An adaptive conversion method and system for electric vehicle charging protocols

Abstract

This invention provides an adaptive conversion method and system for electric vehicle charging protocols, relating to the field of electric vehicle charging technology. The method includes: constructing a two-level heterogeneous computing architecture; capturing the micro-current ripple characteristic waveforms of the communication carrier during the charging connector coupling and communication handshake phases in real time; obtaining a multi-protocol parallel conflict risk sequence; and collecting the full-dimensional operating state vector of the charging pile. It also involves generating an initial protocol fingerprint probability tensor, analyzing it in conjunction with the multi-protocol parallel conflict risk sequence, generating a protocol fusion driving vector, analyzing the nonlinear coupling relationship between the full-dimensional operating state vector of the charging pile and the protocol fusion driving vector, predicting the evolution trajectory of the optimal protocol parameters, performing comparative simulation, and generating a reconfigurable bitstream slice sequence. This invention solves the technical problems of low protocol compatibility and flexibility in existing technologies. It achieves the technical effects of real-time and accurate identification of protocol conflicts, dynamic generation of optimal protocol parameters, and improved protocol adaptability and compatibility.

Description

Technical Field

[0001] This invention relates to the field of electric vehicle charging technology, and specifically to an adaptive conversion method and system for electric vehicle charging protocols. Background Technology

[0002] With the rapid growth of electric vehicle ownership, V2G (Vehicle-to-Grid) technology is gradually becoming an important means of achieving two-way interaction between vehicles and the power grid. However, differences in communication protocols and power control mechanisms among different automakers, national standards, and charging equipment lead to frequent protocol incompatibility issues, severely restricting charging interoperability and system scalability. Existing charging piles typically achieve protocol matching through fixed protocol stacks or manual configuration, lacking the ability to adapt to changes in the operating environment and dynamic demands of power grid scheduling. When protocols are upgraded or new protocol types are added, manual firmware updates or system downtime maintenance are required, resulting in long response cycles and low efficiency.

[0003] Furthermore, current protocol conflict detection methods largely rely on offline testing or rule matching, making it difficult to predict potential conflicts in scenarios with multiple protocols running in parallel, and lacking the ability to proactively analyze real-time scheduling changes. Meanwhile, traditional charging control systems mostly operate at the software level, lacking a dynamic execution mechanism that integrates with reconfigurable hardware. This prevents the use of hardware resources such as FPGAs to achieve online reconfiguration and rapid switching of protocol modules, resulting in insufficient system flexibility.

[0004] Existing technologies suffer from technical problems such as protocol adaptation relying on manual configuration, delayed conflict prediction, and the inability to achieve hardware-level dynamic reconfiguration, resulting in low compatibility and flexibility. Summary of the Invention

[0005] The purpose of this application is to provide an adaptive conversion method and system for electric vehicle charging protocols, which solves the technical problems of low compatibility and flexibility caused by the reliance on manual configuration for protocol adaptation, lag in conflict prediction, and inability to achieve hardware-level dynamic reconfiguration in the prior art.

[0006] In view of the above problems, this application provides an adaptive conversion method and system for electric vehicle charging protocols.

[0007] The first aspect of this application provides an adaptive conversion method for electric vehicle charging protocols. This method includes: deploying a high-frequency micro-current sensor array and a heterogeneous computing protocol fusion unit in a V2G smart charging pile with protocol reconfigurability to construct a two-level heterogeneous computing architecture of protocol perception and protocol fusion; based on the two-level heterogeneous computing architecture, capturing the micro-current ripple characteristic waveform of the communication carrier during the charging connector coupling and communication handshake stages in real time using the high-frequency micro-current sensor array; simultaneously, obtaining the multi-protocol parallel conflict risk sequence within future scheduling cycles through a V2G protocol conflict prediction digital twin deployed in the cloud; and synchronously collecting the full-dimensional operating status of the charging pile through an embedded system within the pile. Based on the microcurrent ripple characteristic waveform of the communication carrier, super-resolution feature extraction is performed to generate an initial protocol fingerprint probability tensor. This tensor is then combined with the multi-protocol parallel conflict risk sequence for spatiotemporal alignment and correlation analysis to dynamically identify key reconfigurable protocol modules and their expected evolution paths, and to generate a protocol fusion driving vector. Based on the charging pile's full-dimensional operating state vector and the protocol fusion driving vector, their nonlinear coupling relationship is analyzed to predict the optimal protocol parameter evolution trajectory under FPGA partial reconfigurability constraints. Based on the optimal protocol parameter evolution trajectory, real-time in-loop compatibility comparison and conflict resolution simulation with the target V2G service protocol are performed in a high-fidelity virtual interactive environment in the cloud, dynamically generating a FPGA partial reconfigurable bitstream slice sequence.

[0008] Optionally, for the microcurrent ripple characteristic waveform of the communication carrier, a complex wavelet transform with a sliding time window is used to extract multidimensional protocol fingerprint features within each time window; through a lightweight protocol fingerprint matching engine, the multidimensional protocol fingerprint features are compared with the protocol feature knowledge graph updated by federated learning to calculate similarity, and a multi-protocol confidence distribution matrix indexed by time is output as the initial protocol fingerprint probability tensor; the multi-protocol parallel conflict risk sequence is decoded into protocol state transition constraints, and the constraints are satisfied by solving the multi-protocol confidence distribution matrix to identify key reconfigurable protocol modules and predict their adjustment direction and magnitude, forming a protocol fusion driving vector.

[0009] Optionally, a composite prediction model is established that includes the full-dimensional operating state vector of the charging pile and the protocol fusion driving vector to analyze the coupling relationship between state disturbances and protocol requirements and predict the adjustment trend of protocol parameters. Based on the adjustment trend, combined with the hardware constraint model of the partially reconfigurable region of the FPGA, a discrete protocol state transition sequence is generated as the optimal protocol parameter evolution trajectory.

[0010] Optionally, the time-frequency domain decoupling of the full-dimensional operating state vector of the charging pile is performed to separate the steady-state baseline component and the transient disturbance component; the protocol fusion driving vector is decomposed into protocol semantic layer adjustment requirements and communication physical layer adjustment requirements; a disturbance-demand coupled state space equation is established, with the transient disturbance component as input and the rate of change of the protocol fusion driving vector as the state variable, and state estimation is performed through a Kalman filter to solve for the optimal adjustment acceleration of the protocol parameters.

[0011] Optionally, a hardware constraint model is established based on the size of the reconfigurable region of the FPGA and the bitstream reconfiguration delay parameters; the optimal adjustment acceleration of the protocol parameters is input into the hardware constraint model to calculate the feasible protocol parameter adjustment increment within the hardware resource and delay boundaries; based on the current protocol state and the protocol parameter adjustment increment, the protocol state coordinates of the future time series are iteratively deduced; the protocol state coordinates are sequence optimized to generate a continuous and hardware-realizable optimal protocol parameter evolution trajectory.

[0012] Optionally, a high-fidelity virtual interactive environment synchronized with physical charging piles, vehicle BMS, and power grid dispatch instructions is constructed; the evolution trajectory of the optimal protocol parameters is injected into the virtual interactive environment in the form of incremental protocol configuration instruction streams to drive the dynamic reconstruction of the virtual protocol stack; the target V2G service protocol is simulated and executed in the virtual interactive environment, and the difference between the output behavior of the virtual protocol stack and the expected behavior of the target protocol is compared in real time to generate a multi-dimensional difference vector; closed-loop feedback optimization is performed based on the multi-dimensional difference vector to dynamically adjust the incremental protocol configuration instruction streams until the multi-dimensional difference vectors converge to a preset threshold range; the finally converged protocol configuration instruction streams are compiled into a partially reconfigurable bitstream slice sequence for the FPGA.

[0013] Optionally, in the high-fidelity virtual interactive environment, a V2G service response simulation model is constructed; the power grid dispatch command is used as the input stimulus, and the power parameters are dynamically adjusted based on the simulated vehicle SOC state. The V2G service response simulation model is run to obtain the standard protocol interaction process; the dynamic protocol stack behavior driven by the incremental instruction stream of the protocol configuration is compared with the standard protocol interaction process in a spatiotemporal alignment; the difference values ​​of the comparison results in multiple dimensions such as protocol parameters, communication timing, and power response curve are calculated to generate the multidimensional difference vector.

[0014] Optionally, a digital twin closed-loop control model is constructed, including a protocol parameter regulator and a communication timing compensator; the multidimensional difference vector is input into the digital twin closed-loop control model to calculate the protocol stack parameter correction amount and the communication timing compensation amount; the protocol stack parameter correction amount and the communication timing compensation amount are rolled optimization through a model predictive control algorithm and iteratively simulated in a virtual environment; when the simulation results show that each component of the multidimensional difference vector is less than a preset threshold, the protocol configuration optimization is determined to be complete, and the current protocol configuration incremental instruction stream is locked.

[0015] Optionally, the currently locked protocol configuration incremental instruction stream is compiled and generated according to the partial reconfigurable region division of the target FPGA, and the corresponding independently loadable bitstream slice file is generated; the bitstream slice file is sent to the heterogeneous computing protocol fusion unit through a secure channel to trigger the FPGA to perform dynamic partial reconfiguration, thereby completing the online fusion and switching of the charging protocol.

