V2G charge and discharge data adaptive calibration method and system for overcharge station

Abstract

The invention discloses a V2G charging and discharging data self-adaptive calibration method and system for an overcharge station, and relates to the field of electric vehicle interaction. The method comprises the following steps: S1, acquiring electrical measurement data, generating a battery state observation reference through an independent physical model algorithm, and calculating a physical consistency score in combination with vehicle state parameters; s2, calculating the original credibility and the process credibility based on the comparison result of the electrical measurement data and the battery state observation benchmark; s3, fusing the original credibility, the process credibility and the physical consistency score, and performing dynamic evolution in combination with the historical credibility record of the data source to generate a dynamic maturity parameter; and S4, mapping a corresponding data correction algorithm from the strategy set according to the dynamic maturity parameter, carrying out self-adaptive correction on preset charging and discharging key data, and outputting a calibration result. And a closed-loop governance system is constructed through independent observation benchmark and data credit dynamic evolution, and credible guarantee and self-adaptive accurate calibration of a V2G data source are realized.

Description

Technical Field

[0001] This invention relates to the field of electric vehicle interaction, specifically to a V2G charging and discharging data adaptive calibration method and system for supercharging stations. Background Technology

[0002] With the rapid development of electric vehicles and supercharging stations, vehicle-to-grid (V2G) technology has become a key means to achieve flexible grid interaction and the absorption of renewable energy. In the operation of supercharging stations, accurate and reliable monitoring and measurement of the charging and discharging status of vehicle batteries are fundamental to ensuring safe vehicle-grid interaction, fair trading, and battery health management under the V2G model.

[0003] Currently, the acquisition of relevant data mainly relies on data reported by the vehicle battery management system (VBS) and the metering data of the charging pile itself. However, in actual operation, the estimation algorithm and sensor accuracy of the VBS vary depending on the vehicle model and usage status, resulting in inherent uncertainties and opacities in the reported battery status data. Meanwhile, although charging pile metering data serves as a basis for trade settlement, it is difficult to independently verify the true internal state of the battery. Under the complex dynamic conditions of frequent, rapid, and bidirectional power interaction in V2G, inconsistencies and incoordination are prone to occur among the aforementioned multi-source data. Traditional calibration methods based on fixed models or single data sources struggle to dynamically adapt to the effects of equipment performance degradation, environmental interference, and diverse battery characteristics, thus limiting the accuracy and reliability of supercharging stations as distributed and flexible resources participating in grid regulation.

[0004] Therefore, how to achieve adaptive and reliable calibration of multi-source heterogeneous charge and discharge data in V2G scenarios has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] Based on the shortcomings of the prior art described above, the purpose of this invention is to provide an adaptive calibration method and system for V2G charging and discharging data for supercharging stations, so as to solve the above-mentioned technical problems.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an adaptive calibration method for V2G charging and discharging data for supercharging stations, comprising:

[0007] S1: Use smart meters to acquire electrical measurement data in real time, generate a battery state observation benchmark through an independent physical model and state estimation algorithm, and calculate a physical consistency score by combining the vehicle state parameters uploaded by the vehicle.

[0008] S2: Calculate the original confidence level and process confidence level based on the comparison results between electrical measurement data and battery state observation benchmark;

[0009] S3: Integrates original credibility, process credibility, and physical consistency scores, and dynamically evolves them by combining historical credibility records from the data source to generate dynamic maturity parameters;

[0010] S4: Based on the dynamic maturity parameters, map the corresponding data correction algorithm from the preset strategy set, adaptively correct the preset key charging and discharging data, and output the calibration results with confidence labels.

[0011] The present invention is further configured such that S1 includes: a data acquisition step, a state observation step, and a consistency calculation step;

[0012] The data acquisition steps include:

[0013] Obtain vehicle identifier, timestamp, transmission delay metadata, state of charge, total voltage, total current, battery temperature and battery health value reported by the vehicle battery management system, and construct vehicle status parameters;

[0014] Simultaneously, the charging pile's built-in smart meter acquires electrical measurement data in real time, including current, voltage, cumulative energy, power, and power factor.

[0015] The present invention is further configured such that the state observation step includes:

[0016] Based on the vehicle identification, the open-circuit voltage-state-of-charge mapping relationship and nominal capacity of the corresponding vehicle model are obtained from the preset battery parameter database;

[0017] During the vehicle's stationary phase, the first state-of-charge reference point is determined based on the voltage data and the open-circuit voltage-state-of-charge mapping relationship in the electrical measurement data.

[0018] During the charging and discharging phase, the current data in the electrical measurement data is integrated and combined with the first state of charge reference point and nominal capacity to generate the first independent state of charge sequence.

[0019] Based on electrical measurement data, by using current data as input and voltage data as observation, and utilizing the battery equivalent circuit model and state observer, the second independent state of charge value and battery internal resistance value are estimated in real time.

[0020] Based on the changes in the first independent state of charge sequence and the second independent state of charge value, the battery health status is estimated by comparing it with the nominal capacity.

[0021] A battery state observation benchmark is constructed by combining the second independent state of charge value, the battery internal resistance value, and the battery health state value.

[0022] The present invention is further configured such that the consistency calculation step is:

[0023] Based on the state of charge and battery health values ​​in the vehicle state parameters, and combined with the second independent state of charge and battery health values ​​in the battery state observation benchmark, a physical consistency score is calculated.

