters involves complex time series processing techniques. Due to clock skew and transmission delays in different data source acquisition devices, directly merging data can lead to time synchronization issues. The alignment module first uses a high-precision time synchronization protocol to minimize the fundamental clock differences between data sources. Then, it uses an interpolation algorithm to resample the non-uniformly sampled data, unifying it onto the same time grid. For key feature points, the system detects their occurrence time in each data source, calculates the time deviation, and generates alignment correction parameters. The time alignment parameter table records the delay compensation value, sampling rate conversion coefficient, and data validity time window for each data source relative to the reference clock. The alignment algorithm pays special attention to the precise moment of feature occurrence, employing a higher-precision time interpolation method near feature points to ensure accurate positioning of features on the time axis. The alignment parameters are also dynamically adjusted based on network conditions to address the impact of transmission delay fluctuations.

[0029] The generation of stage division node numbers relies on the detection of abrupt changes in the integrated parameter sequence. The algorithm monitors the first and second differences of the parameter sequence in real time. When the difference value exceeds a preset threshold, it is marked as a potential stage transition point. Node numbers are assigned in chronological order, with each number corresponding to a stage boundary time point. The numbering system also considers a hierarchical structure, assigning first-level numbers to major stage transition points and second-level numbers to minor fluctuation points, forming a hierarchical representation of stage division. Node numbers are associated with specific charging stage types, such as the start point of the constant current stage and the transition point of the constant voltage stage. Each number carries a detailed stage attribute description, including stage type, expected duration, and typical characteristic patterns. The node number sequence provides a time frame reference for subsequent power calculations. The derivation of the power boundary condition set is based on battery physical characteristics and safety specifications. The boundary conditions for each charging stage include the minimum and maximum allowable power values, as well as the recommended power operating point for that stage. The boundary value calculation comprehensively considers the battery's chemical characteristics, thermal limitations, and electrical constraints. For example, in high-temperature environments, the power upper limit will be appropriately reduced to prevent overheating. The boundary conditions also consider the battery's aging degree; as the number of cycles increases, the power boundary will be adjusted accordingly to extend battery life. The boundary condition set is represented in the form of a piecewise function, with each stage corresponding to a power range interval. The interval boundary can be a fixed value or a dynamic value that varies with parameters such as temperature and SOC. The boundary condition set also defines the power change rate limit during stage transitions to ensure a smooth power transition without any shocks.

[0030] The control parameter index table establishes a mapping relationship between power boundary conditions and specific execution instructions. The index table uses a key-value pair structure for storage, where the key is the hash value of the power boundary condition, and the value is the corresponding combination of control parameters. These control parameters include adjustable parameters such as the charging pile's voltage setting, current limit, and switching frequency. Each parameter has a clearly defined adjustment range and resolution. The index table supports a multi-level indexing mechanism, allowing for rapid retrieval of optimal control parameters based on multiple boundary condition dimensions. The table structure also includes version management functionality, facilitating updates to the control strategy based on changes in battery status. The index table is implemented in memory using an efficient data structure, ensuring fast access during real-time control.

[0031] The safety threshold comparison results are generated through multi-dimensional data comparison. The system compares the real-time monitored waveform features with the preset safety thresholds item by item. The comparison operation is not a simple numerical judgment, but a comprehensive assessment considering the duration and trend of change. For example, a voltage exceedance duration shorter than the set time limit may not be considered abnormal, while a duration exceeding the limit will trigger an alarm. In addition to Boolean normal / abnormal indicators, the comparison results also include a quantitative score of the degree of deviation and trend prediction. The comparison algorithm adopts a hysteresis comparison strategy to avoid frequent state jumps near the threshold. The comparison of important parameters undergoes multiple verifications to ensure the reliability of the results. The comparison results are accompanied by detailed timestamps and contextual information for easy traceability and analysis. The fluctuation feature classification identifier is generated based on signal processing technology. The system performs real-time spectrum analysis and time-frequency analysis on the output waveform of the charging pile, extracting parameters such as characteristic frequency components, harmonic distribution, and modulation characteristics of the waveform. Based on these parameters, the fluctuations are classified into different types, such as low-frequency fluctuations, high-frequency noise, periodic ripples, and random interference. Each type corresponds to different physical causes and processing strategies. The classification identifier uses multidimensional encoding, which includes multiple dimensions such as fluctuation type, amplitude level, and frequency characteristics. The classifier will continuously learn from historical data to optimize classification accuracy. The classification results are associated with processing suggestions, providing a direct basis for optimizing the output.

