Intelligent charging pile system and charging control method

Abstract

The invention discloses an intelligent charging pile system and a charging control method, and relates to the field of charging piles, the intelligent charging pile system comprises a data acquisition module, a health assessment module and a charging strategy optimization module; the data acquisition module is used for acquiring basic state information of a battery of the electric vehicle in real time in the charging process of the electric vehicle, and performing feature extraction on the basic state information to obtain health features; the health assessment module is used for processing the health characteristics by adopting a deep learning model to obtain a real-time battery health state value and a residual life prediction value of the battery; and the charging strategy optimization module is used for performing real-time adjustment on the charging parameters based on the real-time battery health state value of the battery. According to the application, the requirements of plug-and-charge, charge-and-diagnose and diagnosis-and-charge cooperation can be met.

Description

Technical Field

[0001] This application relates to the field of charging piles, and in particular to an intelligent charging pile system and a charging control method. Background Technology

[0002] Current charging piles have limited functionality, providing only "power supply" without "diagnostic capabilities." Most mainstream charging piles only transmit energy and cannot detect battery aging. Furthermore, current technologies generally suffer from three major flaws: "separation of detection and charging," "disconnect between algorithms and hardware," and "lack of diagnostic feedback," failing to meet infrastructure requirements such as plug-and-charge, instant charging and diagnostics, and coordinated charging and diagnostics. Summary of the Invention

[0003] The purpose of this application is to provide an intelligent charging pile system and a charging control method that can meet the needs of plug-and-charge, charging and diagnosis, and diagnosis and charging coordination.

[0004] To achieve the above objectives, this application provides the following solution: In the first aspect, this application provides an intelligent charging pile system, including: a data acquisition module, a health assessment module, and a charging strategy optimization module.

[0005] The data acquisition module is used to acquire basic state information of electric vehicle batteries in real time during the charging process of electric vehicles, and to extract health features from the basic state information.

[0006] The health assessment module uses a deep learning model to process health features and obtain real-time battery health status and remaining life prediction values.

[0007] The charging strategy optimization module is used to adjust the charging parameters in real time based on the real-time battery health status value.

[0008] Secondly, this application provides a charging control method applied to the aforementioned intelligent charging pile system. The charging control method includes: acquiring basic state information of the electric vehicle battery in real time during the charging process of the electric vehicle, and extracting features from the basic state information to obtain health features.

[0009] A deep learning model is used to process health features to obtain real-time battery health status values ​​and predicted remaining life values.

[0010] The charging parameters are adjusted in real time based on the battery's real-time health status value.

[0011] According to the specific embodiments provided in this application, this application has the following technical effects: This application provides an intelligent charging pile system and a charging control method. The data acquisition module is used to acquire the basic state information of the electric vehicle battery in real time during the charging process, and extract health features from the basic state information. The health assessment module is used to process the health features using a deep learning model to obtain the real-time battery health status value and the predicted remaining life value of the battery. It can monitor the state of the lithium battery in real time and accurately while charging, meeting the needs of plug-and-charge and charge-and-diagnose. The charging strategy optimization module is used to adjust the charging parameters in real time based on the real-time battery health status value. Based on the real-time feedback of SOH, the charging parameters are adjusted in real time to meet the needs of diagnosis and charging coordination, extending battery life while ensuring charging speed. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 A flowchart illustrating the use and operation of the intelligent charging pile system provided in this application embodiment.

[0014] Figure 2 A flowchart of the fast protocol adaptive identification method provided in the embodiments of this application.

[0015] Figure 3 This is a flowchart illustrating the internal processing of an intelligent charging pile system provided in an embodiment of this application. Detailed Implementation

[0016] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0017] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0018] In one exemplary embodiment, an intelligent charging pile system is provided, including: a data acquisition module, a health assessment module, and a charging strategy optimization module.

[0019] The data acquisition module is used to acquire basic state information of electric vehicle batteries in real time during the charging process of electric vehicles, and to extract health features from the basic state information.

[0020] The health assessment module uses a deep learning model to process health features and obtain real-time battery health status and remaining useful life (RUL) prediction values.