[0016] A second aspect of this application provides an adaptive conversion system for electric vehicle charging protocols. This system includes: a computing architecture construction module, used to deploy a high-frequency micro-current sensor array and a heterogeneous computing protocol fusion unit in a V2G smart charging pile with protocol reconfigurability, constructing a two-level heterogeneous computing architecture of protocol perception and protocol fusion; a data acquisition module, used to, based on the two-level heterogeneous computing architecture, capture in real time the micro-current ripple characteristic waveform of the communication carrier during the coupling and handshake phases of the charging connector using the high-frequency micro-current sensor array, and simultaneously obtain the multi-protocol parallel conflict risk sequence within future scheduling cycles through a V2G protocol conflict prediction digital twin deployed in the cloud, and synchronously collect the full-dimensional operating state vector of the charging pile through an embedded system within the pile; and a driving module. The vector generation module is used to perform super-resolution feature extraction based on the micro-current ripple characteristic waveform of the communication carrier, generate an initial protocol fingerprint probability tensor, and perform spatiotemporal alignment and correlation analysis in conjunction with the multi-protocol parallel conflict risk sequence to dynamically identify key reconfigurable protocol modules and their expected evolution paths, and fuse them to generate a protocol fusion driving vector. The data prediction module is used to analyze the nonlinear coupling relationship between the charging pile's full-dimensional operating state vector and the protocol fusion driving vector, and predict the optimal protocol parameter evolution trajectory under the FPGA partial reconfigurability constraint. The comparison simulation module is used to perform real-time in-loop compatibility comparison and conflict resolution simulation with the target V2G service protocol in a high-fidelity virtual interactive environment in the cloud based on the optimal protocol parameter evolution trajectory, and dynamically generate a FPGA partial reconfigurable bitstream slice sequence.

[0017] One or more technical solutions provided in this application have at least the following technical effects or advantages: The method provided in this application embodiment deploys a high-frequency micro-current sensor array and a heterogeneous computing protocol fusion unit in a V2G smart charging pile with protocol reconfigurability to construct a two-level heterogeneous computing architecture of protocol perception and protocol fusion. Based on the two-level heterogeneous computing architecture, the high-frequency micro-current sensor array captures the micro-current ripple characteristic waveform of the communication carrier during the charging connector coupling and communication handshake stages in real time. Simultaneously, a V2G protocol conflict prediction digital twin deployed in the cloud is used to obtain the multi-protocol parallel conflict risk sequence within the future scheduling cycle, and the full-dimensional operating state vector of the charging pile is synchronously collected through the embedded system in the pile. Based on the micro-current ripple characteristic waveform of the communication carrier... Super-resolution feature extraction is performed to generate an initial protocol fingerprint probability tensor. This tensor is then combined with the multi-protocol parallel conflict risk sequence for spatiotemporal alignment and correlation analysis to dynamically identify key reconfigurable protocol modules and their expected evolution paths, fusing them to generate a protocol fusion driving vector. Based on the charging pile's full-dimensional operating state vector and the protocol fusion driving vector, their nonlinear coupling relationship is analyzed to predict the optimal protocol parameter evolution trajectory under FPGA partial reconfigurability constraints. Based on this optimal protocol parameter evolution trajectory, real-time in-loop compatibility comparison and conflict resolution simulation with the target V2G service protocol are performed in a high-fidelity virtual interactive environment in the cloud, dynamically generating a FPGA partial reconfigurable bitstream slice sequence. This achieves the technical effects of real-time and accurate identification of protocol conflicts, dynamic generation of optimal protocol parameters, and improved protocol adaptability and compatibility.

[0018] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating an adaptive conversion method for electric vehicle charging protocols provided in this application.

[0021] Figure 2 This is a schematic diagram of an adaptive conversion system for electric vehicle charging protocols provided in this application.

[0022] Figure labeling: Module 11 for computing architecture construction, Module 12 for data acquisition, Module 13 for driving vector generation, Module 14 for data prediction, and Module 15 for comparison and simulation. Detailed Implementation

[0023] This application provides an adaptive conversion method and system for electric vehicle charging protocols, addressing the technical problems of existing technologies such as reliance on manual configuration for protocol adaptation, delayed conflict prediction, and the inability to achieve hardware-level dynamic reconfiguration, resulting in low compatibility and flexibility. It achieves the technical effects of real-time and accurate identification of protocol conflicts, dynamic generation of optimal protocol parameters, and improved protocol adaptability and compatibility.

[0024] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all of them.

[0025] Example 1, as Figure 1 As shown, this application provides an adaptive conversion method for electric vehicle charging protocols, which includes: In V2G smart charging piles with reconfigurable protocols, a high-frequency micro-current sensor array and a heterogeneous computing protocol fusion unit are deployed to construct a two-level heterogeneous computing architecture of protocol perception and protocol fusion.

[0026] Specifically, V2G smart charging piles with reconfigurable protocols refer to charging piles that can dynamically adjust their charging protocols and operating modes according to the different needs of electric vehicles (EVs) and power grid dispatching. The hardware and software of such charging piles are reconfigurable, allowing them to flexibly adapt to various charging tasks based on different charging standards, communication protocols, or operating modes, thereby improving charging efficiency and ensuring compatibility with different vehicle models, power grids, and other smart devices.

[0027] In V2G smart charging piles with reconfigurable protocols, a high-frequency micro-current sensor array and a heterogeneous computing protocol fusion unit are deployed. The high-frequency micro-current sensor array consists of multiple high-precision, high-sensitivity micro-current sensors, enabling data acquisition devices to detect minute current changes at extremely high frequencies, such as thousands or even more times per second. This array is deployed in the communication interface of the charging pile to accurately capture minute changes in current fluctuations during charging, especially the high-frequency ripple characteristics in the current signal. Furthermore, the high-frequency micro-current sensor array employs a differential arrangement, meaning the sensors measure the difference between two current signals to eliminate the influence of external noise, improve measurement accuracy and anti-interference capabilities, and thus obtain higher-precision current waveform data.

[0028] The heterogeneous computing protocol fusion unit is mainly composed of FPGAs (Field-Programmable Gate Arrays), which have reconfigurable hardware characteristics and can dynamically adjust the hardware configuration according to the requirements of different protocols. In the heterogeneous computing protocol fusion unit, the FPGA real-time signal processing channel is responsible for processing the current waveform data transmitted by the high-frequency micro-current sensor array. Through parallel computing and real-time data processing capabilities, the FPGA can quickly extract the micro-current ripple characteristics of the communication carrier during the charging process.

[0029] A high-frequency micro-current sensor array and a heterogeneous computing protocol fusion unit are rigidly connected using equal-length backplane wiring, forming a two-stage heterogeneous computing architecture of protocol sensing and protocol fusion. The equal-length backplane wiring ensures the balance and synchronization of signal transmission, avoiding signal delay issues caused by inconsistent wiring lengths. The rigid connection further enhances the stability and reliability of the connection, avoiding errors and interference in signal transmission and improving the accuracy and real-time performance of the protocol fusion process. Specifically, in the protocol sensing stage of the two-stage architecture, the communication current waveform captured by the micro-current sensor array is used to monitor the communication process between the charging pile and the electric vehicle in real time, providing basic data for protocol identification. In the protocol fusion stage, the FPGA processing capabilities of the heterogeneous computing unit are used to process the acquired signals in real time and reliably.

[0030] A two-level heterogeneous computing architecture of protocol awareness and protocol fusion is constructed. Through high-precision current waveform acquisition and high-speed real-time calculation, the correct identification and optimization adjustment of electric vehicle charging protocols are ensured, thereby improving the compatibility of charging piles with electric vehicles and the power grid.

[0031] Based on the aforementioned two-level heterogeneous computing architecture, the high-frequency micro-current sensor array captures the micro-current ripple characteristic waveform of the communication carrier during the coupling and handshake phases of the charging connector in real time. Simultaneously, the V2G protocol conflict prediction digital twin deployed in the cloud obtains the multi-protocol parallel conflict risk sequence within the future scheduling cycle, and the charging pile's full-dimensional operating status vector is synchronously collected through the embedded system within the charging pile.

[0032] Specifically, multi-dimensional data acquisition is achieved through a two-level heterogeneous computing architecture of protocol awareness and protocol fusion. First, during the coupling and communication handshake phase of the charging connector, a high-frequency micro-current sensor array collects the micro-current ripple characteristic waveform of the communication between the charging pile and the electric vehicle in real time. The coupling and communication handshake phase of the charging connector refers to the initial communication exchange process between the electric vehicle and the charging pile after they are connected, through protocols such as power line carrier communication (PLC). This includes the coupling phase and the communication handshake phase. The coupling phase refers to the signal coupling between the charging pile and the electric vehicle's battery management system (BMS) through power line carrier communication (PLC) when the electric vehicle is physically connected to the charging pile. At this time, a signal transmission channel is established between the charging pile and the electric vehicle, preparing for the subsequent communication handshake.

[0033] After the charging pile and the electric vehicle are successfully physically connected, a communication handshake phase begins. This phase includes authentication, protocol negotiation, and confirmation of charging parameters between the charging pile and the electric vehicle via the PLC protocol. In other words, the charging pile sends signals to the electric vehicle, and after the electric vehicle responds, both parties exchange information via a specific carrier frequency to confirm protocol parameters such as charging mode, charging power, and time. At this time, a high-frequency micro-current sensor array acquires the micro-current ripple characteristic waveform of the communication carrier. This waveform reflects the communication carrier used during the charging process, such as the PLC protocol, power line carrier communication, and the signal exchange during the protocol handshake.

[0034] Key data is collected from multiple sources, including charging piles, battery management systems (BMS), and the power grid. This key data includes, but is not limited to, charging pile data, electric vehicle data, and power grid dispatch data. Charging pile data includes information such as the charging pile's hardware status, communication protocol type, operating mode, current and voltage waveforms, and power output. Electric vehicle data includes vehicle battery status, such as SOC (State of Charge), battery capacity, vehicle charging requirements, and communication protocols supported by the BMS. Power grid dispatch data includes grid load conditions, power grid dispatch tasks, dispatch protocols, and response time requirements.

[0035] Modeling is required for various charging protocols, such as CHAdeMO, CCS, GB / T, power line carrier communication (PLC) protocols, and grid dispatching protocols. Each protocol has different communication characteristics, power control logic, and dispatching requirements, thus requiring the definition of standardized data structures and processing logic for each. The protocol model needs to include the basic protocol flow, protocol characteristics and constraints, and protocol adaptation conditions. The basic protocol flow includes communication and dispatching phases. Communication phases include handshakes, acknowledgments, and data transmission; dispatching phases include power demand and grid response. Protocol characteristics and constraints include communication frequency, data transmission rate, power limit, and charging time. Protocol adaptation conditions include how to interoperate with other protocols, which protocols can be executed in parallel, and which protocols may conflict.