[0032] The anomaly status coding adopts a standardized code system, with each anomaly type corresponding to a unique numerical code. The code structure is designed with a hierarchical classification, where the first digit indicates the major anomaly category (such as electrical anomaly, thermal anomaly, etc.), and subsequent digits identify the specific anomaly subclass. The coding system also includes auxiliary information such as anomaly severity level, occurrence time, and duration, forming a complete anomaly description framework. Anomaly codes are associated with handling plans, and the code parsing module can quickly retrieve the corresponding handling process based on the code. The coding system supports dynamic expansion, facilitating the inclusion of newly discovered anomaly types and ensuring the long-term adaptability of the system.

[0033] Example 2: The operation of the power demand modeling module begins with the feature extraction submodule receiving and processing the data transmitted from the BMS. The data stream enters the processing unit through an isolated communication interface. The transmission protocol typically uses CAN bus or Ethernet communication format. The data packet contains real-time sampled values ​​of voltage, current, and temperature of each unit in the battery pack, as well as system status words and other information. The feature parsing operation first performs integrity verification and parsing on the data packet, extracts valid battery parameter fields, and standardizes the parameter data, converting it to a unified engineering unit system and numerical range. The extraction of the feature quantity set is performed independently for each battery unit. The calculation unit derives a series of feature quantities based on the original parameters, including but not limited to the moving average of voltage sampled values, the short-term trend of current readings, the parameter difference between adjacent sampling points, and statistical features based on time windows, such as variance and extreme values. Each feature quantity is immediately stamped with a high-precision timestamp after being calculated. This timestamp strictly corresponds to the data acquisition time. The time sequence position of the feature quantity in the data stream is recorded using a doubly linked list data structure. Each feature quantity node contains pointers to the previous and next time nodes, forming a chain of feature quantities arranged in chronological order. The key feature recognition algorithm scans the entire feature set to find a subset of features that are significant for characterizing the battery state, such as voltage plateau features and current inflection point features. The algorithm compares the relative relationship between the first occurrence position of these key features in the parameter sequence and the length of the entire data stream, and calculates the time proportion and distribution density of the features. The battery cell type classification operation is performed based on the battery type identifier in the BMS message. Different types of battery cells, such as power lithium-ion batteries and energy lithium iron phosphate batteries, are assigned to different processing channels. The classification process also considers the topological connection relationship of battery cells within the group. The final generated feature distribution result is organized in the form of a multi-dimensional feature vector. The vector dimension corresponds to different types of feature quantities, and the vector element values ​​represent the numerical magnitude and temporal attributes of the feature quantities.

[0034] After receiving the feature distribution results, the sequence construction submodule starts the data processing flow. The identification of key features is based on a predefined feature importance scoring system. The scoring system is established based on historical data of the correlation between feature quantities and battery core parameters such as SOC and SOH. The extraction of data segments corresponding to key features adopts dynamic time window technology. The window size is adaptively adjusted according to the feature type. Shorter windows are used for rapidly changing transient features, while longer observation windows are used for slowly changing steady-state features. The data segment truncation algorithm ensures that each feature segment contains the complete feature evolution process. The time interval is determined based on the duration of the feature quantity and the interval period between adjacent features. The truncation operation retains all the detailed information of the original data. The feature fragment set is constructed using database sharding technology. Each feature fragment is stored as an independent data object. The object attributes include feature type, time range, data source identifier, etc. The reorganization process re-aggregates the scattered feature fragments based on the battery cell identifier. The reorganization algorithm considers the time overlap and gap between feature fragments and uses data fusion technology to fill time gaps and smooth overlapping areas. The output format of the feature quantity time series set is a time series database, where each time point corresponds to a complete set of feature quantity values. The time series set also includes descriptions of the correlation between feature quantities, such as causal relationships, correlation degree, and other additional information.

[0035] The template generation submodule performs in-depth analysis of the feature quantity time series set. The statistical analysis of weight parameter values ​​is based on the frequency of occurrence, numerical stability, and significance of the feature quantity's impact on battery state in historical data. The weight calculation adopts an adaptive learning algorithm, continuously optimizing the weight allocation as data accumulates. The original order of feature quantities in battery parameter data is reconstructed through timestamp sequences. The rearrangement process uses a stable sorting algorithm to maintain the strict monotonicity of the feature quantity time order. The time series rearrangement considers the time deviation between data streams of different battery units and performs cross-unit time series alignment correction. The integration of time series segments of multiple feature quantities within the same battery unit adopts data weaving technology to interweave the time series of different feature quantities into a unified timeline. The sequence splicing algorithm specially handles the boundary regions between feature quantity time segments and uses a smooth transition function to avoid discontinuities at splicing points. The final output of the energy demand feature template is in a structured document format, containing complete feature quantity time series, weight parameter tables, unit relationship diagrams, and other components. The template data adopts a compressed storage format to reduce transmission bandwidth consumption while retaining fast random access capabilities.

