Self-adaptive regulation and control system for vehicle network interaction of large-scale ultrafast charging facility

Abstract

The invention provides a large-scale ultrafast charging facility vehicle network interaction-oriented adaptive regulation and control system, and belongs to the technical field of charging pile load scheduling. During working, a multi-source data acquisition and vehicle network state sensing module acquires and fuses multiple types of data, and a vehicle network interaction strategy optimization module generates a strategy according to the data; the power dynamic distribution module distributes power according to a strategy, and the energy storage coordinated control module optimizes energy storage charging and discharging. The real-time response and feedback module quickly responds to a power grid instruction and corrects a strategy, and the security and protocol adaptation module guarantees security and docking. During use, after the terminal is started, fusion data is automatically collected, a regulation and control strategy is generated and executed, instructions are responded and corrected in real time, and safety and cross-platform adaptation are guaranteed in the whole process.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of vehicle network interaction, and specifically relates to a self-adaptive regulation and control system for large-scale super-fast charging facility vehicle network interaction. BACKGROUND

[0002] In the field of large-scale super-fast charging facility vehicle network interaction, the prior art has many problems. In terms of vehicle network state perception, the traditional technology relies on a single data source, the data acquisition architecture is simple, and there is a lack of effective fusion algorithm, resulting in large time lag error, low SOC measurement accuracy, distorted vehicle network state evaluation, and inability to provide accurate data support for regulation and control.

[0003] The existing regulation and control strategy focuses on a single target and does not consider the balance of interests of the power grid, users and operators, making the interests of the parties in conflict, and it is difficult to simultaneously consider power grid stability, user charging efficiency and operator revenue. In addition, the traditional single-pile exclusive mode cannot break through the transformer capacity limit, and in the large-scale super-fast charging scene, the charging pile utilization rate is low, the user waiting time is long, and the resource waste is serious.

[0004] The energy storage collaborative regulation has defects, and the use of a single battery type either has slow response speed or insufficient capacity, making it difficult to respond to sudden power impact and also not able to play a good role during peak and valley periods, resulting in large power grid capacity investment and high operator electricity purchase cost. SUMMARY

[0005] To solve the problems raised in the background art, the application provides a self-adaptive regulation and control system for large-scale super-fast charging facility vehicle network interaction to solve the problems of unstable charging strategy, low charging pile utilization rate and high cost.

[0006] To achieve the above purpose, the application provides the following technical solution: a self-adaptive regulation and control system for large-scale super-fast charging facility vehicle network interaction, characterized by comprising a multi-source data acquisition and vehicle network state perception module, a vehicle network interaction strategy optimization module, a power dynamic allocation module, an energy storage collaborative regulation module, a real-time response and feedback module and a safety and protocol adaptation module, wherein:

[0007] The multi-source data acquisition and vehicle network state perception module: through distributed sensing network and heterogeneous interface technology, a three-level data acquisition architecture of end-edge-cloud is constructed to collect charging pile operation data, electric vehicle battery state, power distribution network operation parameters and user charging demand information in real time;

[0008] The vehicle-network interaction strategy optimization module: based on real-time state of vehicle-network and user demand prediction, a multi-agent collaborative optimization model is constructed, and a multi-agent reinforcement learning algorithm is used to generate a vehicle-network interaction strategy. This module breaks through the limitations of traditional single-objective optimization and realizes the multi-objective balance of user charging efficiency, grid peak shaving contribution and operator revenue under the premise of meeting the safety constraints of the distribution network.

[0009] The power dynamic allocation module: for the concurrent super-charging demand of large-scale super-fast charging piles, based on the vehicle-network interaction strategy and real-time capacity constraints of the distribution network, a dynamic priority scheduling algorithm is used to realize intelligent power allocation, and through the power pool of shared transformers and energy storage systems, it supports multiple piles to operate at full power at the same time, solving the contradiction between transformer capacity limitation and user super-charging demand.