[0024] The present invention is further configured such that S2 includes: an initial credibility calculation step and a process credibility calculation step;

[0025] The original credibility calculation steps include:

[0026] Obtain the following metadata from the vehicle status parameters: vehicle identifier, timestamp, and transmission delay.

[0027] Based on the vehicle identification, query the preset equipment quality file to obtain the corresponding initial source quality score;

[0028] Calculate the data freshness score based on timestamps and transmission delay metadata;

[0029] Within a preset time alignment window, the total current and total voltage in the vehicle status parameters are compared with the current and voltage at the corresponding time points in the electrical measurement data. When the difference exceeds the preset conflict threshold, a conflict is marked, and the severity of the conflict is assessed based on the difference.

[0030] By integrating the assessment results of initial source quality score, freshness score, and conflict severity, the raw reliability of electrical measurement data and vehicle condition parameters is generated.

[0031] The present invention is further configured such that the process credibility calculation step includes:

[0032] Electrical measurement data and vehicle status parameters with original confidence labels are organized into time-series data segments in chronological order.

[0033] The time series data segments are normalized and input into a pre-trained spatiotemporal feature extraction network to obtain high-dimensional feature vectors.

[0034] The high-dimensional feature vector is input into a pre-trained normal behavior autoencoder, the reconstruction error of the high-dimensional feature vector is calculated, and the reconstruction error is normalized into an error score.

[0035] The process credibility is generated by weighted fusion of the original credibility and error score.

[0036] When the process credibility is lower than the preset process threshold, a process anomaly flag is output.

[0037] The present invention is further configured such that S3 includes:

[0038] Obtain the original credibility, process credibility, and physical consistency scores;

[0039] Based on the vehicle identifier in the vehicle status parameters, query the historical reliability records of this data source in the local archive;

[0040] Based on the preset fusion rules and historical credibility records, the original credibility, process credibility and physical consistency scores are weighted, fused and dynamically adjusted to generate candidate values ​​for dynamic maturity parameters for the current period.

[0041] The maximum permissible range of change for this evolution is determined based on the data source stability index in the historical reliability records;

[0042] The variation range between the dynamic maturity parameter of the previous period and the candidate value of the dynamic maturity parameter of the current period is calculated. The variation range is constrained according to the maximum allowable variation range to generate the final dynamic maturity parameter.

[0043] The present invention is further configured such that the preset key charging and discharging data includes the following vehicle state parameters: state of charge, total voltage, total current and battery health value.

[0044] The present invention is further configured such that S4 includes:

[0045] Based on the dynamic maturity parameters, the preset calibration strategy mapping table is queried to determine the corresponding target data correction algorithm and confidence generation rules;

[0046] The target data correction algorithm is used to correct the state of charge, total voltage, total current and battery health value in the vehicle state parameters to generate calibrated data;

[0047] Based on the confidence generation rules and dynamic maturity parameters, confidence labels corresponding to the calibrated data are generated.

[0048] The present invention also provides a V2G charge and discharge data adaptive calibration system for supercharging stations, the system comprising:

[0049] Observation and verification module: It uses smart meters to acquire electrical measurement data in real time, generates a battery state observation benchmark through an independent physical model and state estimation algorithm, and calculates a physical consistency score by combining the vehicle state parameters uploaded by the vehicle.

[0050] Credibility assessment module: Calculates the original credibility and process credibility based on the comparison results between electrical measurement data and battery state observation benchmark;

[0051] Credit Evolution Module: Integrates original credibility, process credibility, and physical consistency scores, and dynamically evolves the credit based on historical credibility records from the data source to generate dynamic maturity parameters;

[0052] Calibration execution module: Based on the dynamic maturity parameters, it maps the corresponding data correction algorithm from the preset strategy set, performs adaptive correction on the preset key charging and discharging data, and outputs calibration results with confidence labels.

[0053] This invention provides an adaptive calibration method and system for V2G charging and discharging data for supercharging stations. The method comprises the following steps: S1: Real-time acquisition of electrical measurement data using smart meters; generation of a battery state observation benchmark using an independent physical model and state estimation algorithm; calculation of a physical consistency score based on vehicle state parameters uploaded from the vehicle; S2: Calculation of original confidence and process confidence based on the comparison results between the electrical measurement data and the battery state observation benchmark; S3: Fusion of the original confidence, process confidence, and physical consistency score, combined with historical confidence records from the data source, to generate dynamic maturity parameters; S4: Adaptive correction of preset charging and discharging key data by mapping corresponding data correction algorithms from a preset strategy set according to the dynamic maturity parameters, outputting calibration results with confidence labels. The beneficial effects include:

[0054] Create independent observation benchmarks to break data dependency cycles

[0055] By combining metering data from charging stations with an independent algorithm model, a battery state observation benchmark independent of the vehicle's BMS was constructed. This design fundamentally breaks the unidirectional dependence of traditional calibration methods on BMS data, providing a traceable and auditable objective truth reference for the entire system, ensuring the impartiality and reliability of the calibration basis.

[0056] A dynamic evolution model for data credit is established to achieve personalized calibration: By proposing the "data credibility maturity" parameter and its multi-level evolution mechanism, this mechanism dynamically generates and updates a unique credit identifier for each data source by integrating real-time data performance with long-term historical credit records. This enables calibration strategies to shift from a "one-size-fits-all" approach to a "source-specific" approach, significantly improving the accuracy and adaptability of calibration.