[0021] The charging strategy optimization module is used to adjust the charging parameters in real time based on the real-time battery health status value.

[0022] In another exemplary embodiment of this application, the basic state information of the battery includes the battery's voltage, current, and temperature; the data acquisition module specifically includes: a voltage acquisition unit, a current Hall sensor, and a temperature sensor interface.

[0023] The voltage acquisition unit is used to acquire the battery's voltage curve in real time.

[0024] A current Hall sensor is used to acquire the battery's current curve in real time.

[0025] The temperature sensor interface is used to collect the temperature at each sampling point of the battery in real time.

[0026] The filtering module is used to filter the basic state information of the battery.

[0027] The feature extraction module is used to extract features from the basic state information of the filtered battery.

[0028] In another exemplary embodiment of this application, the feature extraction module includes a time-domain feature extraction submodule and a temperature feature extraction submodule.

[0029] The time-domain feature extraction submodule is used to process the filtered battery voltage and current curves to obtain the peak value, slope, and integral area of ​​the voltage curve, as well as the peak value, slope, and integral area of ​​the current curve.

[0030] The temperature feature extraction submodule is used to process the temperature at each sampling point of the filtered battery to obtain the temperature rise rate and thermal distribution gradient.

[0031] In another exemplary embodiment of this application, the deep learning model includes: a shared encoder, a summing module, and a parallel Elastic Net regularized linear regression model and a Micro-Transformer module; the output of the Elastic Net regularized linear regression model and the first output of the Micro-Transformer module are connected through the summing module, and the second output of the Micro-Transformer module is connected to the shared encoder. The first output of the Micro-Transformer module outputs a battery health status value. The shared encoder outputs a predicted remaining lifetime value.

[0032] In another exemplary embodiment of this application, the shared encoder is a single-layer MLP.

[0033] In another exemplary embodiment of this application, the intelligent charging pile system further includes: a two-way communication and protocol adaptation module; the electric vehicle battery management system is connected to the data acquisition module through the two-way communication and protocol adaptation module.

[0034] In practical applications, the intelligent charging pile system further includes: a main control module, a two-way communication and protocol adaptation module, and a data acquisition module connected through the main control module. The two-way communication and protocol adaptation module obtains basic status information of the electric vehicle battery from the electric vehicle battery management system and sends it to the main control module. The main control module then sends the basic status information of the electric vehicle battery to the data acquisition module.

[0035] In another exemplary embodiment of this application, the intelligent charging pile system further includes: a cloud data management platform; the data acquisition module, the health assessment module, and the charging strategy optimization module are all connected to the cloud data management platform.

[0036] In another exemplary embodiment of this application, the charging parameters are adjusted in real time based on the real-time battery health status value. Specifically, if the battery health status value is greater than or equal to a first set threshold, the charging power is adjusted to the maximum allowable current and the charging power is adjusted to full power.

[0037] If the second set threshold is less than or equal to the battery health status value and less than the first set threshold, the charging current will be adjusted to 0.8 times the standard value and the charging power will be adjusted to 90% of the full power.

[0038] If the battery health status value is less than the second set threshold, the charging current will be adjusted to 0.7 times the standard value and the charging power will be adjusted to 70% of the full power.

[0039] In another exemplary embodiment of this application, the bidirectional communication and protocol adaptation module includes three parallel signal monitoring channels; the first signal monitoring channel acquires the CP-PWM signal of the GB / T protocol, the second signal monitoring channel acquires the PLC-SLAC initialization signal of the CCS protocol, and the third signal monitoring channel acquires the basic CAN frame in the CHAdeMO standard.

[0040] This application provides an intelligent charging pile system for electric vehicles that integrates health monitoring and adaptive charging. Applicable to public fast charging stations, battery swapping stations, and residential communities, the system comprises a data acquisition module, a health assessment module, a charging strategy optimization module, a two-way communication and protocol adaptation module, and a cloud-based data management platform. During vehicle charging, the system continuously acquires battery operating status, multimodal operating characteristics, and thermal behavior information, enabling online assessment of the battery's State of Health (SOH). Based on the assessment results, it dynamically adjusts the charging power and charging curve shape, thereby significantly improving battery life and charging safety while ensuring charging efficiency. The entire system is centered on a main control module. By establishing a two-way communication link between the charging pile and the vehicle's battery management system (BMS), it achieves high-frequency data acquisition, health status inference, strategy optimization execution, and full lifecycle data management, constructing a closed-loop intelligent charging technology system of "perception-diagnosis-control." The main control module is connected to the data acquisition module, health assessment module, charging strategy optimization module, and two-way communication and protocol adaptation module.