[0036] Based on key data and protocol models, a virtual simulation environment is constructed between charging piles, vehicles, and the power grid. This virtual simulation environment reflects the behavior of real charging piles through digital twin technology, including the interaction between charging piles and electric vehicles, power grid scheduling and charging scheduling, and the parallel execution of multiple protocols. The virtual simulation environment is used as a digital twin for V2G protocol conflict prediction and deployed in the cloud for multi-protocol parallel simulation. This predicts conflicts between different protocols and obtains the risk sequence of multi-protocol parallel conflicts within future scheduling cycles. Specifically, multiple charging piles, different electric vehicles, and power grid scheduling tasks are simultaneously introduced into the V2G protocol conflict prediction digital twin. Through parallel execution, possible scenarios of protocol conflicts are simulated. For example, multiple charging protocols may compete for power grid resources at the same time, leading to excessive power grid load, or data transmission conflicts between different protocols may cause charging piles to be unable to correctly parse instructions. The V2G protocol conflict prediction digital twin identifies conflict points between different protocols by analyzing the execution status, scheduling requests, and power grid resource allocation of each protocol. Examples of conflicts include protocol conflicts, resource conflicts, and scheduling conflicts. Protocol conflicts refer to the inability of a single charging pile to support multiple communication protocols simultaneously. Resource conflicts refer to the excessive power demand from multiple charging piles on the power grid, leading to an overload on the grid. Scheduling conflicts refer to the overlap between the time windows of grid scheduling tasks and electric vehicle charging tasks, making it impossible to satisfy them simultaneously. Based on simulation results, a multi-protocol parallel conflict risk sequence is generated for future scheduling cycles. This sequence represents the spatiotemporal sequence of potential protocol conflicts that may occur when multiple protocols are executed between charging piles, multiple electric vehicles, and the grid scheduling system. It indicates the potential conflicts between different charging protocols, grid scheduling protocols, and the charging demands of electric vehicles within future scheduling cycles.

[0037] Meanwhile, the embedded system inside the charging pile synchronously collects the full-dimensional operating status vector of the charging pile. The embedded system is an intelligent control system integrated inside the charging pile, such as an embedded processor or FPGA. The full-dimensional operating status vector includes the operating parameters of the charging pile in various aspects of its current working state, including but not limited to input voltage, output current, charging power, battery power, temperature, communication quality, etc.

[0038] The microcurrent ripple characteristic waveform of the communication carrier collected by the high-frequency microcurrent sensor array provides a direct data source for protocol perception, enabling the charging pile to accurately identify the charging protocol currently in use. The multi-protocol parallel conflict risk sequence obtained by the cloud-based V2G protocol conflict prediction digital twin provides forward-looking information for the charging pile to plan protocol adjustment strategies in advance, which helps to avoid problems such as charging interruption caused by protocol conflicts. The full-dimensional operating status vector of the charging pile collected by the embedded system in the pile provides the equipment status basis for protocol adjustment, ensuring that the protocol adjustment will not affect the normal operation of the charging pile. In turn, it provides comprehensive and accurate data support for subsequent protocol adaptive conversion and optimization, improving the stability and intelligence of the charging process.

[0039] Based on the microcurrent ripple characteristic waveform of the communication carrier, super-resolution feature extraction is performed to generate an initial protocol fingerprint probability tensor. Then, combined with the multi-protocol parallel conflict risk sequence, spatiotemporal alignment and correlation analysis are performed to dynamically identify key reconfigurable protocol modules and their expected evolution paths, and a protocol fusion driving vector is generated.

[0040] Furthermore, based on the microcurrent ripple characteristic waveform of the communication carrier, super-resolution feature extraction is performed to generate an initial protocol fingerprint probability tensor. This is then combined with the multi-protocol parallel conflict risk sequence for spatiotemporal alignment and correlation analysis to dynamically identify key reconfigurable protocol modules and their expected evolution paths. A protocol fusion driving vector is generated, including: extracting multi-dimensional protocol fingerprint features within each time window using a sliding time window complex wavelet transform on the microcurrent ripple characteristic waveform of the communication carrier; calculating the similarity between the multi-dimensional protocol fingerprint features and the protocol feature knowledge graph updated by federated learning using a lightweight protocol fingerprint matching engine, outputting a multi-protocol confidence distribution matrix indexed by time as the initial protocol fingerprint probability tensor; decoding the multi-protocol parallel conflict risk sequence into protocol state transition constraints, and solving for these constraints with the multi-protocol confidence distribution matrix to identify key reconfigurable protocol modules and predict their adjustment direction and magnitude, thus forming a protocol fusion driving vector.

[0041] Specifically, for the continuously acquired micro-current ripple characteristic waveform of the communication carrier, according to the time resolution Δt and sliding step size Δs set by the system, the micro-current ripple characteristic waveform of the communication carrier is slidably divided on the time axis according to the fixed window length L=Δt×sampling rate and the window translation step size Δs. In this way, the continuous micro-current ripple characteristic waveform of the communication carrier is divided into multiple partially overlapping or equally spaced sliding time windows, so as to ensure the time continuity of signal analysis and the ability to capture dynamic changes.

[0042] Within each sliding time window, the local communication carrier micro-current ripple signal within that window is extracted as an independent analysis sample, and a complex wavelet transform is performed on it. A suitable complex wavelet basis function, such as the Morlet complex wavelet, is selected based on the characteristics of the communication carrier. Multi-scale convolution operations are then performed on the local communication carrier micro-current ripple signal to obtain complex wavelet coefficients at different scales. The amplitude spectrum, phase spectrum, and energy distribution are then calculated for the complex wavelet coefficients at different scales. The magnitude of the amplitude spectrum coefficients reflects the signal intensity distribution, the phase of the phase spectrum coefficients reflects modulation and symbol transition information, and the energy distribution reflects the sum of squares of the coefficients at each scale, indicating the degree of signal energy concentration.

[0043] Further, the dominant frequency component, local energy peak position, transient change point, and scale correlation features are extracted from the time-frequency representations such as amplitude spectrum, phase spectrum, and energy distribution. The dominant frequency component refers to the frequency corresponding to the maximum amplitude. The local energy peak position is determined by finding local maxima on the time series of the amplitude spectrum. The peak time is located by using neighborhood comparison or second derivative zero-point detection methods. The transient change point refers to the position where the adjacent coefficients change abruptly beyond a threshold. The threshold can be determined by statistical methods, such as using the mean of the difference sequence within the current time window plus twice the standard deviation as the threshold. The scale correlation feature refers to the energy ratio between different scales. The extracted features are combined to form the multidimensional protocol fingerprint feature corresponding to the time window.

[0044] After obtaining the multidimensional protocol fingerprint features within each time window, a lightweight protocol fingerprint matching engine is used. This engine calculates similarity with the protocol feature knowledge graph updated by federated learning to determine whether the currently extracted multidimensional protocol fingerprint features match existing protocol fingerprint models. Federated learning is a distributed learning method that can jointly train models on multiple edge devices, such as different charging piles and vehicles, without centrally storing the dataset. The protocol feature knowledge graph updated by federated learning is a dynamically updated knowledge base. Through federated learning technology, data is collected from multiple charging piles or related devices, and protocol feature information is continuously optimized and expanded, enabling the protocol feature knowledge graph to more comprehensively and accurately reflect the characteristics of various charging protocols, ensuring that new or unseen protocols can also be correctly identified.

[0045] The protocol feature knowledge graph updated through federated learning is obtained by locally training and updating the multidimensional protocol fingerprint features collected on multiple local models deployed on different charging piles, vehicles, or edge nodes. The local model is a protocol feature embedding model based on a lightweight neural network, specifically a one-dimensional convolutional neural network (1D-CNN) structure. This structure maps the multidimensional protocol fingerprint feature vector extracted in each time window into a low-dimensional semantic embedding representation, while outputting the local confidence of the corresponding protocol category. The overall structure of the one-dimensional convolutional neural network consists of one-dimensional convolutional layers, batch normalization layers, activation function layers, local pooling layers, and fully connected embedding layers stacked in sequence. The input is the multidimensional protocol fingerprint feature vector extracted within each time window. The 1D convolutional layer performs sliding convolution operations along the time axis on the feature sequence, extracting local temporal patterns and frequency domain structure information through multiple convolutional kernels to capture periodic changes and abrupt structures in microcurrent ripples. After the convolutional output, a batch normalization layer is applied to standardize the distribution of each batch of features, suppressing internal covariate shifts, improving the stability of the local model training, accelerating convergence, and enhancing the robustness of the local model to differences in data distribution under different devices and sampling conditions. Then, an activation function layer, such as ReLU or GELU, is applied. ReLU enhances nonlinear expressive power, while GELU improves feature smoothness while preserving gradient information, enhancing the neural network's ability to fit complex protocol feature patterns through nonlinear mapping. After feature extraction layer by layer, a local pooling layer is introduced to compress and reduce the dimensionality of convolutional features in local time regions through max pooling, extracting key salient features and reducing redundant information, while also enhancing the local model's tolerance to slight temporal shifts. The obtained high-level features are flattened and input into a fully connected embedding layer, which maps the high-dimensional features into fixed-dimensional protocol embedding vectors. The confidence probability of each protocol category is output through a Softmax layer, which is then used for supervised training and protocol matching and recognition, realizing the structured representation of protocol features and lightweight edge deployment.

[0046] The local model is trained using supervised learning on various charging piles or edge nodes, utilizing locally collected historical protocol fingerprint samples and real protocol labels. Model parameters are optimized using cross-entropy loss or contrastive learning loss functions, bringing samples of the same protocol closer together and distancing samples of different protocols further apart. Self-supervised pre-training can be combined to improve generalization ability for unknown protocol samples. Mini-batch gradient descent, such as the Adam optimizer, is used during training to complete several rounds of local iterations while ensuring local data remains within its domain. The trained local model parameters, such as feature embedding vectors, classification weights, or gradient information, are then uploaded to the cloud or a central aggregation server. The central server uses a weighted aggregation algorithm, such as FedAvg, to fuse updates from local models from different nodes and automatically updates the structural information of protocol nodes, protocol attributes, and protocol relationship edges based on newly added protocol samples. This dynamically expands and optimizes the protocol feature knowledge graph, ensuring that each protocol node in the graph is associated with a corresponding feature embedding representation and behavioral pattern label, achieving continuous evolution and adaptive updates.