[0057] Constructing a closed-loop governance system of "assessment-adjudication-enforcement": This system elevates data calibration from an isolated algorithmic step to a systematic governance process that spans the entire data lifecycle. It mimics the complete logic of "fact investigation, credit assessment, comprehensive adjudication, and precise enforcement," forming an intelligent closed loop with memory, learning, and evolutionary capabilities. This systematically ensures the final quality and credibility of data products in the complex V2G environment.

[0058] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0059] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0060] Figure 1 A flowchart illustrating an adaptive calibration method for V2G charging and discharging data for supercharging stations, as shown in an exemplary embodiment of the present invention;

[0061] Figure 2 This is a schematic diagram illustrating the structure of a V2G charge and discharge data adaptive calibration system for supercharging stations, as an exemplary embodiment of the present invention. Detailed Implementation

[0062] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0063] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0064] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the invention. However, it will be apparent to those skilled in the art that embodiments of the invention may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the invention.

[0065] Example 1:

[0066] An adaptive calibration method for V2G charging and discharging data for supercharging stations, such as Figure 1 As shown, it includes:

[0067] S1: Use smart meters to acquire electrical measurement data in real time, generate a battery state observation benchmark through an independent physical model and state estimation algorithm, and calculate a physical consistency score by combining the vehicle state parameters uploaded by the vehicle.

[0068] S2: Calculate the original confidence level and process confidence level based on the comparison results between electrical measurement data and battery state observation benchmark;

[0069] S3: Integrates original credibility, process credibility, and physical consistency scores, and dynamically evolves them by combining historical credibility records from the data source to generate dynamic maturity parameters;

[0070] S4: Based on the dynamic maturity parameters, map the corresponding data correction algorithm from the preset strategy set, adaptively correct the preset key charging and discharging data, and output the calibration results with confidence labels.

[0071] The present invention is further configured such that S1 includes: a data acquisition step, a state observation step, and a consistency calculation step;

[0072] The data acquisition steps include:

[0073] Obtain vehicle identifier, timestamp, transmission delay metadata, state of charge, total voltage, total current, battery temperature and battery health value reported by the vehicle battery management system, and construct vehicle status parameters;

[0074] The system simultaneously utilizes the smart meters built into the charging pile to acquire real-time electrical measurement data, including current, voltage, accumulated energy, power, and power factor. Specifically, the data acquisition step synchronously acquires raw data from the vehicle battery management system and the charging pile, laying a multi-source, heterogeneous data foundation and ensuring the integrity and synchronization of all subsequent processing objects. The state observation step generates an observation benchmark for battery state based on the charging pile data using an independent algorithm, thereby creating an auditable source of "objective facts" independent of the vehicle battery management system and solving the circular dependency problem of calibration. The consistency calculation step compares the data reported by the vehicle battery management system with the independent observation benchmark, quantifies its deviation, and generates a "physical consistency score" that can directly measure the reliability of the vehicle battery management system data, providing a core basis for subsequent credit assessment and calibration. The data acquisition process is implemented as follows: First, the charging pile's built-in communication gateway establishes a connection with the vehicle following standard protocols such as ISO 15118. It periodically parses the vehicle's unique identifier, millisecond-accurate timestamps, estimated network transmission latency, and core parameters including state of charge, total battery pack voltage, total current, temperature, and health status from the battery management system's data frames. These parameters constitute the vehicle's required status parameters locally, while simultaneously recording the precise reception time of each data frame. Simultaneously, the charging pile's main controller reads real-time measured output current, voltage, accumulated energy, power, and power factor from calibrated smart meters at a higher frequency via industrial buses, including Modbus TCP, forming electrical measurement data. To achieve time alignment between these two types of heterogeneous data, all devices within the system access a network time protocol service to synchronize their clocks. Using a sliding time window algorithm, the estimated and corrected timestamps of the battery management system data and the meter data timestamps are matched using nearest neighbor matching within the window to form time-synchronized data pairs. Subsequently, a rapid physical range-based reasonableness check is performed on the data pairs, such as checking whether the voltage and current are within the reasonable range specified by the equipment. Finally, the verified and time-aligned vehicle status parameters and electrical measurement data will be sequentially transmitted to the subsequent processing modules as a complete data unit.

[0075] The present invention is further configured such that the state observation step includes:

[0076] Based on the vehicle identification, the open-circuit voltage-state-of-charge mapping relationship and nominal capacity of the corresponding vehicle model are obtained from the preset battery parameter database;

[0077] During the vehicle's stationary phase, the first state-of-charge reference point is determined based on the voltage data and the open-circuit voltage-state-of-charge mapping relationship in the electrical measurement data.

[0078] During the charging and discharging phase, the current data in the electrical measurement data is integrated and combined with the first state of charge reference point and nominal capacity to generate the first independent state of charge sequence.

[0079] Based on electrical measurement data, by using current data as input and voltage data as observation, and utilizing the battery equivalent circuit model and state observer, the second independent state of charge value and battery internal resistance value are estimated in real time.

[0080] Based on the changes in the first independent state of charge sequence and the second independent state of charge value, the battery health status is estimated by comparing it with the nominal capacity.