[0041] During system operation, the data acquisition module undertakes the task of real-time sensing of key underlying parameters. This module integrates high-precision voltage and current sampling units and a temperature sensor interface, enabling it to synchronously acquire and capture battery terminal voltage curves, current curves, and temperatures at various sampling points on the battery with millisecond-level time resolution. The data acquisition module processes the raw signals through built-in filtering and feature extraction algorithms, ensuring the consistency, accuracy, and traceability of the data input to the backend. Furthermore, the system can apply small-amplitude AC perturbation signals during the constant current phase or under specific conditions. The data acquisition module then synchronously samples the voltage and current curves with high precision. The time-domain feature extraction submodule extracts the frequency domain impedance characteristics of the battery (peak value, slope, and integral area of ​​the voltage curve, and peak value, slope, and integral area of ​​the current curve), providing higher-resolution aging characteristics for the health assessment module.

[0042] Based on data obtained from the data acquisition module, the health assessment module can analyze the battery's aging status in real time, generate a State of Health (SOH) index, and identify potential risks. This module employs a lightweight neural network model, enabling the system to perform real-time health inference under low computing power. The health assessment module not only generates the current SOH value but also performs regression predictions on the battery's remaining lifespan.

[0043] After the health status is inferred, the charging strategy optimization module dynamically adjusts the charging parameters based on the current State of Health (SOH), achieving true adaptive charging control. When the system detects that the SOH has dropped to a set threshold, the strategy optimization module automatically reduces the charging power and adjusts the constant current stage current to ensure the battery is in a controlled charging environment best suited to its current health status. All adjustments are executed in real time under a closed-loop control mechanism and maintain communication synchronization with the vehicle's Battery Management System (BMS) to ensure the accuracy and stability of the strategy implementation. After charging is completed, the system uploads the full-cycle collected data, health assessment results, and strategy execution records to the cloud data management platform for subsequent battery life trend modeling, cross-user strategy optimization, and sharing of charging equipment models across the entire network, thereby further improving the overall system performance.

[0044] The use and operation process of the intelligent charging pile system provided in this application are as follows: Figure 1 As shown, the process includes: Step 1: When inserting the charging gun, simultaneously connect the battery pack and the charging pile measurement system (intelligent charging pile system). The system completes protocol adaptive recognition within 2 seconds.

[0045] Step Two: The main control module checks the BMS, data acquisition module, and communication module (bidirectional communication and protocol adaptation module) for proper functioning and reads basic battery information. The main control module loads sensor configurations, data sampling plans, and safety thresholds, placing the smart charging pile system into high-frequency monitoring mode to prepare for subsequent health inference and dynamic strategy execution. Basic battery information includes: battery type, battery model, rated capacity, maximum continuous charging rate, and maximum continuous discharging rate.

[0046] Step 3: The data acquisition module synchronously acquires information such as battery terminal voltage, current and temperature with millisecond-level time resolution, and calculates the peak value, slope and integral area of ​​the voltage curve and current curve, as well as the temperature rise rate and other characteristics.

[0047] Step 4: Health Assessment Module ( Figure 1 and Figure 2 The AI ​​algorithm module (shown in the middle) uses features collected from the battery to make real-time inferences about the battery's aging status. Employing a lightweight deep learning model, the system can achieve online inference of battery health status and prediction of remaining lifespan under low computing power conditions, ensuring data quality and traceability. The health assessment module can also determine basic information such as battery type and identify the battery based on the basic battery information uploaded by the main control module.

[0048] Step 5: The battery health status assessment results (battery health status value) are promptly fed back to the charging strategy optimization module. Based on the health assessment results, key parameters are adjusted in real time to trigger an adaptive charging strategy.