[0036] The data acquisition submodule of the multi-source integration module simultaneously monitors multiple data source inputs. The charging pile output parameter set is obtained through a real-time communication interface, including electrical parameters such as the effective value of the output voltage, the peak value of the output current, and the power factor. The environmental monitoring data stream comes from a sensor network deployed at the charging site, and the monitored variables include physical quantities such as ambient temperature and humidity, cooling fan status, and connector temperature. The time alignment operation adopts a combination of hardware time synchronization and software interpolation. A precise clock protocol is used to reduce the basic time deviation between data sources. For data points with inconsistent sampling times, a spline interpolation algorithm is used to generate values ​​on a unified time grid. The time alignment parameter table records detailed information such as the original sampling time, the aligned timestamp, and the interpolation coefficients for each data source. The multi-source data time series index is constructed as a multi-layer index structure. The top layer is an absolute time index, and the lower layer is a relative time offset index for each data source. The index structure supports efficient range query and point query operations, meeting the requirements of real-time processing for data access speed.

[0037] The fusion processing submodule operates based on the aligned multi-source dataset. Feature matching is performed in three dimensions: temporal matching of data from different sources at the same time, spatial matching of sensor readings at the same physical location, and logical matching of monitoring parameters of the same type. The matching algorithm employs a multi-level filtering strategy, first performing coarse-grained time window matching, then fine-grained numerical correlation matching. Feature-weighted fusion dynamically adjusts the weighting coefficients based on the confidence weight of the feature and data quality indicators. The weighting process considers the reliability differences of different data sources, assigning higher weights to data sources with high reliability. The calculation of the integrated parameter value for each time series point uses a robust estimation algorithm to reduce the impact of outliers on the fusion result. The generation of the integrated parameter sequence adopts a streaming processing mode, updating the sequence content in real time when new data points arrive. The sequence storage uses a circular buffer structure to balance historical data retention and memory usage efficiency.

[0038] The verification feedback submodule continuously monitors the quality of the integrated parameter sequence. Difference analysis between the feature matching results and the original electricity demand feature template is conducted from multiple perspectives, including numerical deviation statistics, trend consistency checks, and outlier distribution comparisons. Difference calculation employs a multi-index comprehensive evaluation method, with each index quantifying the degree of deviation between the matching result and the expected pattern. Data quality assessment indicators include sub-indicators such as integrity score, consistency level, and timeliness score. These indicators comprehensively reflect the reliability level of the data fusion process. The feedback mechanism is designed as a closed-loop control structure, with quality assessment results sent back to the fusion processing submodule in real time. The fusion algorithm dynamically adjusts the feature weighting strategy and matching threshold parameters based on the feedback information, achieving self-optimization of the data fusion effect. The feedback signal is also used to trigger data source quality assessment, identify data sources with degraded performance, and take compensatory measures. The data flow between modules adopts a pipeline architecture, with the output of the previous module serving as the input of the next module. The modules employ a multi-threaded parallel processing mechanism, and computationally intensive tasks such as feature matching and sequence alignment are assigned to dedicated processing units for execution. Data uses a unified encoding format in each processing stage to reduce format conversion overhead. An error handling mechanism monitors the success status of each processing step and automatically initiates a data recovery process when an anomaly occurs, ensuring that the system can still degrade to operation even when some data is abnormal.

[0039] Example 3: See Figure 2 The dynamic adjustment module operates based on the input of an integrated parameter sequence, which contains time-series information fused from multi-source data. The module performs in-depth analysis of the sequence through a stage division sub-module. This sub-module uses a pattern recognition algorithm to scan the integrated parameter sequence and identify nodes where characteristic patterns change during the charging process. These nodes typically correspond to transition points in the charging stages, such as the critical position when switching from constant current charging to constant voltage charging. The algorithm calculates the local statistical features of the sequence based on the sliding window technique, such as mean, variance, and gradient changes. When these feature values ​​exceed a preset sensitivity threshold, they are marked as potential stage boundaries. The generation of stage boundary time series point numbers uses an incremental numbering method, with each number corresponding to a unique timestamp. The stage division node list is stored in the form of a data structure, containing the time coordinates of the boundary points, stage type identifiers, and confidence scores. The list is updated in real time as new data flows in, ensuring the timeliness of stage division.