[0081] A battery state observation benchmark is constructed by combining the second independent state of charge (SOC) value, battery internal resistance value, and battery health state value. Specifically, the state observation steps are implemented through the following methods: First, the system accesses the vehicle model battery parameter database stored in the cloud or locally based on the vehicle's unique identifier: the vehicle identifier. Through database query operations, it obtains a complete data table of battery open-circuit voltage and SOC values ​​that perfectly match the vehicle model, as well as the factory-calibrated nominal battery capacity value. When the vehicle is connected to a charging station but not charging or discharging, the system continuously monitors the current measurement value from the smart meter. If the absolute value of the current is less than 0.5% of the nominal capacity for three consecutive minutes, the vehicle is determined to be in a static state. The arithmetic mean of the voltage measurement values ​​during this period is calculated, and this average voltage value is used as the current battery open-circuit voltage. The corresponding precise SOC value is matched in the pre-obtained data table using a lookup table method, and this precise SOC value is recorded as the first SOC benchmark point. After entering the charging and discharging phase, the system acquires current measurements at a sampling frequency of no less than 10 Hz. It then uses a trapezoidal numerical integration method to integrate the current time series point by point, obtaining the cumulative net charge input or net discharge from the reference point. This cumulative net charge input or net discharge is divided by the battery's nominal capacity obtained from the database; the resulting change is the change in state of charge (SOC). This SOC change is then algebraically added to the first SOC reference point, thereby generating and outputting a continuous first independent SOC sequence in real time. Simultaneously, the system activates an extended Kalman filter state observer based on a first-order equivalent circuit model. This observer uses the real-time acquired current measurements as system input and the real-time acquired voltage measurements as observation comparisons, recursively updating its internal state vector in each processing cycle. One of the observer's core state variables is the estimated second independent SOC, which simultaneously identifies and outputs the battery's estimated ohmic internal resistance online. When the system detects the end of a charging or discharging session and the difference between the start and end values ​​of the first independent state of charge sequence exceeds 50%, a valid capacity test segment is considered complete. The total charge change in this segment is calculated, divided by the battery's nominal capacity, and then multiplied by 100% to calculate the health state estimate characterizing the battery's capacity retention rate. Finally, the system combines the second independent state of charge estimate and the battery ohmic resistance estimate output by the extended Kalman filter observer at each time step with the battery health state estimate calculated after the session, aligns them with the timestamp, and packages them into a battery state observation baseline data package.

[0082] The present invention is further configured such that the consistency calculation step is:

[0083] Based on the state of charge (SOC) and battery health values ​​in the vehicle's state parameters, and combined with the second independent SOC and battery health values ​​in the battery state observation benchmark, a physical consistency score is calculated. Specifically, the consistency calculation steps are implemented as follows: The system first extracts two sets of corresponding core parameters: one set is the SOC and battery health values ​​reported by the vehicle's battery management system at the most recent time point; the other set is contained in the second independent SOC estimate and battery health estimate in the battery state observation benchmark generated at the corresponding time point by the state observation steps. Subsequently, the system calculates the absolute differences between these two sets of corresponding parameters, namely, the SOC difference (obtained by calculating the difference between the SOC value and the second independent SOC estimate) and the health difference (obtained by calculating the difference between the battery health value and the battery health estimate). Then, the system normalizes these two differences by dividing the SOC difference by a range factor 1 (representing 100% SOC) and the health difference by a range factor 1 (representing 100% SOH), thus obtaining two normalized difference values ​​between 0 and 1. Next, the system performs a weighted fusion of the normalized differences in state of charge (SOC) and health status (HS). The weight of HHS is typically lower than that of SOC because health status changes slowly and there are inherent biases between different estimation methods. Specifically, the default weights are set to 0.7 for SOC and 0.3 for HHS. This weighted fusion produces an initial consistency index between 0 and 1, with lower values ​​indicating higher consistency. Finally, the system maps the initial consistency index to a physical consistency score between 0 and 1 using a predefined monotonically decreasing function, such as an exponential decay function, where 1 represents perfect consistency and 0 represents severe deviation.

[0084] The present invention is further configured such that S2 includes: an initial credibility calculation step and a process credibility calculation step;

[0085] The original credibility calculation steps include:

[0086] Obtain the following metadata from the vehicle status parameters: vehicle identifier, timestamp, and transmission delay.

[0087] Based on the vehicle identification, query the preset equipment quality file to obtain the corresponding initial source quality score;

[0088] Calculate the data freshness score based on timestamps and transmission delay metadata;

[0089] Within a preset time alignment window, the total current and total voltage in the vehicle status parameters are compared with the current and voltage at the corresponding time points in the electrical measurement data. When the difference exceeds the preset conflict threshold, a conflict is marked, and the severity of the conflict is assessed based on the difference.