[0049] Step Six: Generate a digital certification report based on the remaining battery life prediction and push it to the user's app. Upload the data to the cloud data management platform and iterate the lightweight deep learning model weekly to improve its cross-vehicle generalization capabilities. The digital certification report includes basic battery information, battery health status, remaining battery life prediction, and safety and risk warnings.

[0050] This application provides a more specific intelligent charging pile system, and provides a detailed description of the above-mentioned intelligent charging pile system. The intelligent charging pile system in this embodiment takes the charging pile side main control module as the core, and realizes continuous monitoring of electric vehicle power battery and execution of optimal charging strategy through parallel data acquisition link, health inference link and strategy control link.

[0051] The data acquisition module includes a high-precision voltage acquisition unit, a current Hall sensor, and a temperature sensor interface. The module has millisecond-level sampling capability, enabling it to capture the vehicle battery's voltage curves, current curves, and temperature information in real time.

[0052] The health assessment module uses an embedded lightweight deep learning model to fuse and calculate features, outputting real-time SOH values ​​and predicted remaining lifespan. This module can run on low-computing-power platforms and completes inference within milliseconds, enabling online health monitoring.

[0053] The charging strategy optimization module adjusts charging parameters in real time based on the State of Harm (SOH) value output by the health assessment module, including charging power, constant current stage current magnitude, and constant voltage stage cutoff voltage. The module incorporates a health-level control strategy, applying different charging strategies to three SOH levels. Specifically, this strategy considers an SOH ≥ 90% as a healthy state, capable of withstanding standard fast charging and performing full-power charging; when SOH ≤ 80% < 90% is considered an early degradation state, requiring moderate load reduction, with the charging current adjusted to 0.8 times the standard value, performing 90% power charging; and when SOH < 80% is considered an accelerated aging state, the current is adjusted to 0.7 times the standard value, requiring significant protection, and performing 70% power charging.

[0054] The bidirectional communication module (bidirectional communication and protocol adaptation module) is responsible for information exchange between the main control module and the vehicle BMS. Through the "open diagnostic link" mechanism based on public signals, it achieves real-time capture and synchronous updates of key operating states. The cloud-based data management platform is used to record full-cycle data, model inference results, and strategy execution logs, enabling cross-user and cross-site data sharing and health model updates.

[0055] Work process as follows Figure 1 As shown, after the user inserts the charging gun, the system completes protocol identification within 2 seconds, recognizes different protocols, and responds uniformly. After connecting to the charging pile, the BMS reports a set of basic information to the pile according to the protocol, and the pile, as the receiving end, records the charging start state. The data acquisition module is activated to extract health characteristics of voltage, current, and temperature, and outputs SOH and remaining lifespan through a deep learning model, triggering an adaptive charging strategy. At the same time, a digital authentication report is generated and pushed to the user's APP. The entire process is seamless for the user, and the charging experience is consistent with traditional charging piles. This embodiment, through hardware-software-cloud collaboration, achieves charging as a health check, charging as maintenance, and charging as authentication for the first time, injecting new intelligent momentum into new energy infrastructure.

[0056] like Figure 2 As shown, this application proposes a rapid protocol adaptive identification method (bidirectional communication and protocol adaptation module) for multi-standard electric vehicles, aiming to solve the compatibility bottleneck commonly faced by charging piles in international applications. Different countries and automakers use significantly different DC charging protocols, and traditional smart charging pile systems often rely on the vehicle's BMS's private fields or encrypted diagnostic interfaces to correctly identify vehicle requirements, leading to difficulties in cross-model and cross-regional deployment. This application constructs a "protocol-independent" identification mechanism, enabling the charging pile to accurately determine the charging standard used and extract key public fields within approximately 2 seconds of vehicle connection, achieving plug-and-play automated health monitoring and control.

[0057] This application designs three parallel signal monitoring channels to acquire CP-PWM signals from the GB / T protocol, PLC-SLAC initialization signals from the CCS protocol, and basic CAN frames from the CHAdeMO standard, respectively. The main control unit simultaneously starts three monitoring threads upon system startup to capture the characteristics of each channel in real time, achieving millisecond-level asynchronous parallel recognition.