[0040] After receiving the node list of the stage division, the power calculation submodule starts the calculation process. The extreme values ​​of power demand in each stage interval are obtained by the peak detection algorithm within the time window. The algorithm searches the data segment corresponding to each stage to find the local maximum and minimum values ​​of power. The extreme value detection takes into account the smoothness of the data and adopts noise reduction processing to avoid misjudgment caused by noise. The calculation of the power transition threshold between adjacent stages introduces mathematical formulas for quantification.

[0041]

[0042] in: Represents the transition power threshold. The proportional gain is determined by the battery type. This represents the maximum energy requirement in stage n. This represents the minimum energy requirement for the (n-1)th stage. It is the time difference parameter between stages. The calibration offset is learned from historical data. The power boundary condition set is generated by combining battery rated parameters, such as the maximum allowable charging current and voltage limits. The boundary conditions are represented in interval form, including the upper and lower limits of power operation for each stage. The set is stored in a hierarchical structure to support fast query and update.

[0043] The instruction generation submodule maps power boundary conditions to the actual control parameters of the charging pile. The matching process is based on the output capability curve of the charging pile, which describes the efficiency characteristics at different operating points. The construction of the control parameter index table adopts hash mapping technology to map the power value range to specific control instructions, such as switching frequency setpoints or voltage reference values. After the instruction set is generated, it is encapsulated in digital signal format, including time stamps and target parameter values. The transmission of the instruction set is completed through a real-time communication bus to ensure low-latency control.

[0044] The activation of the safety boundary determination module and the execution of the dynamic control instruction set are synchronized. The signal monitoring submodule continuously collects the characteristic parameters of the charging pile output waveform. During the acquisition process, a high-precision ADC converter is used to digitize the analog signal. The waveform characteristics include voltage ripple coefficient, current harmonic distortion, and phase offset. Battery status data fed back by the BMS is received through an isolated communication interface. After parsing the data packets, key parameters such as cell voltage, temperature, and health status indicators are extracted. Digital signal processing algorithms, such as Fast Fourier Transform, are used for frequency domain analysis to extract waveform fluctuation characteristics. Battery anomaly feature encoding is based on a predefined pattern library, classifying abnormal signals into categories such as overvoltage, undervoltage, and overtemperature. The threshold comparison submodule calls the safety threshold parameter table stored in non-volatile memory. This table is loaded from the configuration file during system initialization and contains various limits specified by battery safety standards. The comparison operation adopts a parallel processing architecture, comparing multiple features with thresholds simultaneously. The feature state comparison result set records the actual value, threshold, and deviation of each comparison item. The comparison algorithm introduces hysteresis comparison logic to prevent boundary oscillations. The comparison of important parameters is performed by multiple sampling averaging to improve reliability.

[0045] The status identifier generation submodule classifies comparison results into levels: anomaly levels are categorized into three severity categories: attention, warning, and danger. The classification rules consider the duration and trend of the deviation. Boundary status identifiers are encoded using a binary bitmap format, with each bit representing an anomaly state. The identifier update frequency is synchronized with the data acquisition cycle to ensure real-time performance. Data exchange between modules uses a shared memory area. The instruction sets generated by the modules are dynamically adjusted and written to the shared area. The safety boundary determination module reads and processes this information. An error handling mechanism monitors the module execution status, triggering a watchdog timer for reset in case of an anomaly. The entire system's timing is synchronized by a high-precision clock source. The execution sequence of each submodule is managed by a task scheduler, with scheduling based on priority allocation. Critical tasks such as safety determinations have the highest priority. Data flow adopts a pipelined processing mode, with the output of the previous module buffered for use by subsequent modules to reduce waiting time. System resource usage is monitored in real time, and a cleanup program is automatically triggered when memory or CPU usage exceeds a threshold. All operations are recorded in a circular log for post-event analysis.

[0046] Example 4: See Figure 3 The operation of the optimized output module is based on the input of the boundary state identifier, which comes from the real-time output of the safety boundary determination module. The module analyzes the abnormal state distribution pattern through the strategy generation submodule. For example, in a specific charging scenario, the BMS feedback signal shows that multiple cells in the battery pack have abnormal temperature gradients. The temperature reading rises from 25 degrees Celsius to 40 degrees Celsius in a short period of time. At the same time, the charging pile output waveform detects an increase in current ripple. The strategy generation submodule first identifies the time distribution characteristics of this abnormal state and finds that the temperature abnormality begins 300 seconds after the start of charging and continues to exist for the next 200 seconds. Spatially, the abnormality is concentrated in the middle area cells of the battery pack. The pattern analysis unit establishes an abnormal state distribution map based on these data. The map uses the time axis as the horizontal axis and the battery cell position as the vertical axis, marking the contour distribution of the abnormal intensity. The map shows the pattern of the abnormality spreading from the central cell to the periphery. The parameter mapping unit queries the set of feature quantities corresponding to the time interval in the power demand feature template and finds that the voltage feature quantity shows a prolonged plateau period during this period, while the current feature quantity shows an increased fluctuation frequency. A mapping relationship table between abnormal states and feature quantities is established. See Table 1 for details. The feature quantities associated with each type of abnormality and their influence weights are recorded.