[0090] The system integrates the initial source quality score, freshness score, and conflict severity assessment results to generate the initial credibility of electrical measurement data and vehicle status parameters. Specifically, the initial credibility calculation step assesses the reliability and immediate consistency of the data source, constituting a static single-point review; the process credibility calculation step analyzes whether data behavior patterns conform to historical normality, constituting a dynamic time-series review. The initial credibility calculation step is implemented as follows: First, the system extracts the vehicle identifier, timestamp, and network transmission latency from the received vehicle status parameter data packet. Next, using the vehicle identifier as an index, the system queries the locally stored or cloud-synchronized device quality profile database. This database pre-records the vehicle model corresponding to the identifier and the historical performance data of the battery management system. The system extracts two key values: a basic trust score based on long-term statistics and the mean absolute error (MAE) between the historical reported values ​​and high-precision reference values ​​of the data source. If no data corresponding to the vehicle identifier is found, the vehicle is considered a new user, and the basic trust score is set to a default value of 100, with the MAE set to a default value of 0.05. The initial source quality score is calculated using the formula: Initial Source Quality Score = Base Trust Score / (1 + 10 × Mean Absolute Error). Subsequently, the system calculates the data freshness score, using an exponential decay function to handle transmission delay. The specific formula is: Freshness Score = exp(-Transmission Delay / 1000), where the score is 1 when the delay is 0, and decreases as the delay increases. During consistency comparison, the system uses a 200-millisecond time alignment window, pairing the total voltage and total current in the vehicle status parameters within the window with the voltage and current values ​​in the electrical measurement data that have the closest timestamps. The system dynamically calculates the conflict threshold based on the characteristic parameters of the data source pre-stored in the equipment quality file. This threshold is calculated by the following formula: Dynamic threshold = Basic threshold + α × |Battery temperature - Reference temperature| + β × (|Current| / Nominal current), where the reference temperature is 25 degrees Celsius by default, the nominal current is extracted from the equipment quality file database, and α and β are predefined coefficients, specifically extracted from the equipment quality file database. If the absolute value of the difference between the paired values ​​exceeds the dynamic threshold, a conflict is marked. The severity of the conflict is calculated using a piecewise function: the severity is zero when the difference does not exceed the threshold, increases linearly from 0 to 1 when the difference exceeds the threshold but is within twice the threshold, and is fixed at 1 when the difference exceeds twice the threshold. Finally, the system performs a weighted fusion of the initial source quality score, freshness score, and the value obtained by subtracting the conflict severity from 1. The weights of the three are set to 0.5, 0.3, and 0.2, respectively. The weighted sum is input into an S-shaped function for standardization, and finally outputs a value between 0 and 1 as the original credibility of the vehicle state parameter and the corresponding electrical measurement data.

[0091] The present invention is further configured such that the process credibility calculation step includes:

[0092] Electrical measurement data and vehicle status parameters with original confidence labels are organized into time-series data segments in chronological order.

[0093] The time series data segments are normalized and input into a pre-trained spatiotemporal feature extraction network to obtain high-dimensional feature vectors.

[0094] The high-dimensional feature vector is input into a pre-trained normal behavior autoencoder, the reconstruction error of the high-dimensional feature vector is calculated, and the reconstruction error is normalized into an error score.

[0095] The process credibility is generated by weighted fusion of the original credibility and error score.

[0096] When the process reliability is lower than the preset process threshold, a process anomaly flag is output. Specifically, the process reliability calculation steps are implemented through the following method: First, the system constructs a time-series data segment: based on the current processing time, continuous data with a length of 5 seconds and a sampling rate of 10 Hz is extracted backward, i.e., a data window containing 50 consecutive time points; each data point in this window is composed of the attached original reliability label, the current and voltage values ​​in the electrical measurement data, and the state of charge and total voltage values ​​in the vehicle status parameters. Subsequently, the acquired data segment is normalized. For the sequences of different physical quantities such as current and voltage, their historical statistical mean is subtracted and divided by their historical standard deviation, mapping all values ​​to a similar scale. The processed time series segments are input into a pre-trained lightweight spatiotemporal feature extraction network, which consists of consecutive one-dimensional convolutional layers and gated recurrent unit layers. The one-dimensional convolutional layers use three types of convolutional kernels with widths of 3, 5, and 7 to scan the time series data in parallel and extract local fluctuation patterns and short-term features. The output of the convolutional layers is fed into a single-layer gated recurrent unit with 32 hidden units to capture the long-term dependencies and dynamic trends of the sequence. The network finally outputs a 128-dimensional high-dimensional feature vector as a condensed representation of the system's overall operating mode within that time period. Subsequently, this high-dimensional feature vector is input into a pre-trained normal behavior autoencoder, which consists of a symmetric encoder and decoder. The encoder is a three-layer fully connected network that compresses the 128-dimensional input layer by layer to 64 dimensions, then 32 dimensions, and finally forms a 16-dimensional bottleneck layer feature vector. The decoder performs the reverse process to reconstruct the 16-dimensional vector back to 128 dimensions. This autoencoder is trained in the cloud using massive amounts of historical normal charging session data. The training goal of the autoencoder is to make the reconstructed output as close as possible to the original input. In online computation, the original reconstruction error is obtained by calculating the Euclidean distance between the high-dimensional feature vector and its autoencoder reconstruction output. Then, the original reconstruction error is divided by a dynamically updated normalization factor—the normalization factor is set to the 95th percentile of the reconstruction errors of the most recent 100 normal sessions—thus converting the error into an error score between 0 and 1.5. The higher the score, the more abnormal the current behavior pattern. After obtaining the error score, the system weights and fuses it with the original credibility of the corresponding data window (the median of the original credibility of each point in the window). The calculation formula is: process credibility = 0.6 × original credibility + 0.4 × (1 - error score). This weighting scheme is used to balance the reliability of the data source and the normality of the behavior. When the error score is high, even if the original credibility is high, the process credibility will be significantly reduced.The final generated process credibility is also a value between 0 and 1. During the system's implementation, the system continuously monitors the process credibility. If the value is lower than the preset process anomaly threshold of 0.4, a process anomaly flag is immediately output to the system log and monitoring interface, triggering subsequent manual review or system degradation processing.