[0058] The "protocol independence" identification mechanism can be divided into three layers. The first layer is "protocol feature judgment," where the bidirectional communication and protocol adaptation module quickly classifies the vehicle's corresponding protocol family based on the publicly available characteristics of the protocols. The second layer is "protocol structure parsing," where the bidirectional communication and protocol adaptation module automatically switches to the matching parser to extract publicly available parameters such as vehicle required voltage, current, battery temperature, maximum allowable power, and stage status from the initial negotiation message. The third layer is "unified protocol format," which maps heterogeneous fields output by different protocols into a unified session structure, allowing it to be directly used by the backend health assessment module and charging strategy optimization module without needing to be aware of underlying protocol differences. Through this module, the charging pile system can achieve rapid adaptation to electric vehicles across brands, providing a reliable foundation for intelligent health monitoring, adaptive charging strategies, and cross-regional deployment, significantly improving the system's versatility and engineering feasibility.

[0059] For the health assessment module, the process is as follows: Figure 3 As shown, this module obtains the real-time battery health status value and remaining life prediction value through feature data (health features). The raw signal is preprocessed through built-in filtering and feature extraction algorithms to obtain feature data. First, the filtering algorithm removes noise interference to ensure signal smoothness; the feature extraction stage further mines the value of the signal, extracting time-domain features such as peak value, slope, and integral area from the voltage and current curves, and calculating the temperature rise rate and thermal distribution gradient from the temperature data.

[0060] For the deep learning model of the health assessment module, this application proposes a two-stage hybrid architecture of "Elastic Net + Micro-Transformer". The first stage uses Elastic Net regularized linear regression as a globally interpretable baseline model, and the innovative design of "Elastic Net + Micro-Transformer" cleverly combines the interpretability of linear models with the powerful expressive power of nonlinear neural networks. The first stage uses the ElasticNet regularized linear regression model as a globally interpretable baseline model. This model effectively handles the collinearity problem in multidimensional feature data by combining L1 and L2 regularization terms, while achieving feature selection and parameter shrinkage. This makes the model highly robust and interpretable when capturing the linear relationship in the battery aging process. However, battery aging often involves complex nonlinear dynamics, which are difficult for these linear models to fully model. To make up for this deficiency, the output of the first stage is used as the basis, and the second stage introduces a Micro-Transformer module to generate high-precision correction terms.

[0061] The Micro-Transformer module is a miniaturized Transformer architecture (which employs known methods to sequentially prune, quantize, schedule memory, and accelerate operators of the original Transformer) specifically optimized for embedded systems. It inherits the mechanisms of the Transformer, enabling it to capture long-term dependencies and non-linear interactions in data sequences without requiring a large parameter size.

[0062] The vector composed of health features of the Elastic Net regularized linear regression model at time t. The above method achieves a fast, sparse, and robust initial estimate of SOH, yielding the original SOH (SOH0). Its optimization objective function is: ,in, .

[0063] W is the model's parameter vector (weight vector); argmin() represents taking the minimum value; X represents the laboratory-calibrated true SOH value; X is the historical feature matrix. The regularization strength; This represents the weight ratio between L1 and L2 regularization. || ||1 represents the L1 norm, || || 2 denoted by , and b denotes the square of the norm.

[0064] To compensate for the limitations of linear models in modeling nonlinear aging dynamics, this stage introduces a Micro-Transformer module to process the feature sequence (a vector composed of health features at time t). Perform temporal residual learning and output high-precision correction terms. The Micro-Transformer module employs relative position encoding to avoid dependence on absolute timestamps, improving generalization ability across SOC ranges. Output residual correction is also implemented. Final SOH prediction: .

[0065] Design a RUL regression head (shared encoder) using Transformer's multi-head attention mechanism, sharing the last hidden state of Transformer, integrating with a single-layer MLP, and combining the current SOH with the cumulative loop count synchronized with the cloud. Perform regression: .