[0047] The lightweight protocol fingerprint matching engine calculates the similarity between the multi-dimensional protocol fingerprint features extracted within the current time window and the feature embedding vectors corresponding to each protocol node in the protocol feature knowledge graph updated by federated learning. It employs lightweight calculation methods such as cosine similarity, Euclidean distance, inverse distance, or dot product normalization to calculate the matching degree between the current multi-dimensional protocol fingerprint features and all protocol category features in the protocol feature knowledge graph, obtaining the matching confidence score for each protocol category within the current time window. The confidence scores of all protocol categories are arranged chronologically to form a multi-protocol confidence distribution matrix indexed by time and categorized by protocol category. Each row of this matrix corresponds to a time window, and each column corresponds to the confidence score of a protocol. This multi-protocol confidence distribution matrix is ​​used as the initial protocol fingerprint probability tensor, which carries a high-precision time label.

[0048] The multi-protocol parallel conflict risk sequence is expanded along the time axis. Based on the information contained in the multi-protocol parallel conflict risk sequence, such as timestamps, conflicting protocol pairs, conflict types, and durations, core information such as conflict types, conflict occurrence time intervals, and involved protocol combinations is extracted and formalized into constraint expressions with time attributes. For example, within the time interval T, the state of protocol A cannot be in high-power mode and communication mode of protocol B at the same time. Thus, prohibited state transition edges or restricted transition edges in the protocol state diagram are constructed to form computable protocol state transition constraints.

[0049] The protocol state transition constraints are coupled with a multi-protocol confidence distribution matrix. The confidence matrix is ​​treated as the probability distribution of each protocol state at the current time, and this matrix is ​​input as an optimization variable into the constraint satisfaction solution model. This model is a constrained optimization model derived from the state transition constraints, such as probability sum constraints, state mutual exclusion constraints, and time continuity constraints. Its objective function is typically set to maximize the overall confidence or minimize the conflict cost function while satisfying all conflict constraints. By introducing the constraint satisfaction problem (CSP) or constrained optimization algorithms, such as the Lagrange multiplier method, to solve for the protocol state combination that maximizes confidence while satisfying conflict constraints, the protocol state change nodes with the greatest impact on the overall system are selected.

[0050] After obtaining the optimal protocol state combination, the gradient of the confidence distribution before and after optimization is calculated. This involves differential analysis of the probability changes of each protocol state, combined with Lagrange multipliers or constraint sensitivity coefficients, to evaluate the marginal contribution of each protocol state variable to the objective function. If a protocol state shows a significant probability adjustment under constraints and contributes the most to the reduction of the conflict function, the protocol module to which that state belongs is identified as a key reconfigurable protocol module. A key reconfigurable protocol module refers to a protocol functional unit within the protocol architecture that supports dynamic parameter configuration or functional customization. Its parameters can be dynamically adjusted through software configuration or FPGA reconfiguration to achieve real-time optimization of protocol behavior. Based on the direction of state change in the optimization results (i.e., probability increase or decrease) and the magnitude of the gradient of the corresponding variable, the parameter adjustment direction and magnitude of the key reconfigurable protocol module are obtained, forming a protocol fusion driving vector. This vector represents the specific adjustments to the protocol parameters that need to be made in the future, guiding how to adjust protocol parameters according to current charging demand, grid load, and vehicle status to ensure protocol compatibility, avoid conflicts, and improve overall charging efficiency.

[0051] By accurately extracting and analyzing the micro-current ripple characteristics of communication carriers, key features, potential conflicts, and optimized paths in charging protocols can be dynamically identified, thereby adjusting protocol parameters in real time. Combined with a multi-protocol parallel conflict risk sequence, protocol conflicts can be accurately predicted and prevented within future scheduling cycles, ensuring efficient collaboration between charging piles, electric vehicles, and the power grid. This, in turn, ensures the flexibility, intelligence, and adaptability of electric vehicle charging protocol conversion, and enables intelligent optimization of charging pile and power grid scheduling.

[0052] Based on the full-dimensional operating state vector and protocol fusion driving vector of the charging pile, the nonlinear coupling relationship between the two is analyzed, and the evolution trajectory of the optimal protocol parameters under the reconfigurable constraints of the FPGA part is predicted.

[0053] Furthermore, based on the full-dimensional operating state vector and protocol fusion driving vector of the charging pile, the nonlinear coupling relationship between the two is analyzed, and the optimal protocol parameter evolution trajectory under the reconfigurable constraints of the FPGA part is predicted. This includes: establishing a composite prediction model containing the full-dimensional operating state vector and protocol fusion driving vector of the charging pile, analyzing the coupling relationship between state disturbances and protocol requirements, and predicting the adjustment trend of protocol parameters; based on the adjustment trend, combined with the hardware constraint model of the reconfigurable region of the FPGA part, a discrete protocol state transition sequence is mapped and generated as the optimal protocol parameter evolution trajectory.

[0054] Furthermore, a composite prediction model is established, comprising a full-dimensional operating state vector of the charging pile and a protocol fusion driving vector. This model analyzes the coupling relationship between state disturbances and protocol requirements, and predicts the adjustment trend of protocol parameters. The process includes: decoupling the full-dimensional operating state vector of the charging pile in the time-frequency domain to separate the steady-state baseline component and the transient disturbance component; decomposing the protocol fusion driving vector into protocol semantic layer adjustment requirements and communication physical layer adjustment requirements; establishing a disturbance-requirement coupled state-space equation, using the transient disturbance component as input and the rate of change of the protocol fusion driving vector as the state variable, and performing state estimation through a Kalman filter to solve for the optimal adjustment acceleration of the protocol parameters.

[0055] Furthermore, based on the aforementioned adjustment trend and combined with the hardware constraint model of the partially reconfigurable region of the FPGA, a discrete protocol state transition sequence is generated as the optimal protocol parameter evolution trajectory. This includes: establishing a hardware constraint model based on the size of the partially reconfigurable region of the FPGA and the bitstream reconfiguration delay parameters; inputting the optimal adjustment acceleration of the protocol parameters into the hardware constraint model to calculate the feasible protocol parameter adjustment increment within the hardware resource and delay boundaries; iteratively extrapolating the protocol state coordinates of future time series based on the current protocol state and the protocol parameter adjustment increment; and performing sequence optimization on the protocol state coordinates to generate a continuous and hardware-realizable optimal protocol parameter evolution trajectory.

[0056] Specifically, the charging pile's full-dimensional operating state vector and protocol fusion driving vector are uniformly aligned and normalized on a time axis. Normalization can be achieved using z-score or minimum-maximum normalization methods. A composite prediction model is constructed, which is a hierarchical structure model including a state decoupling layer, a vector decomposition layer, and a coupling equation layer. The state decoupling layer performs time-frequency domain decoupling on the charging pile's full-dimensional operating state vector. Through Fourier transform, the full-dimensional operating state vector is mapped to a time-frequency distribution matrix. A low-frequency cutoff threshold is set in the frequency domain, and low-frequency components, such as the fundamental frequency and its slowly changing parts, are extracted as steady-state baseline components to characterize long-term operating trends. High-frequency components or frequency abrupt change segments are extracted as transient disturbance components to characterize short-term impacts such as grid harmonics, electromagnetic interference, or load abrupt changes. Then, through inverse transform, two independent time series are reconstructed to achieve the separation of steady-state baseline components and transient disturbance components.

[0057] The vector decomposition layer decomposes the protocol fusion driving vector into protocol semantic layer adjustment requirements and communication physical layer adjustment requirements. First, a parameter classification mapping table is established based on the functional attributes of the protocol parameters. Upper-layer control parameters, such as authentication mechanisms, handshake timeouts, retransmission strategies, and load balancing logic, are classified as protocol semantic layer requirements, while lower-layer communication parameters, such as carrier frequency, modulation rate, time slot length, and signal threshold, are classified as communication physical layer requirements. Each component in the protocol fusion driving vector is then projectively decomposed according to the classification rules, forming two sub-vectors. Orthogonal decomposition methods are then used to reduce dimensionality and remove redundant correlations, ensuring that semantic layer and physical layer requirements remain relatively independent in mathematical space.

[0058] The coupled equation layer establishes a disturbance-demand coupled state-space equation to analyze the coupling relationship between state disturbances and protocol requirements, and to predict the adjustment trend of protocol parameters. When establishing the disturbance-demand coupled state-space equation, the transient disturbance component is defined as the input variable, and the rate of change of the protocol fusion driving vector, i.e., the first-order time derivative, is defined as the state variable. The discrete-time state equation is constructed as: x(k+1)=Ax(k)+Bu(k)+w(k), where x(k) represents the rate of change of protocol parameters, u(k) represents the transient disturbance input, A is the state transition matrix, B is the disturbance coupling matrix, and w(k) is the process noise. The observation equation is set as y(k)=Cx(k)+v(k), where y(k) is the actual measured adjustment of protocol parameters, and v(k) is the observation noise. The Kalman filter recursive algorithm is adopted to calculate the optimal state estimate by predicting and updating iteratively. At each time step, the minimum mean square error estimate of the protocol parameter change rate is obtained. Then, the change rate is integrated to obtain the optimal adjustment acceleration of the protocol parameter, which is used to predict the adjustment trend of the protocol parameter, thereby realizing dynamic prediction and smooth optimization of the protocol parameter evolution path.

[0059] Based on the size of the reconfigurable region of the FPGA, such as area size, number of logic units, on-chip storage capacity, and bitstream reconfiguration delay parameters, a hardware constraint model is established. The number of logic units refers to the upper limit of LUT / FF / DSP occupancy. Bitstream reconfiguration delay parameters include single reconfiguration time and minimum reconfiguration time interval. The hardware constraint model is a set of inequalities that include resource capacity constraints and time constraints. For example, the change in logic resources corresponding to a single parameter adjustment must not exceed the remaining capacity of the PRR, and the time interval between two adjacent state transitions must not be less than the reconfiguration delay.