[0097] The present invention is further configured such that S3 includes:

[0098] Obtain the original credibility, process credibility, and physical consistency scores;

[0099] Based on the vehicle identifier in the vehicle status parameters, query the historical reliability records of this data source in the local archive;

[0100] Based on the preset fusion rules and historical credibility records, the original credibility, process credibility and physical consistency scores are weighted, fused and dynamically adjusted to generate candidate values ​​for dynamic maturity parameters for the current period.

[0101] The maximum permissible range of change for this evolution is determined based on the data source stability index in the historical reliability records;

[0102] The system calculates the variation range between the dynamic maturity parameter of the previous period and the candidate value of the dynamic maturity parameter of the current period, constrains the variation range according to the maximum allowable variation range, and generates the final dynamic maturity parameter. Specifically, the system first obtains three key inputs calculated from the previous steps: original credibility, process credibility, and physical consistency score. Simultaneously, based on the vehicle identifier contained in the currently processed vehicle status parameters, the system queries the locally stored historical credibility record archive, which is a time-expanding sequence. Each record entry contains the dynamic maturity parameter value of the data source in the historical period, the sequence of each input score, and the calculated stability index. The first step in the evolution process is to generate the candidate value of the dynamic maturity parameter for the current period. The system dynamically weights and fuses the three input scores according to a preset fusion rule. The specific rule is as follows: First, based on the process credibility and physical consistency score sequences of the most recent 20 periods in the historical record, the corresponding coefficient of variation is calculated, which is used as the "short-term volatility" indicator. The weighting formula is: Candidate value = w1 * original credibility + w2 * process credibility + w3 * physical consistency score. The weights w1, w2, and w3 are not fixed but dynamically adjusted based on the "identity" of the data source: For stable data sources with complete records (number of records > 50) and low short-term volatility (coefficient of variation < 0.1), the weights are biased towards process credibility and physical consistency scores (e.g., w1 = 0.2, w2 = 0.4, w3 = 0.4), emphasizing behavioral and physical consistency; for new data sources or data sources with high volatility (coefficient of variation > 0.3), the weights are biased towards original credibility (e.g., w1 = 0.5, w2 = 0.3, w3 = 0.2), relying more on the assessment of its underlying source. This dynamic adjustment logic is the core of the preset fusion rules. Next, the system calculates the historical stability index of the data source to determine the maximum allowable maturity change range. The stability index is jointly determined by the long-term trend stability of the dynamic maturity parameter in the historical record (e.g., the standard deviation of the first derivative over the past 100 periods) and short-term volatility. The formula for calculating the maximum allowable change range is: Maximum allowable change range = Upper limit of basic change * (1 - stability index). The basic change limit is set at 0.2 (i.e., 20%), and the stability index is normalized to between 0 and 1 (1 representing the most stable). Therefore, for a completely new or extremely unstable data source, the maximum allowable change can be close to 0.2, allowing for rapid learning and adjustment; for an extremely stable data source, the maximum allowable change may be less than 0.05 to prevent interference from accidental noise. Subsequently, the system calculates the final dynamic maturity parameter after constraints. The absolute change value is calculated as: |Current period candidate value - Previous period maturity parameter|.The final dynamic maturity parameter is generated according to the following rules: if the absolute change value is less than or equal to the maximum allowable change range, then the dynamic maturity parameter = the candidate value of the current period; if the absolute change value is greater than the maximum allowable change range, then the dynamic maturity parameter = the maturity parameter of the previous period + sign(candidate value of the current period - maturity parameter of the previous period) * maximum allowable change range, that is, the change range is clamped to the maximum allowable value, while the direction remains unchanged. This constraint mechanism effectively prevents the maturity parameter from jumping due to drastic fluctuations in a single score, ensuring the smoothness and reliability of its evolution. Finally, the system uses the dynamic maturity parameter calculated this time to update the local archive: the dynamic maturity parameter, the original credibility input in this period, the process credibility, the physical consistency score, and the calculated stability index and actual change Δ are added together as a new record to the historical credibility record of the vehicle identifier, providing a basis for the evolution of the next period. If there is no historical archive for the queried vehicle identifier, it is regarded as a new data source, the maturity parameter of the previous period is initialized to 0.5, and a new archive containing the current record is created. The entire process achieves a closed-loop, adaptive, and robust evolution of dynamic maturity parameters.

[0103] The present invention is further configured such that the preset key charging and discharging data includes the following vehicle state parameters: state of charge (SCC), total voltage, total current, and battery health value. Specifically, SCC, total voltage, total current, and battery health value are chosen as key charging and discharging data because these four parameters together constitute the four essential technical pillars of safety, metering, control, and lifespan management in V2G interaction. Among them, SCC, as a battery capacity indicator, determines the charging and discharging boundaries and scheduling strategies. The accuracy of this value is crucial for preventing overcharging and over-discharging safety risks, eliminating user range anxiety, and resolving settlement disputes. Total voltage and total current are the direct objects of real-time power control and the metering source for economic settlement. The accuracy of these two parameters directly relates to the quality of grid command compliance and the fairness of economic transactions. Battery health value represents the slow-changing state of battery capacity decay and internal resistance increase due to use, providing a necessary correction benchmark for accurately interpreting SCC changes, correcting voltage model parameters, and evaluating long-term battery performance.