[0010] The energy storage collaborative control module: linking lithium iron phosphate battery energy storage and supercapacitor energy storage, a double-layer energy storage system with long-term energy storage and short-term buffer is constructed, and a model predictive control algorithm is used to optimize the charging and discharging plan. This module releases energy storage power to support super-charging during the peak load of the grid, avoiding the investment in capacity expansion of the grid; it uses valley electricity for charging during the low load period, reducing the operator's electricity purchase cost; in the event of sudden power impact, the supercapacitor responds quickly to stabilize the power fluctuation impact on the grid.

[0011] Real-time response and feedback module: through edge computing nodes, the grid command is quickly responded, and a closed-loop control is formed combined with terminal sensing data;

[0012] Safety and protocol adaptation module: a three-layer security protection system of data-communication-device is constructed to ensure data privacy and system security during vehicle-network interaction; at the same time, a multi-protocol conversion interface is built-in to support seamless connection with grid platforms at all levels, vehicle enterprise systems and government supervision platforms, meeting the cross-agent collaboration and policy compliance requirements.

[0013] Optionally, the multi-source data acquisition and vehicle-network state sensing module includes a multi-source heterogeneous data acquisition sub-module and a vehicle-network state fusion sensing sub-module.

[0014] The multi-source heterogeneous data acquisition sub-module collects real-time charging power, voltage and current through Hall sensors deployed on the direct current side of the charging pile; communicates with the electric vehicle BMS system through the OBD-II interface to obtain battery SOC, SOH, real-time temperature and charging and discharging cutoff voltage; collects side voltage, current, power factor and frequency through the distribution network FTU; at the same time, it accesses the charging reservation time, target SOC and payment mode submitted by the user through the APP, supports CAN bus, 5G, LoRa and other multi-protocol parallel transmission;

[0015] The vehicle-to-network (V2N) state fusion perception submodule employs a weighted Bayesian fusion algorithm for data calibration and fusion to address the time lag and noise interference present in multi-source data. First, various types of data are preprocessed, and then a V2N state vector is generated using a dynamic weight allocation formula. The V2N state vector is as follows:

[0016]

[0017] in, Let be the credibility weight of the i-th type of data source. The original data value, This is the historical moving average of this type of data. =3 is the dynamic correction coefficient. The fused state vector includes the grid safety margin, charging demand intensity and equipment health, providing accurate state input for subsequent regulation.

[0018] Optionally, the vehicle-to-grid interaction strategy optimization module includes: a multi-agent game optimization submodule and an intelligent charging and discharging prediction submodule;

[0019] The multi-party game optimization submodule constructs a non-cooperative game model among the power grid, users, and operators, clarifying the payoff functions and constraints for each party. The power grid aims to minimize network losses and achieve frequency stability; its payoff function is as follows:

[0020]

[0021] in, For the target load of the power grid, This is the actual load. The line loss is caused by V2G discharge. =0.6, which is the peak-shaving weight. =0.4, which is the network loss weight.

[0022] The user's objective is to minimize charging time and battery degradation; the revenue function is:

[0023]

[0024] in, =0.5, which is the target weight of SOC. =0.3, representing the time weight. =0.2, which is the battery loss weight. This refers to the percentage of the battery's current remaining charge relative to its total capacity. The percentage of battery charge that the user expects to reach after charging is complete. This refers to the cumulative charging time from the start of charging to the current moment. The amount of loss due to battery health status is estimated using a cycle count model.

[0025] The operator aims to maximize the operating income, and the income function is:

[0026] wherein, = 0.7 is the weight of power sales income, = 0.3 is the weight of V2G subsidy, is the charging power, is the charging price, is the power purchase from the grid, is the power purchase price, is the V2G discharge power, is the V2G subsidy standard.

[0027] The three parties balance the interests through Nash equilibrium, and meet:

[0028] The intelligent charging and discharging prediction sub-module adopts an LSTM algorithm to predict future charging load and V2G available capacity. The model input features include historical charging and discharging data, user travel regularity, meteorological data, and date type. The prediction accuracy is evaluated by the relative error mean:

[0029]

[0030] wherein, T is the number of prediction periods, is the actual charging and discharging power at t moment, is the predicted power.

[0031] Optionally, in some embodiments, the power dynamic allocation module further comprises a priority evaluation sub-module and a dynamic power allocation sub-module.