[0047] Table 1: Mapping Relationship between Abnormal States and Feature Quantities

[0048] The adjustment strategy unit generates output waveform adjustment parameter combinations based on the mapping table. For abnormal temperature gradients, the parameter combinations include reducing the charging current amplitude by 10% and increasing the switching frequency by 5kHz. For abnormal voltage ripples, the equivalent series resistance compensation value of the filter capacitor is increased. The parameter combinations are stored in the adjustment parameter table in the form of digital instructions. The table structure supports parallel storage of multiple sets of parameters for easy and rapid switching. The strategy generation process considers the superposition effect of abnormal states. When multiple abnormalities exist simultaneously, the adjustment strategy uses a weighted fusion algorithm to generate comprehensive parameter settings to ensure the coordination of adjustment actions. After receiving the adjustment parameter table, the output control submodule initiates the execution process. The drive parameter calculation unit converts the adjustment parameters into drive signals for the power devices. For example, it converts the current reduction command into a specific value for the PWM duty cycle. The conversion process is based on the hardware characteristic curve of the charging pile. The real-time adjustment unit monitors the power boundary conditions corresponding to the current charging stage. For example, the current limit in the constant current stage is 100A, and the drive signal parameters are dynamically adjusted to keep it within a safe range. The protection execution unit monitors the actual output of the power devices through a high-speed comparator circuit. When the deviation between the output current and the set value exceeds the tolerance value of 2A, protective adjustments are immediately initiated, such as gradually reducing the output power or switching to a standby control mode. The output control adopts a closed-loop feedback mechanism, and the drive signal parameters are updated every 10 milliseconds to ensure real-time response.

[0049] The adaptive adjustment submodule continuously evaluates the output performance. The matching degree calculation unit collects the output waveform characteristics of the charging pile, including voltage steady-state error and current tracking accuracy. Simultaneously, it receives battery response data from the BMS, such as voltage change rate and temperature rise rate. It calculates the matching degree index between the waveform characteristics and the battery response. The index value is derived using a correlation coefficient algorithm, ranging from 0 to 1; a higher value indicates a better matching degree. The deviation analysis unit compares the matching degree index with the preset target value of 0.9, identifying the time interval in which the deviation occurs. For example, if the matching degree drops to 0.7 between the 400th and 500th seconds of charging, the analysis reveals that this is mainly due to current response delay. The parameter update unit dynamically adjusts the output waveform adjustment parameter table based on the analysis results. For example, it increases the proportional gain of the current loop by 15% and decreases the integral time constant by 20%. The update cycle is synchronized with the charging stage, and the validity of the parameters is re-evaluated at each stage transition. The entire module's implementation takes into account changes in the actual operating environment, such as grid voltage fluctuations or load abrupt changes. The module has a built-in self-learning function, optimizing adjustment parameters through historical data recording. The anomaly handling mechanism employs a multi-level recovery strategy: minor anomalies only require parameter adjustments, while severe anomalies trigger tiered protection. Data communication utilizes redundant channels to ensure reliable transmission of control commands. The module records detailed operation logs during operation, including timestamps, parameter values, and execution results for each adjustment action, supporting offline analysis and optimization. System resource management adopts a dynamic allocation strategy, prioritizing computing resources for real-time tasks and executing non-critical tasks in the background. The module's initialization process includes self-checks and parameter calibration to ensure rapid entry into operational status after power-on.

[0050] In a specific application example, suppose a certain type of electric vehicle battery experiences localized overheating during fast charging. The optimization output module parses the abnormal code E-102 activation from the boundary state identifier. The strategy generation submodule queries the mapping table and learns that the current output needs adjustment. The output control submodule gradually reduces the charging current from 150A to 135A while increasing the switching frequency to improve heat dissipation. The adaptive adjustment submodule detects that the temperature rise is slowing down and gradually restores the current output to 140A. The entire adjustment process is completed within 30 seconds, and the battery temperature stabilizes within a safe range. The coordinated operation of the modules ensures the safety and efficiency of the charging process. Similar processing logic is applicable to various abnormal scenarios, such as voltage surges and current spikes. The software implementation of the modules adopts modular programming, with each submodule running as an independent task and exchanging data through a message queue. The hardware platform is based on a multi-core processor architecture, with computationally intensive tasks allocated to dedicated cores and real-time control tasks executed by hardware accelerators. The system supports online upgrades, allowing for updates to the control algorithm based on battery technology developments. The debugging interface provides detailed status monitoring functions, facilitating on-site fault diagnosis and performance optimization.