[0104] The present invention is further configured such that S4 includes:

[0105] Based on the dynamic maturity parameters, the preset calibration strategy mapping table is queried to determine the corresponding target data correction algorithm and confidence generation rules;

[0106] The target data correction algorithm is used to correct the state of charge, total voltage, total current and battery health value in the vehicle state parameters to generate calibrated data;

[0107] Based on the confidence generation rules and dynamic maturity parameters, confidence labels corresponding to the calibrated data are generated. Specifically, firstly, the system queries a pre-defined calibration strategy mapping table based on the dynamic maturity parameters. This calibration strategy mapping table is a piecewise linear mapping relationship, dividing the dynamic maturity parameters into four continuous intervals and associating them with different processing strategies: when the parameter value is between 0 and 0.3, it maps to the "conservative observation substitution" strategy; between 0.3 and 0.6, it maps to the "weighted fusion" strategy; between 0.6 and 0.9, it maps to the "model-guided correction" strategy; and between 0.9 and 1.0, it maps to the "minimal intervention" strategy. Each strategy predefines a corresponding target data correction algorithm and confidence generation rules. Taking the most typical "weighted fusion" strategy as an example, when the algorithm is determined to be using this strategy, the target data correction algorithm performs the following operations: For the state of charge (SOC) in the vehicle state parameters, the algorithm calls the second independent SOC estimate generated in the state observation step, and performs a weighted average of the reported SOC and this independent estimate. The weight is determined by the dynamic maturity parameter; the higher the maturity, the higher the weight of the reported value. The calculation formula is: calibrated SOC = dynamic maturity parameter × reported SOC + (1 - dynamic maturity parameter) × independent estimated SOC. For total voltage and total current, the algorithm performs a weighted average of the vehicle's reported value and the corresponding value in the electrical measurement data, with the weight allocation method being the same as above. For battery health values, the algorithm adopts a long-term tracking method. After every five complete charge-discharge cycles, the algorithm smoothly updates the latest health state estimate calculated based on the observation benchmark with the current value, with the smoothing coefficient set to 0.2. Subsequently, confidence labels are generated according to the confidence generation rules of the "weighted fusion" strategy. These rules stipulate that the basic confidence is directly equal to the dynamic maturity parameter. Based on this, the algorithm calculates the overall dispersion among the calibrated data, the original reported data, and the independent observation data: if the calibrated data is closer to the independent observation data, the base confidence level is increased by up to 10%; if it is closer to the original reported data, the base confidence level remains unchanged; if the calibrated data is in between and the differences among the three are significant, the base confidence level is decreased by up to 5%. A confidence value between 0 and 1 is generated and converted into three text labels: "High" (≥0.8), "Medium" (0.5-0.8), and "Low" (<0.5), which are then appended to the calibrated data record. The entire calibration process, along with the generated calibrated data and confidence labels, is recorded in an immutable log with timestamps and digital signatures, completing this adaptive calibration. If the dynamic maturity parameter is below 0.1, the system will not perform calibration and will directly output an anomaly flag.

[0108] Example 2:

[0109] Please see Figure 2This exemplary V2G charge / discharge data adaptive calibration system for supercharging stations includes:

[0110] Observation and verification module: It uses smart meters to acquire electrical measurement data in real time, generates a battery state observation benchmark through an independent physical model and state estimation algorithm, and calculates a physical consistency score by combining the vehicle state parameters uploaded by the vehicle.

[0111] Credibility assessment module: Calculates the original credibility and process credibility based on the comparison results between electrical measurement data and battery state observation benchmark;

[0112] Credit Evolution Module: Integrates original credibility, process credibility, and physical consistency scores, and dynamically evolves the credit based on historical credibility records from the data source to generate dynamic maturity parameters;

[0113] Calibration execution module: Based on the dynamic maturity parameters, it maps the corresponding data correction algorithm from the preset strategy set, performs adaptive correction on the preset key charging and discharging data, and outputs calibration results with confidence labels.

[0114] It should be noted that the V2G charge / discharge data adaptive calibration system for supercharging stations provided in the above embodiments and the V2G charge / discharge data adaptive calibration method for supercharging stations provided in the above embodiments belong to the same concept. The specific operation methods of each module and unit have been described in detail in the method embodiments and will not be repeated here. In practical applications, the V2G charge / discharge data adaptive calibration system for supercharging stations provided in the above embodiments can be assigned to different functional modules as needed, that is, the internal structure of the system can be divided into different functional modules to complete all or part of the functions described above. This is not a limitation here.

[0115] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An adaptive calibration method for V2G charging and discharging data for supercharging stations, characterized in that, include: S1: Use smart meters to acquire electrical measurement data in real time, generate a battery state observation benchmark through an independent physical model and state estimation algorithm, and calculate a physical consistency score by combining the vehicle state parameters uploaded by the vehicle. S2: Calculate the original confidence level and process confidence level based on the comparison results between electrical measurement data and battery state observation benchmark; S3: Integrates original credibility, process credibility, and physical consistency scores, and dynamically evolves them by combining historical credibility records from the data source to generate dynamic maturity parameters; S4: Based on the dynamic maturity parameters, map the corresponding data correction algorithm from the preset strategy set, adaptively correct the preset key charging and discharging data, and output the calibration results with confidence labels.

2. The adaptive calibration method for V2G charging and discharging data for supercharging stations according to claim 1, characterized in that, S1 includes: a data acquisition step, a state observation step, and a consistency calculation step; The data acquisition steps include: Obtain vehicle identifier, timestamp, transmission delay metadata, state of charge, total voltage, total current, battery temperature and battery health value reported by the vehicle battery management system, and construct vehicle status parameters; Simultaneously, the charging pile's built-in smart meter acquires electrical measurement data in real time, including current, voltage, cumulative energy, power, and power factor.