[0066] The network architecture of this application consists of two stages. The first stage, an Elastic Net regularized linear regression model, provides a fast and interpretable linear baseline. The second stage, a Micro-Transformer module, achieves high-precision nonlinear refinement through temporal residual learning. Together, they form a fast, accurate, and cost-effective inference chain. This model has powerful temporal modeling capabilities, effectively capturing nonlinear and long-range dependent aging dynamics, and significantly improving prediction accuracy under complex operating conditions.

[0067] This application proposes a closed-loop adaptive charging strategy based on real-time feedback of battery health status, aiming to maximize battery cycle life while ensuring fast charging speed. Traditional fixed CC-CV (constant current-constant voltage) charging ignores battery aging status, leading to accelerated SEI film growth, lithium dendrite formation, and cathode structure collapse induced by high-power charging when SOH < 80%, increasing the capacity decay rate by 15%–30%. This strategy addresses this by: when SOH ≥ 90%, the battery is in a healthy state and can withstand standard fast charging, allowing for full-power charging; when SOH ≤ 80% and < 90%, the battery is in an early degradation state, requiring appropriate load reduction, adjusting the charging current to 0.8 times the standard value, and charging at 90% power; when SOH < 80%, the battery is in an accelerated aging state, adjusting the current to 0.7 times the standard value, requiring significant protection, and charging at 70% power. The SOH output by the health assessment module is used as a decision variable to dynamically adjust the charging power, constant current stage current magnitude, and constant voltage charging time, achieving precise matching. This strategy utilizes a mathematical control model driven by segmented SOH thresholds to achieve a shift from "passive charging" to "adaptive charging." To ensure the long-term accuracy stability and continuous performance improvement of the smart charging pile system under global heterogeneous vehicle fleets, diverse chemical systems, and dynamic aging conditions, this system constructs a cloud-edge-end-to-end collaborative autonomous evolutionary closed loop. This mechanism uses anonymous encrypted data uploads as the core input and a cross-chemical system generalization model as the output, enabling the system to continuously evolve with global vehicle fleet data, moving from static deployment to an adaptive learning ecosystem. As the system grows with global vehicle fleet data, SOH errors are optimized and reduced, and lifespan is extended and improved, thoroughly constructing a positive cycle ecosystem of "data-intelligence-lifespan," laying a reliable, evolvable, and profitable intelligent infrastructure for the trillion-dollar battery lifecycle service market.

[0068] The system workflow includes: Step 1: When charging is started, the main control module initializes the system and establishes a communication link between the battery and the charging pile.

[0069] Step 2: The signal acquisition module collects multi-dimensional charging characteristic parameters in real time.

[0070] Step 3: The health assessment module calculates the battery's state of health (SOH) and determines whether there is any abnormal degradation.

[0071] Step 4: If performance degradation is detected, the charging strategy optimization module dynamically adjusts the charging power, etc.

[0072] Step 5: The system completes adaptive charging and records full-cycle data for subsequent health trend modeling and cloud-based collaborative optimization.

[0073] Compared with the prior art, this application has the following technical effects: First, it realizes the advanced technology of simultaneous charging and inspection.

[0074] Traditional State of Health (SOH) testing relies on full-capacity charge-discharge tests in laboratories, taking 4-8 hours, requiring specialized equipment and technicians, and is costly and not feasible for routine use. This application, however, uses a non-invasive, high-frequency charging gun to acquire raw voltage-current-temperature curves, combined with a deep learning model at the charging station for data analysis and feature extraction, enabling prediction of battery health status and remaining battery life. Integrating real-time SOH detection, it overcomes the limitations of traditional passive charging, transforming the charging station from an "energy pump" into a "battery health manager."

[0075] II. Overcome the data barriers of BMS.