[0051] Example 5: The drive parameter calculation unit of the output control submodule continuously receives the output waveform adjustment parameter table from the strategy generation submodule. This table stores the adjustment instructions corresponding to different abnormal states in array form. For example, when the system detects an abnormal rise in battery temperature, the adjustment parameter table may include specific values ​​such as reducing the charging current by 5% and increasing the switching frequency by 2kHz. The calculation unit first parses the meaning and priority of these adjustment parameters and converts them into drive signal parameters that the power device can directly execute. For switching devices such as insulated gate bipolar transistors, the drive signal parameters mainly include the pulse width modulation duty cycle of the gate voltage and the switching frequency setting value. The conversion process is based on the transfer characteristic curve of the power device, which describes the correspondence between the control parameters and the output electrical characteristics. The real-time adjustment unit acquires the power boundary conditions corresponding to the current charging stage via a high-speed data bus. These conditions are periodically refreshed from the dynamic adjustment module in the form of data packets. For example, during constant current charging, the boundary conditions may limit the output current to be maintained between 100A and 120A. The adjustment algorithm scans the changes in boundary conditions every 5 milliseconds, dynamically calculating the allowable fluctuation range of the drive signal parameters. When the calculated drive parameters exceed the safe range, the adjustment unit automatically clamps them near the boundary value and submits a parameter over-limit alarm to the system. The protection execution unit operates independently at the hardware level and includes a dedicated comparator and status monitoring circuit. This unit samples the actual output value of the power device at a 100-microsecond cycle. For example, it collects the actual output current through a current sensor and monitors the output voltage ripple through a voltage probe. The measured values ​​are compared with the theoretical expected values ​​of the drive signal parameters in real time. When the deviation exceeds the tolerance value for three consecutive sampling cycles (e.g., the output current deviation is continuously greater than 3A), the protection circuit immediately activates a graded response mechanism. First, it attempts to fine-tune the drive parameters for compensation. If the deviation continues to increase, it gradually reduces the output power until it triggers complete shutdown protection.

[0052] The matching degree calculation unit of the adaptive adjustment submodule synchronously collects the output waveform characteristics of the charging pile and the BMS feedback data. The output waveform characteristics are extracted by a digital signal processor, including frequency domain characteristics such as fundamental amplitude, harmonic distortion rate, and phase angle. The battery response data fed back by the BMS includes time domain characteristics such as the slope of terminal voltage change, dynamic change of internal resistance, and temperature rise curve. The calculation unit adopts a sliding time window mechanism, taking a calculation cycle every 200 milliseconds, normalizing the waveform characteristics and battery response data within the window, and then calculating the matching degree index of the two through a pattern recognition algorithm. This index is a dimensionless value between 0 and 1, representing the degree of agreement between the output waveform and the expected battery response. The deviation analysis unit maintains a first-in-first-out historical data queue, continuously recording the matching degree index and its timestamp for the most recent 30 sampling periods. When the system detects that the current matching degree index is lower than the preset target value of 0.85, the analysis unit activates the deviation analysis process. First, it locates the starting time point of the matching degree decline, and then extracts data segments from 10 periods before and after that time point for pattern recognition. The analysis focuses on the rate of change, duration, and periodicity of the matching degree. The recognition results are output in the form of feature vectors, indicating the severity level and trend of the deviation. The parameter update unit uses an incremental learning strategy to update the output waveform adjustment parameter table. This unit calculates the direction and magnitude of parameter adjustment based on the deviation analysis results. For example, when it is identified that the decline in matching degree is due to current response delay, the parameter update unit will appropriately increase the proportional gain of the current loop, but limits the single adjustment magnitude to no more than 20% of the original value. The updated parameters are officially written into the adjustment parameter table only after validity verification, and the adjustment log is saved for subsequent analysis.