[0060] The optimal adjustment acceleration of the protocol parameters obtained through state-space equations and Kalman filtering is input into the hardware constraint model. This optimal adjustment acceleration a(k) is used as a continuous dynamic input, and numerical integration is performed over a discrete time step Δt to calculate the theoretical continuous adjustment amount. For example, the theoretical parameter change at the current time step is obtained by first recursively calculating v(k) = v(k−1) + a(k)Δt, and then recursively calculating Δptheory(k) = v(k)Δt using displacement. This theoretical parameter change is then input into a pre-established FPGA hardware constraint model. The FPGA hardware constraint model includes resource capacity constraints, adjustment granularity constraints, and reconfiguration delay constraints. Resource capacity constraints include the upper limit of the number of LUTs available for a single partial reconfiguration; adjustment granularity constraints include the minimum parameter change unit; and reconfiguration delay constraints include the time interval between two adjacent state transitions not being less than the reconfiguration delay.

[0061] Within each discrete time step, it is determined whether the theoretical adjustment amount exceeds the maximum single resource change amount or whether a reconfiguration time conflict is triggered. If it exceeds the resource limit, the parameter change amount is truncated and limited to the maximum supportable range. If it is lower than the minimum adjustment granularity, quantization alignment is performed and rounded to the nearest executable granularity. If the reconfiguration interval constraint is violated, delayed scheduling is performed, postponing the adjustment instruction to the next executable moment that meets the time window. In this way, the continuous theoretical adjustment trend is transformed into discrete protocol parameter adjustment increments that can be executed within the boundaries of resources, granularity, and latency.

[0062] Based on this, using the current protocol state as the initial coordinates, feasible protocol parameter adjustments are incrementally added, and a forward recursive approach is used to extrapolate the protocol state coordinates within multiple future time windows. The protocol state coordinates refer to the state positions in a multi-dimensional parameter space determined by parameters such as communication rate, power limit, time slot length, and timeout threshold. The protocol state sequence undergoes sequence optimization processing, employing a smoothing filter method to calculate the state change amplitude between adjacent time steps. If the direction of change at a transition point reverses within a short period (e.g., +0.5ms followed immediately by -0.5ms), it is identified as an oscillating transition, and frequent oscillating transition points are eliminated. This ensures that the state transition is continuous on the time axis and meets the reconfigurability constraints, ultimately generating a continuous, monotonous, or piecewise smooth optimal protocol parameter evolution trajectory. This trajectory is not only continuous in time without sudden jumps or interruptions but also fully utilizes the hardware resources of the reconfigurable region of the FPGA, achieving optimal adjustment of protocol parameters while meeting bitstream reconfiguration latency requirements.

[0063] By establishing a hardware constraint model, the hardware characteristics of the reconfigurable region of the FPGA are fully considered, ensuring that the generated protocol parameter adjustment scheme can be smoothly implemented on actual hardware. Based on the calculation of feasible parameter adjustment increments using the optimal adjustment acceleration, the adjustment of protocol parameters can be reasonably controlled according to actual needs. This ensures both the adaptability and flexibility of the protocol, while avoiding problems such as insufficient hardware resources or excessive reconfiguration delays caused by excessively rapid or large parameter adjustments. Iterative deduction of protocol state coordinates can clearly and accurately reflect the changing trend of the protocol state over time. By performing sequence optimization on the protocol state coordinates, the final generated optimal protocol parameter evolution trajectory can guide the reconfigurable region of the FPGA to achieve efficient and stable protocol parameter adjustment and adaptive conversion, improving the performance and reliability of the entire V2G electrical system.

[0064] Based on the evolution trajectory of the optimal protocol parameters, real-time in-loop compatibility comparison and conflict resolution simulation with the target V2G service protocol are performed in a high-fidelity virtual interactive environment in the cloud, and a partially reconfigurable bitstream slice sequence of the FPGA is dynamically generated.

[0065] Furthermore, based on the optimal protocol parameter evolution trajectory, real-time in-loop compatibility comparison and conflict resolution simulation with the target V2G service protocol are performed in a high-fidelity virtual interactive environment in the cloud, dynamically generating a partially reconfigurable bitstream slice sequence for the FPGA. This includes: constructing a high-fidelity virtual interactive environment synchronized with physical charging piles, vehicle BMS, and power grid dispatch instructions; injecting the optimal protocol parameter evolution trajectory into the virtual interactive environment in the form of incremental protocol configuration instruction streams to drive dynamic reconstruction of the virtual protocol stack; simulating the execution of the target V2G service protocol in the virtual interactive environment and comparing the differences between the output behavior of the virtual protocol stack and the expected behavior of the target protocol in real time to generate a multi-dimensional difference vector; performing closed-loop feedback optimization based on the multi-dimensional difference vector, dynamically adjusting the incremental protocol configuration instruction streams until the multi-dimensional difference vector converges to a preset threshold range; and compiling the finally converged protocol configuration instruction stream into a partially reconfigurable bitstream slice sequence for the FPGA.

[0066] Specifically, by deploying multi-source data acquisition interfaces in physical charging piles, vehicle BMS, and the power grid, real-time operational data such as power data, voltage and current waveforms, communication messages, protocol state machine states, and power grid dispatch instructions are synchronously transmitted to a cloud-based digital twin platform via high-speed communication links. Based on this real-time data, a hierarchical digital twin model is established, including a physical layer electrical equivalent model, a communication layer protocol behavior model, a control layer state machine model, and a dispatch layer load response model. The physical layer electrical equivalent model simulates power response and energy conversion processes; the communication layer protocol behavior model simulates handshake procedures, parameter negotiation, and message interaction; and the control layer state machine model simulates protocol state transition logic. Through parameter identification algorithms and online error correction mechanisms, the parameters of multiple twin models are continuously and dynamically updated, ensuring that the power curves, timing responses, and state transition trajectories of the virtual twin model maintain a high degree of accuracy in fitting the measured data of the real system within the allowable error range. Simultaneously, a real-time feedback correction mechanism and data assimilation algorithm, such as Kalman filtering or particle filtering, are introduced to synchronously correct key variables in the virtual environment. This ensures that the virtual model can accurately reflect the dynamic changes of the physical system under different operating conditions, thereby constructing a high-fidelity virtual interactive environment with real-time synchronization capabilities and high consistency with the real operating environment. The high-fidelity virtual interactive environment uses digital twin technology to synchronize and map the physical charging pile status, vehicle BMS behavior, and power grid dispatch instructions in real time. This ensures that the virtual system is consistent with the actual system in terms of power curves, communication timing, and protocol state machine transitions. High fidelity refers to maintaining a high-precision fitting capability within the error range of parameters, dynamic response characteristics, and physical system, thereby ensuring the reliability and accuracy of the simulation results.

[0067] The optimal protocol parameter evolution trajectory is converted into a structured incremental protocol configuration instruction stream. First, the optimal protocol parameter evolution trajectory is discretized over time, dividing the continuous parameter change curve into several key change points according to time nodes. The parameter difference between adjacent time windows is calculated to obtain the incremental change of each protocol parameter. This incremental change is then encoded into a standardized control instruction format according to predefined protocol field mapping rules. For example, communication time slot adjustments, power limit changes, and timeout adjustments are mapped to identifiable parameter fields, with timestamps, execution priorities, and dependency identifiers added to form a structured control data packet. The predefined protocol field mapping rules refer to establishing a unified parameter-field mapping table and encoding standard for protocol parameter types, controllable hardware resources, and protocol state machine structures. This maps physical changes in protocol parameters, such as power increments, timing offsets, and window length changes, into standardized data fields and parsable control field formats, ensuring that parameter increments generated by different modules and different time trajectories can be uniformly identified and parsed by the virtual protocol stack and FPGA reconfiguration module. For example, the protocol field mapping rules stipulate that communication time slot adjustment is mapped to the control field Tag=0x01, with a data format of a 16-bit signed integer representing the time slot increment in μs; power limit change is mapped to Tag=0x02, with a data format of a 32-bit floating-point number representing the power increase or decrease in kW, and an attached power change direction flag; handshake timeout adjustment is mapped to Tag=0x03, with a data format of a 16-bit unsigned integer representing the time extension or reduction in milliseconds, and an additional priority field and dependent module ID; protocol function module start / stop is mapped to Tag=0x04, using a 1-bit Boolean value to represent module state switching. Through serialization and version identification mechanisms, the parameter increments corresponding to all time nodes are arranged in chronological order and encapsulated into protocol configuration increment instructions that can be parsed by the virtual protocol stack. The protocol configuration increment instruction stream refers to the set of control instructions after time-series encoding of protocol parameter changes, used to drive the virtual protocol stack to update and switch parameters according to a predetermined trajectory, and trigger the dynamic reconstruction of the protocol state machine in the virtual environment.

[0068] In a virtual interactive environment, the target V2G service protocol is executed synchronously with the virtual protocol stack reconstructed based on incremental instruction streams. Real-time simulation compares the output behavior of the virtual protocol stack with the expected behavior of the target protocol in dimensions such as protocol parameters, communication timing, and power response. A multi-dimensional difference vector is calculated, containing multiple quantitative indicators such as protocol parameter differences, communication timing differences, and power response differences, used to measure the degree of matching between the current protocol configuration and the target protocol standard. Closed-loop feedback optimization is performed based on the multi-dimensional difference vector, dynamically adjusting the incremental instruction stream of the protocol configuration to continuously approach a preset threshold range, achieving gradual convergence and compatibility optimization of protocol behavior until the difference meets the stable convergence condition. Once the stable convergence condition is met, the final stable protocol configuration instruction stream is compiled and segmented according to the FPGA resource partitioning structure, generating bitstream slice sequences suitable for reconfigurable regions of the FPGA. These bitstream slice sequences refer to binary configuration file fragments generated for independently reconfigurable logical regions in the FPGA, used to achieve online replacement and dynamic loading of local hardware functions, thereby completing the mapping and deployment of protocol optimization results to the hardware layer.