[0076] Electrochemical impedance spectroscopy (EIS) for SOH monitoring in charging stations heavily relies on raw data uploaded by the vehicle's BMS, including SOH and temperature profiles. This dependence on BMS authorization leads to a complete loss of autonomous detection capabilities at the charging pile level, creating severe data silos. Furthermore, current charging pile SOH detection requires access to a private ID or diagnostic service to obtain core data such as raw voltage, internal resistance, state of charge, and SOH, posing a high risk of privacy breaches. However, OEMs generally block or encrypt these interfaces for algorithm protection, business competition, and privacy compliance. This application is the first to use only standard protocol public signals and a 2-second protocol adaptive recognition mechanism to enable plug-in testing of any global brand EV without requiring OEMs to provide any private IDs, fundamentally eliminating data silos and privacy disputes. The bidirectional communication and protocol adaptation module automatically generates a standardized session structure containing only publicly available fields (maximum current, voltage, and battery chemistry type inference), completely shielding the heterogeneity of the underlying protocols. This enables subsequent algorithms to perform protocol-independent reasoning. This solution allows smart charging pile systems to be globally universal, with single firmware compatible with multiple mainstream standards, reducing the deployment cycle from 12 months to 1 week. Furthermore, it enables an open ecosystem, outputting standardized health interfaces to provide a trusted data foundation for public charging networks, battery swapping operators, and insurance companies, fostering a trillion-dollar battery lifecycle service market. It can adaptively identify charging protocols for various publicly available signals, extracting basic vehicle charging parameters from signal messages and mapping them to a unified session structure that shields the differences in underlying protocols. This allows charging piles to understand the charging requests of any electric vehicle in a short time.

[0077] III. Implement optimized charging strategies.

[0078] Electrochemical impedance spectroscopy (EIS) for SOH monitoring at charging stations only provides charging station location recommendations without any intervention, failing to adjust charging strategies in real time to prevent accelerated battery degradation. This application pioneers a closed-loop optimized charging strategy driven by real-time SOH feedback. It instantly converts detection results into execution instructions, dynamically adjusting charging power, constant current stage current magnitude, and appropriate constant voltage stage cutoff time. This ensures fast charging speed while extending battery life. The user app generates health recommendations based on the battery health status in the digital certification report and pushes these recommendations in real time (e.g., "SOH decreased by 1.2% this month, low-temperature slow charging is recommended"), forming a positive data loop between users, charging stations, and the cloud. This closed-loop optimized charging strategy not only extends individual battery life but also provides reliable data on remaining battery life for secondary use, battery swapping, and battery insurance, elevating charging stations from "energy pumps" to "battery health stewards," and opening up a trillion-dollar battery lifecycle service market.

[0079] IV. Electrochemical impedance spectroscopy (EHS) is highly sensitive to environmental conditions. Temperature, voltage window, and SOC distribution all affect the shape of the impedance curve. Although charger-based battery SOH testing devices propose a "specific frequency impedance real part ratio" to reduce SOC dependence, they still exhibit unstable responses under low temperature, fast charging, and high-power pulsation scenarios. Especially at 10Hz high-frequency recording, the rapid transient charging current superimposed with the AC excitation signal causes noise peaks in the impedance calculation model, resulting in fluctuations in SOH estimation during certain cycles. This application effectively suppresses noise and enhances robustness through filtering and feature extraction.

[0080] V. Machine Learning-Driven Charging Pile SOH Evaluation System Due to the extremely small size of the TinyML model, its feature representation ability is limited, and it can only handle a small number of input variables, such as current integral, transient voltage step, and equivalent internal resistance. Under complex operating conditions, such as changes in aging rate, migration across battery models, and missing multimodal data, the model's generalization ability is insufficient, making it difficult to achieve the expressive power of deep learning models and identify multiple degradation modes or capture rapid failure trends in advance. Unlike TinyML's limited variables, this application's data acquisition module captures multimodal data at millisecond-level resolution; full-cycle data (acquisition data, health results, strategy records) is uploaded to the cloud platform, and the lightweight model is iterated weekly to solve the problem of insufficient model generalization under complex operating conditions, making it difficult to achieve the expressive power of deep learning models and identify multiple degradation modes or capture rapid failure trends in advance.

[0081] VI. Electrochemical impedance spectroscopy for SOH monitoring in charging stations lacks a local certification mechanism, and the recommended results cannot generate compliance reports that meet standards such as UL and CCC, still requiring additional manual verification; this application can generate digital certification reports and push them to the user's APP and upload the data to the cloud.

[0082] This application also provides a charging control method for the above-mentioned intelligent charging pile system, the method comprising: acquiring basic state information of electric vehicle battery in real time during electric vehicle charging process, and extracting health features from the basic state information.