[0053] In a specific operational scenario, the charging system performs fast charging for a certain type of ternary lithium battery pack. The initial setting of the charging pile is an output current of 150A and a voltage of 400V. At the 8-minute mark of charging, the drive parameter calculation unit receives an instruction from the adjustment parameter table, requesting an increase in the switching frequency from 20kHz to 22kHz to improve output ripple. The calculation unit immediately consults the IGBT driver chip's datasheet, converting the frequency parameters into corresponding register configuration values. Simultaneously, the real-time adjustment unit monitors the system in a constant current charging phase, where the power boundary condition requires current fluctuations not to exceed ±5% of the set value. Therefore, it performs compliance verification on the drive parameters. At this point, the protection execution unit detects abnormal fluctuations in the actual output current through sensors, with the maximum instantaneous value reaching 158A, exceeding the tolerance threshold of 3A. The protection unit immediately initiates a level-one response, temporarily reducing the PWM duty cycle by 2% and simultaneously sending a warning signal to the main control system. Meanwhile, the matching degree calculation unit of the adaptive adjustment submodule found that the matching degree between the output current waveform and the voltage response curve reported by the BMS dropped from 0.9 to 0.75. The deviation analysis unit tracked that the drop started at 7 minutes and 45 seconds and showed a continuous deterioration trend. Based on the analysis results, the parameter update unit adjusted the integral time constant of the current loop from 0.1 seconds to 0.08 seconds. After two adjustment cycles, the matching degree gradually recovered to 0.88, and the system returned to stable operation.

[0054] The entire module adopts a distributed computing architecture. The output control submodule is deployed on an FPGA chip to achieve nanosecond-level response, while the adaptive adjustment submodule runs on a DSP processor to perform complex algorithm calculations. The two submodules exchange data through dual-port RAM to ensure a balance between real-time performance and computing power. The system also features a parameter version management mechanism, which retains historical versions after each parameter update and supports rapid rollback when new parameters cause performance degradation. All important operations are logged, including parameter modification time, values ​​before and after modification, and operator identification information, facilitating subsequent traceability and analysis. The module has self-diagnostic capabilities, periodically checking the working status of each unit and automatically switching to a redundant backup unit when an anomaly is detected.

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

[0056] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A charging pile and vehicle BMS integrated power supply controller, characterized in that, The controller includes: The power demand modeling module acquires battery parameter data transmitted by the vehicle's BMS, analyzes battery voltage and current thresholds, extracts multiple sets of battery cell state features, generates a battery demand sequence based on the order of appearance of the features, and constructs a power demand feature template. Based on the power demand feature template, the multi-source integration module collects charging pile output parameters and environmental monitoring data, matches feature quantities in the battery demand sequence, and fuses multi-source data to generate an integrated parameter sequence. The dynamic adjustment module calculates the charging stage division threshold based on the integrated parameter sequence, identifies the power boundary conditions of different charging sections, and generates a dynamic control instruction set. The safety boundary determination module, based on the dynamic control instruction set, monitors the output fluctuation characteristics of the charging pile and the BMS feedback signal, and generates a boundary status identifier by comparing the safety threshold range. The optimization output module adjusts the output waveform parameters of the charging pile according to the boundary state identifier, matches the characteristic quantity change trend in the battery demand sequence, and generates an optimized power supply strategy.

2. The integrated power supply controller for charging piles and vehicle BMS according to claim 1, characterized in that, The power demand feature template includes a battery cell state encoding structure, a feature quantity time sequence arrangement set, and voltage and current weight parameters. The integrated parameter sequence includes multi-source data mapping relationships, feature matching result sets, and time sequence alignment parameters. The dynamic control instruction set includes stage division node numbers, power boundary condition sets, and control parameter indexes. The boundary state identifier includes safety threshold comparison results, fluctuation feature classification identifiers, and abnormal state codes. The optimized power supply strategy includes output waveform adjustment parameters, feature quantity matching relationship tables, and power supply mode switching instructions.

3. The integrated power supply controller for charging piles and vehicle BMS according to claim 1, characterized in that, The power demand modeling module includes: The feature extraction submodule acquires battery parameter data transmitted by the vehicle BMS, performs feature parsing operations on the parameter data, extracts the feature set of each battery cell, records the temporal position of each feature in the data stream, compares the relationship between the first occurrence position of key features in the parameter sequence and the data length, classifies them according to battery cell type, and obtains the feature distribution results. The sequence construction submodule extracts the data segments corresponding to the key features in the battery parameter data based on the feature distribution results, truncates the data based on the time interval of the feature in the data stream, constructs a feature fragment set based on the position of the data truncated for each feature, and reassembles it with the battery unit to which the feature belongs to obtain the feature time series set. The template generation submodule calculates the weight parameter values ​​of all features based on the time series set of features, performs time series rearrangement processing on the feature segment set based on the original order of features in the battery parameter data, and splices the time series segments of multiple features in the same battery cell according to the first occurrence position to obtain the power demand feature template.