[0069] By constructing a high-fidelity virtual interactive environment, protocol compatibility and conflict resolution can be simulated and tested without interfering with the actual physical system, reducing the risks and costs of actual deployment. Real-time in-loop compatibility comparison and conflict resolution simulation based on the optimal protocol parameter evolution trajectory can promptly identify potential problems in actual protocol operation. Closed-loop feedback optimization dynamically adjusts protocol parameters to ensure protocol compatibility and stability. Finally, a partially reconfigurable bitstream slice sequence is generated for the FPGA, enabling the FPGA to dynamically adjust the protocol processing logic according to actual needs, improving the stable operation and efficient management of electric vehicle charging.

[0070] Furthermore, the target V2G service protocol is simulated and executed in the virtual interactive environment, and the differences between the output behavior of the virtual protocol stack and the expected behavior of the target protocol are compared in real time to generate a multi-dimensional difference vector. This includes: constructing a V2G service response simulation model in the high-fidelity virtual interactive environment; using power grid dispatch instructions as input stimuli and dynamically adjusting power parameters based on the simulated vehicle SOC state, running the V2G service response simulation model to obtain the standard protocol interaction process; performing a spatiotemporal alignment comparison between the dynamic protocol stack behavior driven by the incremental instruction stream of the protocol configuration and the standard protocol interaction process; calculating the difference values ​​of the comparison results in multiple dimensions such as protocol parameters, communication timing, and power response curves to generate the multi-dimensional difference vector.

[0071] Specifically, a V2G service response simulation model is constructed within a high-fidelity virtual interactive environment. First, the standard flow of the V2G service protocol is formally modeled, converting the handshake phase, parameter negotiation phase, power scheduling phase, and state feedback phase into a computable state machine structure. A unified behavioral model is then established by combining grid dispatch rules and vehicle BMS energy management logic. Simultaneously, a vehicle SOC dynamic evolution model and a battery equivalent circuit model are constructed. The SOC change trend is calculated in real time using energy conservation equations and charging / discharging power constraints. Grid-side dispatch commands are injected into the system as external input excitation. Based on this, a power response model is established, mapping protocol parameters to physical power output. This ensures that protocol state changes trigger synchronous updates of the power curve and communication behavior. Furthermore, key physical constraints such as system latency, communication delay, and power ramp-up rate are introduced through parametric modeling to improve simulation accuracy. A real-time data synchronization mechanism is used to correct the operating data of physical charging piles and the parameters of the virtual model. Error feedback and online parameter update algorithms, such as Kalman filtering or parameter identification methods, are used to continuously correct the output of the V2G service response simulation model. This results in a high-fidelity V2G service response simulation model that accurately reflects the dynamic behavior of the real system and supports protocol behavior verification and conflict detection. The V2G service response simulation model simulates the behavioral logic of the standard V2G service protocol under typical scheduling scenarios. Internally, it includes a power grid scheduling command parsing module, a vehicle SOC dynamic evolution model, and a power control state machine model. The SOC dynamic model performs real-time calculations of the charging and discharging process based on the battery equivalent circuit and energy conservation equations, ensuring that power changes and remaining battery capacity remain physically consistent during the simulation.

[0072] Dispatch commands from the power grid side, such as peak shaving and valley filling commands, power feedback commands, or load limiting commands, are used as input excitations to the V2G service response simulation model. Based on the simulated vehicle State of Charge (SOC) state, the charging and discharging power parameters are dynamically adjusted under different SOC conditions. The V2G service response simulation model is run to generate a standard protocol interaction process conforming to the V2G specification. This standard protocol interaction process includes a complete sequence of behaviors such as handshake procedures, parameter negotiation results, power scheduling curves, and communication message timing. Simultaneously, a virtual dynamic protocol stack driven by incremental protocol configuration command streams is run synchronously and spatiotemporally aligned using a unified time base, i.e., aligning timestamps and state nodes to ensure that the execution trajectories of the two protocols are compared point-by-point at the same scheduling event trigger point. After alignment, the differences in protocol parameter dimensions, communication timing dimensions, and power response curve dimensions are calculated. Protocol parameter dimensions include power upper limit deviation and timeslot length difference; communication timing dimensions include message delay difference and handshake completion time difference; and the power response curve dimension represents the mean square error or dynamic deviation between the actual power curve and the standard curve. After normalizing the differences across multiple dimensions, a multidimensional difference vector is formed. This vector quantifies the degree of matching between the current dynamic protocol stack behavior and the target standard protocol. For example, when the power grid issues a peak shaving command requiring the discharge power to be stabilized at 20kW when the vehicle's SOC is 60%, the power curve generated after running the V2G service response simulation model is 20±0.5kW, and the handshake completion time is 150ms, according to the standard protocol interaction process. The dynamic protocol stack driven by the incremental protocol configuration command stream results in a power output of 22kW and a handshake completion time of 180ms. After spatiotemporal alignment, the calculated power deviation is 2kW, the handshake delay difference is 30ms, and the average communication message delay difference is 5ms. These differences are normalized to form a multidimensional difference vector, which serves as the input for subsequent closed-loop adjustments.

[0073] By constructing a V2G service response simulation model and obtaining the standard protocol interaction process, an objective and accurate reference standard is provided for evaluating protocol stack behavior. Spatiotemporally aligning and comparing the dynamic protocol stack behavior with the standard protocol interaction process allows for precise identification of the differences between the two, avoiding evaluation errors caused by time-space mismatches. A comprehensive evaluation of protocol stack behavior from multiple dimensions generates a multidimensional difference vector, enabling timely identification of protocol issues and targeted adjustments, thereby improving protocol compatibility and flexibility and ensuring the stable and efficient operation of V2G charging stations.

[0074] Furthermore, closed-loop feedback optimization is performed based on the multidimensional difference vector to dynamically adjust the incremental instruction flow of protocol configuration until the multidimensional difference vector converges to a preset threshold range. This includes: constructing a digital twin closed-loop control model containing a protocol parameter regulator and a communication timing compensator; inputting the multidimensional difference vector into the digital twin closed-loop control model to calculate the protocol stack parameter correction amount and the communication timing compensation amount; performing rolling optimization on the protocol stack parameter correction amount and the communication timing compensation amount through a model predictive control algorithm and iteratively simulating in a virtual environment; when the simulation results show that all components of the multidimensional difference vector are less than the preset threshold, the protocol configuration optimization is determined to be complete, and the current incremental instruction flow of protocol configuration is locked.

[0075] Specifically, a unified state feedback structure is established within a high-fidelity virtual interactive environment. A multi-dimensional difference vector is used as the system error signal input, constructing a dynamic control framework with protocol parameters as control variables and the operating state as the controlled object. The protocol parameter regulator, based on an error-driven mechanism, establishes fine-tunable parameter update functions for adjustable protocol parameters such as power limits, communication rates, and handshake timeout thresholds. Parameter corrections are calculated using PID control or gradient-optimized adaptive algorithms to achieve adaptive adjustment at the protocol structure level. The communication timing compensator constructs a time deviation model to address timing errors such as handshake delay, message transmission deviation, and power response lag. A time error estimator is introduced to align the actual communication timestamps with the standard protocol time series, and the time offset is dynamically corrected through feedforward prediction and lag compensation mechanisms.

[0076] Based on this, the protocol parameter regulator and communication timing compensator are embedded in a unified state-space model. By establishing state transition equations and observation equations, closed-loop feedback linkage between error, control quantity and system response is realized, thereby forming a digital twin closed-loop control model that can be updated in real time. The digital twin closed-loop control model refers to combining the simulation results in the virtual interactive environment with the real-time optimization control algorithm. Through the state feedback mechanism, the protocol behavior and the target protocol are continuously aligned, so that the virtual protocol stack can automatically adjust parameters according to the deviation.

[0077] A multidimensional difference vector is input as the input signal to the digital twin closed-loop control model. Based on the difference vector, the digital twin closed-loop control model constructs an error state equation, calculates the protocol stack parameter correction and communication timing compensation, and uses these as control variables for optimization. In the optimization process, a model predictive control (MPC) algorithm is introduced to perform rolling optimization of the protocol stack parameter correction and communication timing compensation for several future time steps. That is, at each time step, the future state trajectory is calculated according to the MPC algorithm, and the optimal control sequence is solved under constraints such as parameter upper and lower limits, hardware resource constraints, and maximum compensation delay, thereby achieving optimal adjustment of protocol parameter correction and timing compensation.

[0078] The optimized control commands are re-injected into the virtual environment for iterative simulation. New multidimensional difference vectors are continuously calculated, and the optimization process is repeated until all difference values ​​converge to a preset threshold range. At this point, the protocol behavior is deemed to have reached an acceptable level of consistency with the target V2G service protocol, and the current protocol configuration incremental command stream is locked as the final stable result. For example, the initial simulation results show that the multidimensional difference vectors are: protocol parameter difference = 0.15, communication timing difference = 0.12, and power response difference = 0.18, all exceeding the preset threshold of 0.05. After inputting the multidimensional difference vectors into the closed-loop control model, the MPC algorithm calculates that the protocol parameters need to be adjusted by lowering the power limit by 5kW, shortening the signal delay by 20ms, and increasing the message buffer time by 10ms as corrections. The difference vectors are recalculated in the next round of virtual simulation. After three iterations, the difference vectors converge to 0.03, 0.02, and 0.01, all below the preset threshold of 0.05. At this point, the optimization is deemed complete, the current protocol configuration incremental command stream is locked, and a stable control version that can be reconfigured for FPGA is output.

[0079] By constructing a digital twin closed-loop control model, the operation of the protocol can be comprehensively and accurately simulated, and the impact of changes in protocol parameters and communication timing on protocol performance can be perceived in real time. The multidimensional difference vector is input into the model to calculate correction and compensation amounts, providing specific directions and numerical basis for protocol optimization. Using model predictive control algorithms for rolling optimization and virtual iterative simulation allows for dynamic adjustment of optimization strategies, ensuring the protocol maintains good performance under different scenarios. Finally, when the multidimensional difference vector converges to a preset threshold range, the incremental command stream of the protocol configuration is locked, guaranteeing the stability and reliability of the protocol and enabling the V2G charging pile to operate efficiently and safely.