[0083] A deep learning model is used to process health features to obtain real-time battery health status values ​​and predicted remaining life values.

[0084] The charging parameters are adjusted in real time based on the battery's real-time health status value.

[0085] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0086] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0087] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A smart charging pile system, characterized in that, The intelligent charging pile system includes: Data acquisition module, health assessment module, and charging strategy optimization module; The data acquisition module is used to acquire basic state information of electric vehicle batteries in real time during the charging process of electric vehicles, and to extract health features from the basic state information. The health assessment module is used to process health features using a deep learning model to obtain real-time battery health status values ​​and remaining life prediction values. The charging strategy optimization module is used to adjust the charging parameters in real time based on the real-time battery health status value.

2. The intelligent charging pile system according to claim 1, characterized in that, Basic state information of a battery includes its voltage, current, and temperature; The data acquisition module specifically includes: Voltage acquisition unit, current Hall sensor, and temperature sensor interface; The voltage acquisition unit is used to acquire the battery's voltage curve in real time; A current Hall sensor is used to acquire the battery's current curve in real time. The temperature sensor interface is used to collect the temperature at each sampling point of the battery in real time; The filtering module is used to filter the basic state information of the battery; The feature extraction module is used to extract features from the basic state information of the filtered battery.

3. The intelligent charging pile system according to claim 1, characterized in that, The deep learning model includes a shared encoder, a summing module, and a parallel Elastic Net regularized linear regression model and a Micro-Transformer module. The output of the Elastic Net regularized linear regression model and the first output of the Micro-Transformer module are connected through the summing module, and the second output of the Micro-Transformer module is connected to the shared encoder.

4. The intelligent charging pile system according to claim 1, characterized in that, The intelligent charging pile system also includes: a two-way communication and protocol adaptation module; the electric vehicle battery management system is connected to the data acquisition module through the two-way communication and protocol adaptation module.

5. The intelligent charging pile system according to claim 1, characterized in that, The intelligent charging pile system also includes: a cloud data management platform; the data acquisition module, health assessment module, and charging strategy optimization module are all connected to the cloud data management platform.

6. The intelligent charging pile system according to claim 2, characterized in that, The feature extraction module includes: a time-domain feature extraction submodule and a temperature feature extraction submodule; The time-domain feature extraction submodule is used to process the filtered battery voltage and current curves to obtain the peak value, slope, and integral area of ​​the voltage curve, as well as the peak value, slope, and integral area of ​​the current curve. The temperature feature extraction submodule is used to process the temperature at each sampling point of the filtered battery to obtain the temperature rise rate and thermal distribution gradient.

7. The intelligent charging pile system according to claim 1, characterized in that, The charging parameters are adjusted in real time based on the battery's real-time health status value, specifically as follows: If the battery health status value is greater than or equal to the first set threshold, the charging power will be adjusted to the maximum allowable current and the charging power will be adjusted to full power. If the second set threshold is less than or equal to the battery health status value and less than the first set threshold, then the charging current will be adjusted to 0.8 times the standard value and the charging power will be adjusted to 90% of the full power. If the battery health status value is less than the second set threshold, the charging current will be adjusted to 0.7 times the standard value and the charging power will be adjusted to 70% of the full power.

8. The intelligent charging pile system according to claim 4, characterized in that, The bidirectional communication and protocol adaptation module includes three parallel signal monitoring channels; the first signal monitoring channel acquires the CP-PWM signal of the GB / T protocol, the second signal monitoring channel acquires the PLC-SLAC initialization signal of the CCS protocol, and the third signal monitoring channel acquires the basic CAN frame in the CHAdeMO standard.

9. The intelligent charging pile system according to claim 3, characterized in that, The shared encoder is a single-layer MLP.

10. A charging control method, characterized in that, The charging control method is applied to the intelligent charging pile system according to any one of claims 1-9, and the charging control method includes: During the charging process of electric vehicles, the basic state information of the electric vehicle battery is acquired in real time, and health features are obtained by extracting features from the basic state information. A deep learning model is used to process health features to obtain real-time battery health status values ​​and remaining life prediction values. The charging parameters are adjusted in real time based on the battery's real-time health status value.

Images