4. The integrated power supply controller for charging piles and vehicle BMS according to claim 1, characterized in that, The multi-source integration module includes: Based on the power demand characteristic template, the data acquisition submodule combines the charging pile output parameter set and the environmental monitoring data stream to perform time-series alignment of all data sources according to their acquisition timestamps, establish a time-series alignment parameter table, and obtain multi-source data time-series index values. The fusion processing submodule obtains the aligned multi-source data set based on the multi-source data time series index value, performs feature matching operation on the multi-source data of each time series point, generates the integrated parameter value of each time series point using the feature weighted fusion method, and integrates the parameter values ​​of all time series points to generate an integrated parameter sequence. The verification feedback submodule compares the feature matching results with the feature quantity distribution in the original power demand feature template based on the integrated parameter sequence, generates data quality evaluation indicators, and feeds them back to the fusion processing submodule.

5. The integrated power supply controller for charging piles and vehicle BMS according to claim 1, characterized in that, The dynamic adjustment module includes: The stage division submodule analyzes the characteristic patterns of the battery charging stage based on the integrated parameter sequence, divides the charging stage intervals according to the characteristic pattern change nodes, marks the boundary time sequence point numbers of each stage, and generates a stage division node list. The power calculation submodule divides the node list according to the stage, collects the extreme values ​​of power demand in each stage interval, calculates the power transition threshold between adjacent stages, and generates a set of power boundary conditions in combination with the battery rated parameters. The instruction generation submodule matches the output capability parameters of the charging pile with the set of power boundary conditions, generates an index table of control parameters for each stage, and integrates the control parameters of all stages to generate a dynamic control instruction set.

6. The integrated power supply controller for charging piles and vehicle BMS according to claim 1, characterized in that, The security boundary determination module includes: Based on the dynamic control instruction set, the signal monitoring submodule collects the output waveform characteristic parameters of the charging pile and the battery status data fed back by the BMS in real time, and extracts the waveform fluctuation characteristic quantity and battery abnormal characteristic code. The threshold comparison submodule, based on the waveform fluctuation feature quantity and the battery abnormal feature code, calls the preset safety threshold parameter table, performs a step-by-step comparison operation between the feature quantity and the threshold parameter, and generates a feature state comparison result set. The status identifier generation submodule classifies the abnormality level of the feature status based on the feature status comparison result set, and generates boundary status identifiers according to the abnormality level encoding.

7. The integrated power supply controller for charging piles and vehicle BMS according to claim 1, characterized in that, The optimized output module includes: The strategy generation submodule analyzes the abnormal state distribution pattern based on the boundary state identifier and generates an output waveform adjustment parameter table by combining the characteristic quantity change trend in the power demand characteristic template. The output control submodule adjusts the driving parameters of the charging pile power output device based on the output waveform adjustment parameter table to match the power boundary conditions corresponding to the current charging stage. The adaptive adjustment submodule monitors the matching degree between the output waveform and the BMS feedback signal, and dynamically updates the output waveform adjustment parameters based on the matching degree deviation.

8. The integrated power supply controller for charging piles and vehicle BMS according to claim 7, characterized in that, The strategy generation submodule includes: The pattern analysis unit identifies the time interval and spatial distribution characteristics of the abnormal state distribution based on the abnormal state code in the boundary state identifier, and establishes an abnormal state distribution map. Based on the abnormal state distribution map, the parameter mapping unit queries the set of feature quantities for the corresponding time interval in the power demand feature template and establishes a mapping relationship table between abnormal states and feature quantities. The adjustment strategy unit generates a combination of output waveform adjustment parameters for different abnormal states based on the mapping table.

9. The integrated power supply controller for charging piles and vehicle BMS according to claim 7, characterized in that, The output control submodule includes: The drive parameter calculation unit converts the adjustment parameters into drive signal parameters of the power device based on the output waveform adjustment parameter table, including frequency modulation parameters and duty cycle adjustment parameters; The real-time adjustment unit dynamically adjusts the output value of the drive signal parameters according to the power boundary conditions corresponding to the current charging stage; The protection execution unit monitors the deviation between the actual output of the power device and the drive signal parameters, and initiates protective adjustment when the deviation exceeds the tolerance value.

10. The integrated power supply controller for charging piles and vehicle BMS according to claim 7, characterized in that, The adaptive adjustment submodule includes: The matching degree calculation unit collects the output waveform characteristics of the charging pile and the battery response data fed back by the BMS in real time, and calculates the matching degree index between the waveform characteristics and the battery response. The deviation analysis unit identifies the time interval and characteristic pattern of the deviation based on the deviation between the matching index and the preset target value. Based on the deviation analysis results, the parameter update unit dynamically updates the parameter values ​​in the output waveform adjustment parameter table.

Images