[0080] Furthermore, the method also includes: configuring the currently locked protocol configuration incremental instruction stream, dividing the target FPGA into partially reconfigurable regions, compiling and generating corresponding, independently loadable bitstream slice files; sending the bitstream slice files to the heterogeneous computing protocol fusion unit through a secure channel, triggering the FPGA to perform dynamic partial reconfiguration, and completing the online fusion and switching of the charging protocol.

[0081] Specifically, the currently locked protocol configuration incremental instruction stream undergoes structural parsing and hardware mapping transformation. Based on the changes in protocol module parameters involved in the incremental instruction stream, the functional units that need to be updated are identified and mapped to the corresponding logic resource partitions in the pre-defined Partially Reconfigurable Regions (PRRs) of the target FPGA. According to the region constraint files, netlist definitions, and resource binding rules in the FPGA project, the updated modules are resynthesized, placed, and routed. Local configuration data is generated only for the logic structure within the specified PRR range, while keeping other static region logic unaffected. During the compilation process, hardware development toolchains, such as Xilinx's Vivado or Intel's Quartus, are used to convert protocol parameter changes into synthesizable hardware description code or IP parameter configurations based on the FPGA's hardware architecture and resource availability. Combined with region boundary constraints and interface signal fixing rules, configuration files conforming to bitstream format specifications are generated. The local configuration results generated for each reconfigurable region are encapsulated into independently loadable bitstream slice files, embedding version numbers, region identifiers, and integrity verification information to ensure that the slice file can be individually identified and dynamically loaded by the FPGA reconfiguration controller, enabling local updates and online replacement of protocol functional modules. The bitstream slice file is a binary configuration fragment generated for a specific PRR, containing only reconfiguration information for local logic resources to avoid affecting global logic.

[0082] The generated bitstream slice file is distributed through a secure channel. The secure channel adopts an encrypted communication mechanism, such as TLS / SSL or an identity authentication mechanism based on a hardware encryption module, to ensure that data tampering or illegal injection is prevented during transmission. When the heterogeneous computing protocol fusion unit receives a valid bitstream slice file, the dynamic reconfiguration controller inside the FPGA pauses the operation of the corresponding PRR according to the region address and reconstruction control instruction, loads the new bitstream slice, and completes the replacement of local logic resources, thereby realizing the online update and function switching of the electric vehicle charging protocol and completing the real-time fusion and dynamic reconstruction of the protocol at the hardware level.

[0083] By compiling the incremental instruction stream of the protocol configuration into a bit stream slice file and dynamically loading it using the partially reconfigurable characteristics of the FPGA, the charging protocol can be updated quickly and accurately without interrupting the operation of the charging pile. This enables online fusion and switching between different protocols, which not only improves the adaptability and flexibility of protocol conversion and meets the diverse needs of charging protocols in different scenarios, but also reduces the downtime of V2G charging piles and improves their reliability and availability.

[0084] Example 2, based on the same inventive concept as the adaptive conversion method for electric vehicle charging protocols in the foregoing examples, such as... Figure 2As shown, this application provides an adaptive conversion system for electric vehicle charging protocols, wherein the adaptive conversion system for electric vehicle charging protocols includes: The computing architecture construction module 11 is used to deploy a high-frequency micro-current sensor array and a heterogeneous computing protocol fusion unit in a V2G smart charging pile with protocol reconfigurability to construct a two-level heterogeneous computing architecture of protocol perception and protocol fusion. The data acquisition module 12 is used to capture the micro-current ripple characteristic waveform of the communication carrier during the coupling and communication handshake stages of the charging connector in real time through the high-frequency micro-current sensor array based on the two-level heterogeneous computing architecture. At the same time, it obtains the multi-protocol parallel conflict risk sequence in the future scheduling cycle through the V2G protocol conflict prediction digital twin deployed in the cloud, and synchronously collects the full-dimensional operating status vector of the charging pile through the embedded system in the pile. The driving vector generation module 13 is used to generate the micro-current ripple characteristic waveform of the communication carrier. The system performs super-resolution feature extraction to generate an initial protocol fingerprint probability tensor, and combines it with the multi-protocol parallel conflict risk sequence for spatiotemporal alignment and correlation analysis to dynamically identify key reconfigurable protocol modules and their expected evolution paths, and fuses them to generate a protocol fusion driving vector; the data prediction module 14 is used to analyze the nonlinear coupling relationship between the charging pile's full-dimensional operating state vector and the protocol fusion driving vector, and predict the optimal protocol parameter evolution trajectory under the FPGA partial reconfigurability constraint; the comparison simulation module 15 is used to perform real-time in-loop compatibility comparison and conflict resolution simulation with the target V2G service protocol in a high-fidelity virtual interactive environment in the cloud based on the optimal protocol parameter evolution trajectory, and dynamically generate the FPGA partial reconfigurable bitstream slice sequence.

[0085] Furthermore, the driving vector generation module 13 is also used to: extract multi-dimensional protocol fingerprint features within each time window by using a sliding time window complex wavelet transform on the microcurrent ripple feature waveform of the communication carrier; calculate the similarity between the multi-dimensional protocol fingerprint features and the protocol feature knowledge graph updated by federated learning through a lightweight protocol fingerprint matching engine, and output a multi-protocol confidence distribution matrix indexed by time as the initial protocol fingerprint probability tensor; decode the multi-protocol parallel conflict risk sequence into protocol state transition constraints, and solve the constraint satisfaction with the multi-protocol confidence distribution matrix to identify key reconfigurable protocol modules and predict their adjustment direction and magnitude, forming a protocol fusion driving vector.

[0086] Furthermore, the data prediction module 14 is also used to: establish a composite prediction model containing the full-dimensional operating state vector of the charging pile and the protocol fusion driving vector, analyze the coupling relationship between state disturbance and protocol requirements, and predict the adjustment trend of protocol parameters; based on the adjustment trend, combined with the hardware constraint model of the reconfigurable region of the FPGA, map and generate a discrete protocol state transition sequence as the optimal protocol parameter evolution trajectory.

[0087] Furthermore, the data prediction module 14 is also used to: decouple the full-dimensional operating state vector of the charging pile in the time-frequency domain, and separate the steady-state baseline component and the transient disturbance component; decompose the protocol fusion driving vector into protocol semantic layer adjustment requirements and communication physical layer adjustment requirements; establish a disturbance-demand coupled state space equation, take the transient disturbance component as input, take the rate of change of the protocol fusion driving vector as the state variable, perform state estimation through a Kalman filter, and solve for the optimal adjustment acceleration of the protocol parameters.

[0088] Furthermore, the data prediction module 14 is also used to: establish a hardware constraint model based on the size of the reconfigurable region of the FPGA and the bitstream reconfiguration delay parameters; input the optimal adjustment acceleration of the protocol parameters into the hardware constraint model to calculate the feasible protocol parameter adjustment increment within the hardware resource and delay boundaries; iteratively deduce the protocol state coordinates of the future time series based on the current protocol state and the protocol parameter adjustment increment; perform sequence optimization on the protocol state coordinates to generate a continuous and hardware-realizable optimal protocol parameter evolution trajectory.

[0089] Furthermore, the comparison simulation module 15 is also used to: construct a high-fidelity virtual interactive environment synchronized with physical charging piles, vehicle BMS, and power grid dispatch instructions; inject the evolution trajectory of the optimal protocol parameters into the virtual interactive environment in the form of incremental protocol configuration instruction streams to drive the dynamic reconstruction of the virtual protocol stack; simulate the execution of the target V2G service protocol in the virtual interactive environment and compare the differences between the output behavior of the virtual protocol stack and the expected behavior of the target protocol in real time to generate a multi-dimensional difference vector; perform closed-loop feedback optimization based on the multi-dimensional difference vector, dynamically adjust the incremental protocol configuration instruction streams until the multi-dimensional difference vectors converge to a preset threshold range; and compile the finally converged protocol configuration instruction streams into a partially reconfigurable bitstream slice sequence for the FPGA.

[0090] Furthermore, the comparison simulation module 15 is also used to: construct a V2G service response simulation model in the high-fidelity virtual interactive environment; take the power grid dispatch command as the input stimulus and dynamically adjust the power parameters based on the simulated vehicle SOC state, run the V2G service response simulation model, and obtain the standard protocol interaction process; perform a spatiotemporal alignment comparison between the dynamic protocol stack behavior driven by the incremental instruction stream of the protocol configuration and the standard protocol interaction process; calculate the difference values ​​of the comparison results in multiple dimensions such as protocol parameters, communication timing, and power response curves, and generate the multidimensional difference vector.

[0091] Furthermore, the comparison simulation module 15 is also used to: construct a digital twin closed-loop control model including a protocol parameter regulator and a communication timing compensator; input the multidimensional difference vector into the digital twin closed-loop control model to calculate the protocol stack parameter correction amount and the communication timing compensation amount; perform rolling optimization on the protocol stack parameter correction amount and the communication timing compensation amount through a model predictive control algorithm, and iteratively simulate in a virtual environment; when the simulation results show that each component of the multidimensional difference vector is less than a preset threshold, determine that the protocol configuration optimization is complete, and lock the current protocol configuration incremental instruction stream.

[0092] Furthermore, the system is also used to: configure the incremental instruction stream of the currently locked protocol, divide it according to the partially reconfigurable region of the target FPGA, compile and generate corresponding, independently loadable bitstream slice files; send the bitstream slice files to the heterogeneous computing protocol fusion unit through a secure channel, trigger the FPGA to perform dynamic partial reconstruction, and complete the online fusion and switching of the charging protocol.