3. The adaptive calibration method for V2G charging and discharging data for supercharging stations according to claim 2, characterized in that, The state observation steps include: Based on the vehicle identification, the open-circuit voltage-state-of-charge mapping relationship and nominal capacity of the corresponding vehicle model are obtained from the preset battery parameter database; During the vehicle's stationary phase, the first state-of-charge reference point is determined based on the voltage data and the open-circuit voltage-state-of-charge mapping relationship in the electrical measurement data. During the charging and discharging phase, the current data in the electrical measurement data is integrated and combined with the first state of charge reference point and nominal capacity to generate the first independent state of charge sequence. Based on electrical measurement data, by using current data as input and voltage data as observation, and utilizing the battery equivalent circuit model and state observer, the second independent state of charge value and battery internal resistance value are estimated in real time. Based on the changes in the first independent state of charge sequence and the second independent state of charge value, the battery health status is estimated by comparing it with the nominal capacity. A battery state observation benchmark is constructed by combining the second independent state of charge value, the battery internal resistance value, and the battery health state value.

4. The adaptive calibration method for V2G charging and discharging data for supercharging stations according to claim 3, characterized in that, The consistency calculation steps are as follows: Based on the state of charge and battery health values ​​in the vehicle state parameters, and combined with the second independent state of charge and battery health values ​​in the battery state observation benchmark, a physical consistency score is calculated.

5. The adaptive calibration method for V2G charging and discharging data for supercharging stations according to claim 1, characterized in that, S2 includes: the original credibility calculation step and the process credibility calculation step; The original credibility calculation steps include: Obtain the following metadata from the vehicle status parameters: vehicle identifier, timestamp, and transmission delay. Based on the vehicle identification, query the preset equipment quality file to obtain the corresponding initial source quality score; Calculate the data freshness score based on timestamps and transmission delay metadata; Within a preset time alignment window, the total current and total voltage in the vehicle status parameters are compared with the current and voltage at the corresponding time points in the electrical measurement data. When the difference exceeds the preset conflict threshold, a conflict is marked, and the severity of the conflict is assessed based on the difference. By integrating the assessment results of initial source quality score, freshness score, and conflict severity, the original credibility of electrical measurement data and vehicle condition parameters is generated.

6. The adaptive calibration method for V2G charging and discharging data for supercharging stations according to claim 5, characterized in that, The process reliability calculation steps include: Electrical measurement data and vehicle status parameters with original confidence labels are organized into time-series data segments in chronological order. The time series data segments are normalized and input into a pre-trained spatiotemporal feature extraction network to obtain high-dimensional feature vectors. The high-dimensional feature vector is input into a pre-trained normal behavior autoencoder, the reconstruction error of the high-dimensional feature vector is calculated, and the reconstruction error is normalized into an error score. The process credibility is generated by weighted fusion of the original credibility and error score. When the process credibility is lower than the preset process threshold, a process anomaly flag is output.

7. The adaptive calibration method for V2G charging and discharging data for supercharging stations according to claim 1, characterized in that, S3 includes: Obtain the original credibility, process credibility, and physical consistency scores; Based on the vehicle identifier in the vehicle status parameters, query the historical reliability records of this data source in the local archive; Based on the preset fusion rules and historical credibility records, the original credibility, process credibility and physical consistency scores are weighted, fused and dynamically adjusted to generate candidate values ​​for dynamic maturity parameters for the current period. The maximum permissible range of change for this evolution is determined based on the data source stability index in the historical reliability records; The variation range between the dynamic maturity parameter of the previous period and the candidate value of the dynamic maturity parameter of the current period is calculated. The variation range is constrained according to the maximum allowable variation range to generate the final dynamic maturity parameter.

8. The adaptive calibration method for V2G charging and discharging data for supercharging stations according to claim 1, characterized in that, The preset key charging and discharging data include the following vehicle status parameters: state of charge, total voltage, total current, and battery health value.

9. The adaptive calibration method for V2G charging and discharging data for supercharging stations according to claim 8, characterized in that, S4 includes: Based on the dynamic maturity parameters, the preset calibration strategy mapping table is queried to determine the corresponding target data correction algorithm and confidence generation rules; The target data correction algorithm is used to correct the state of charge, total voltage, total current and battery health value in the vehicle state parameters to generate calibrated data; Based on the confidence generation rules and dynamic maturity parameters, confidence labels corresponding to the calibrated data are generated.

10. A V2G charge / discharge data adaptive calibration system for supercharging stations, used to implement the V2G charge / discharge data adaptive calibration method for supercharging stations as described in any one of claims 1-9, characterized in that, include: Observation and verification module: It uses smart meters to acquire electrical measurement data in real time, generates a battery state observation benchmark through an independent physical model and state estimation algorithm, and calculates a physical consistency score by combining the vehicle state parameters uploaded by the vehicle. Credibility assessment module: Calculates the original credibility and process credibility based on the comparison results between electrical measurement data and battery state observation benchmark; Credit Evolution Module: Integrates original credibility, process credibility, and physical consistency scores, and dynamically evolves the credit based on historical credibility records from the data source to generate dynamic maturity parameters; Calibration execution module: Based on the dynamic maturity parameters, it maps the corresponding data correction algorithm from the preset strategy set, performs adaptive correction on the preset key charging and discharging data, and outputs calibration results with confidence labels.

Images