[0093] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The electric vehicle charging protocol adaptive conversion method and specific examples in the foregoing embodiment one are also applicable to the electric vehicle charging protocol adaptive conversion system in this embodiment. Through the foregoing detailed description of the electric vehicle charging protocol adaptive conversion method, those skilled in the art can clearly understand the electric vehicle charging protocol adaptive conversion system in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0094] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0095] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. An adaptive conversion method for electric vehicle charging protocols, characterized in that, The method includes: In V2G smart charging piles with reconfigurable protocols, a high-frequency micro-current sensor array and a heterogeneous computing protocol fusion unit are deployed to construct a two-level heterogeneous computing architecture of protocol perception and protocol fusion. Based on the aforementioned two-level heterogeneous computing architecture, the high-frequency micro-current sensor array captures the micro-current ripple characteristic waveform of the communication carrier during the coupling and handshake phases of the charging connector in real time. Simultaneously, the V2G protocol conflict prediction digital twin deployed in the cloud obtains the multi-protocol parallel conflict risk sequence within the future scheduling cycle, and the charging pile's full-dimensional operating status vector is synchronously collected through the embedded system in the pile. Based on the microcurrent ripple characteristic waveform of the communication carrier, super-resolution feature extraction is performed to generate an initial protocol fingerprint probability tensor. Then, combined with the multi-protocol parallel conflict risk sequence, spatiotemporal alignment and correlation analysis are performed to dynamically identify key reconfigurable protocol modules and their expected evolution paths, and to generate a protocol fusion driving vector. Based on the full-dimensional operating state vector and protocol fusion driving vector of the charging pile, the nonlinear coupling relationship between the two is analyzed, and the evolution trajectory of the optimal protocol parameters under the reconfigurable constraints of the FPGA part is predicted. Based on the evolution trajectory of the optimal protocol parameters, real-time in-loop compatibility comparison and conflict resolution simulation with the target V2G service protocol are performed in a high-fidelity virtual interactive environment in the cloud, and a partially reconfigurable bitstream slice sequence of the FPGA is dynamically generated.

2. The electric vehicle charging protocol adaptive conversion method as described in claim 1, characterized in that, Based on the microcurrent ripple characteristic waveform of the communication carrier, super-resolution feature extraction is performed to generate an initial protocol fingerprint probability tensor. This tensor is then combined with the multi-protocol parallel conflict risk sequence for spatiotemporal alignment and correlation analysis to dynamically identify key reconfigurable protocol modules and their expected evolution paths. Finally, a protocol fusion driving vector is generated, including: For the microcurrent ripple characteristic waveform of the communication carrier, a complex wavelet transform with a sliding time window is used to extract the multidimensional protocol fingerprint features within each time window; The lightweight protocol fingerprint matching engine calculates the similarity between the multi-dimensional protocol fingerprint features and the protocol feature knowledge graph updated by federated learning, and outputs a multi-protocol confidence distribution matrix indexed by time as the initial protocol fingerprint probability tensor. The multi-protocol parallel conflict risk sequence is decoded into protocol state transition constraints, and the constraints are satisfied by solving the multi-protocol confidence distribution matrix. Key reconfigurable protocol modules are identified, and their adjustment direction and magnitude are estimated to form a protocol fusion driving vector.

3. The electric vehicle charging protocol adaptive conversion method as described in claim 1, characterized in that, Based on the full-dimensional operating state vector and protocol fusion driving vector of the charging pile, the nonlinear coupling relationship between the two is analyzed, and the evolution trajectory of the optimal protocol parameters under the reconfigurable constraints of the FPGA part is predicted, including: A composite prediction model is established, which includes the full-dimensional operating state vector of charging piles and the protocol fusion driving vector. The coupling relationship between state disturbances and protocol requirements is analyzed, and the adjustment trend of protocol parameters is predicted. Based on the aforementioned adjustment trend, and combined with the hardware constraint model of the reconfigurable region of the FPGA, a discrete protocol state transition sequence is generated as the optimal protocol parameter evolution trajectory.

4. The electric vehicle charging protocol adaptive conversion method as described in claim 3, characterized in that, A composite prediction model is established, incorporating a full-dimensional operational state vector of the charging pile and a protocol fusion driving vector. This model analyzes the coupling relationship between state disturbances and protocol requirements, and predicts the adjustment trend of protocol parameters, including: The time-frequency domain decoupling of the full-dimensional operating state vector of the charging pile is performed to separate the steady-state baseline component and the transient disturbance component. The protocol fusion driving vector is decomposed into protocol semantic layer adjustment requirements and communication physical layer adjustment requirements; A disturbance-demand coupled state-space equation is established, with transient disturbance components as input and the rate of change of the protocol fusion driving vector as the state variable. The state is estimated using a Kalman filter, and the optimal adjustment acceleration of the protocol parameters is solved.

5. The electric vehicle charging protocol adaptive conversion method as described in claim 4, characterized in that, Based on the aforementioned adjustment trend, and combined with the hardware constraint model of the partially reconfigurable region of the FPGA, a discrete protocol state transition sequence is generated as the optimal protocol parameter evolution trajectory, including: A hardware constraint model is established based on the size of the reconfigurable region of the FPGA and the bitstream reconfiguration delay parameters. The optimal adjustment acceleration of the protocol parameters is input into the hardware constraint model to calculate the feasible protocol parameter adjustment increment within the hardware resource and delay boundaries. Based on the current protocol state and the protocol parameter adjustment increment, the protocol state coordinates of future time series are iteratively deduced; The protocol state coordinates are sequence optimized to generate a continuous and hardware-implementable optimal protocol parameter evolution trajectory.

6. The electric vehicle charging protocol adaptive conversion method as described in claim 1, characterized in that, Based on the evolution trajectory of the optimal protocol parameters, real-time in-loop compatibility comparison and conflict resolution simulation with the target V2G service protocol are performed in a high-fidelity virtual interactive environment in the cloud, dynamically generating a partially reconfigurable bitstream slice sequence for the FPGA, including: Construct a high-fidelity virtual interactive environment that is synchronized with physical charging piles, vehicle BMS, and power grid dispatch instructions; The evolution trajectory of the optimal protocol parameters is injected into the virtual interactive environment in the form of an incremental instruction stream for protocol configuration, driving the dynamic reconstruction of the virtual protocol stack. The target V2G service protocol is simulated and executed in the virtual interactive environment, and the difference between the output behavior of the virtual protocol stack and the expected behavior of the target protocol is compared in real time to generate a multi-dimensional difference vector. Based on the multidimensional difference vector, closed-loop feedback optimization is performed to dynamically adjust the incremental instruction flow of the protocol configuration until the multidimensional difference vector converges to a preset threshold range. The finally converged protocol configuration instruction stream is compiled into a sequence of partially reconfigurable bitstream slices for the FPGA.

7. The electric vehicle charging protocol adaptive conversion method as described in claim 6, characterized in that, The target V2G service protocol is simulated and executed in the virtual interactive environment. The differences between the virtual protocol stack output behavior and the expected behavior of the target protocol are compared in real time to generate a multi-dimensional difference vector, including: In the high-fidelity virtual interactive environment, a V2G service response simulation model is constructed; Using grid dispatch commands as input incentives and dynamically adjusting power parameters based on simulated vehicle SOC status, the V2G service response simulation model is run to obtain the standard protocol interaction process. The dynamic protocol stack behavior driven by the incremental instruction stream of the protocol configuration is compared with the interaction process of the standard protocol in a spatiotemporal alignment. The differences in the comparison results across multiple dimensions, including protocol parameters, communication timing, and power response curves, are calculated to generate the multidimensional difference vector.

8. The electric vehicle charging protocol adaptive conversion method as described in claim 7, characterized in that, Closed-loop feedback optimization is performed based on the multidimensional difference vector, dynamically adjusting the incremental instruction flow of the protocol configuration until the multidimensional difference vector converges to a preset threshold range, including: Construct a digital twin closed-loop control model that includes a protocol parameter regulator and a communication timing compensator; The multidimensional difference vector is input into the digital twin closed-loop control model to calculate the protocol stack parameter correction amount and the communication timing compensation amount; The protocol stack parameter correction amount and communication timing compensation amount are optimized by rolling optimization using a model predictive control algorithm and iteratively simulated in a virtual environment. When the simulation results show that all components of the multidimensional difference vector are less than the preset threshold, the protocol configuration optimization is determined to be complete, and the current protocol configuration incremental instruction stream is locked.

9. The electric vehicle charging protocol adaptive conversion method as described in claim 8, characterized in that, The method further includes: The currently locked protocol configuration incremental instruction stream is compiled and generated according to the partially reconfigurable region of the target FPGA, and the corresponding independently loadable bit stream slice file is generated. The bitstream slice file is sent to the heterogeneous computing protocol fusion unit via a secure channel, triggering the FPGA to perform dynamic partial reconstruction, thereby completing the online fusion and switching of the charging protocol.

10. An adaptive conversion system for electric vehicle charging protocols, characterized in that, The step of implementing the electric vehicle charging protocol adaptive conversion method according to any one of claims 1 to 9, wherein the electric vehicle charging protocol adaptive conversion system comprises: The computing architecture building module is used to deploy a high-frequency micro-current sensor array and a heterogeneous computing protocol fusion unit in V2G smart charging piles with protocol reconfigurability, and to build a two-level heterogeneous computing architecture of protocol perception and protocol fusion. The data acquisition module is used to capture the microcurrent ripple characteristic waveform of the communication carrier during the coupling and handshake stages of the charging connector in real time through the high-frequency microcurrent sensor array based on the two-level heterogeneous computing architecture. At the same time, it obtains the multi-protocol parallel conflict risk sequence in the future scheduling cycle through the V2G protocol conflict prediction digital twin deployed in the cloud, and synchronously collects the full-dimensional operation status vector of the charging pile through the embedded system in the pile. The driving vector generation module is used to perform super-resolution feature extraction based on the microcurrent ripple characteristic waveform of the communication carrier, generate an initial protocol fingerprint probability tensor, and perform spatiotemporal alignment and correlation analysis in combination with the multi-protocol parallel conflict risk sequence to dynamically identify key reconfigurable protocol modules and their expected evolution paths, and fuse them to generate a protocol fusion driving vector. The data prediction module is used to analyze the nonlinear coupling relationship between the full-dimensional operating state vector and the protocol fusion driving vector of the charging pile, and predict the evolution trajectory of the optimal protocol parameters under the reconfigurable constraints of the FPGA part. The comparison simulation module is used to perform real-time in-loop compatibility comparison and conflict resolution simulation with the target V2G service protocol in a high-fidelity virtual interactive environment in the cloud, based on the evolution trajectory of the optimal protocol parameters, and dynamically generate a partially reconfigurable bitstream slice sequence for the FPGA.

Images