[0032] The priority evaluation sub-module constructs a three-dimensional priority evaluation system to dynamically adjust the priority of user charging urgency, battery health status, and grid load rate. The user urgency is calculated by the difference between the reservation time and the current time; the battery health status is obtained by BMS data, and the battery with SOH≤80% reduces the priority to reduce the cycle loss; when the grid load rate exceeds 0.8, the priority of the grid-friendly user is improved. The priority calculation formula is:

[0033]

[0034] wherein, is the user urgency, the user charging within 10 minutes is 1, is 0.9, the user charging after 1 hour is 0.3, is the battery health status, is the grid load rate.

[0035] The dynamic power allocation submodule allocates the total available power of the power grid and the energy storage system based on the priority results obtained by the priority evaluation submodule using a proportional fairness algorithm. According to the priority proportion of each charging pile, the total available power is allocated to each charging pile. During the allocation process, it is strictly ensured that the allocated power of each charging pile does not exceed its own demand power, and also does not exceed the total available power of the system, so as to realize the reasonable allocation of power. When the system detects that the power grid load rate is overloaded, the power reduction mechanism is triggered immediately, and the charging piles with lower priority are preferentially reduced in power. The reduction range is determined according to the extent of the overload of the power grid load rate.

[0036] Optionally, in some embodiments, the energy storage coordination and control module further comprises an energy storage state monitoring submodule and an energy storage and energy storage and charging coordination optimization submodule.

[0037] The energy storage state monitoring submodule collects key parameters of the energy storage system in real time, including lithium battery SOC, charging and discharging power, cycle number and temperature; voltage, instantaneous power and health status of supercapacitor. The available capacity of the energy storage system is calculated as follows:

[0038]

[0039] Wherein = 20% is a protection threshold to ensure that the charging and discharging depth of the energy storage system does not exceed 80%, refers to the actual available capacity of the energy storage system, refers to the rated capacity of the energy storage system.

[0040] The energy storage and charging coordination optimization submodule uses the MPC algorithm to rollingly optimize the energy storage charging and discharging plan for the future period. During the optimization process, the target is to make the grid interaction power as close as possible to the target flattening power and the energy storage power change rate as small as possible. Through such an optimization method, the stable operation of the power grid is ensured, the impact of power fluctuation is reduced, the loss of energy storage equipment due to frequent charging and discharging is reduced, and the service life of the equipment is prolonged.

[0041] Optionally, the real-time response and feedback module further comprises a power grid instruction analysis submodule and a closed-loop correction submodule.

[0042] The power grid instruction analysis submodule receives power grid control instructions, including frequency modulation signals, reactive power compensation requirements and emergency load shedding commands, through the power distribution grid dispatching interface, and converts the frequency modulation signals into power adjustment amount:

[0043]

[0044] Wherein, = 100kW / Hz is the frequency response coefficient, is the frequency modulation signal.

[0045] The closed-loop correction submodule compares the actual response power with the grid instruction target value, calculates the deviation between the two, and records the reasons for the deviation in detail, and optimizes the strategy in the next control period.

[0046] Compared with the prior art, the present application has the following beneficial effects:

[0047] In the vehicle network state perception, through the end-edge-cloud three-level acquisition architecture and the weighted Bayesian fusion algorithm, multi-dimensional data is integrated, the time delay error is reduced, the SOC measurement accuracy is improved, high-fidelity data foundation is provided for regulation and control, and the problem of vehicle network state evaluation distortion is effectively solved.

[0048] The three-party game multi-objective collaborative optimization mechanism in the application solves the three-party benefit balance point through Nash equilibrium, realizes the collaborative improvement of user charging efficiency and operator income while ensuring the stability of the power grid frequency, and solves the problem of multi-party interest conflict;

[0049] The dynamic allocation strategy of the power pool sharing in the application combines the priority evaluation system and the proportional fairness algorithm, supports the simultaneous operation of multiple super-charging piles, greatly improves the utilization rate of the charging pile, and solves the problems of resource waste and user waiting caused by single pile monopoly;

[0050] The double-layer structure of lithium iron phosphate battery and super capacitor in the application optimizes the charging and discharging plan through MPC algorithm, reduces the capacity expansion cost of power grid by releasing energy at peak, reduces the power purchase cost by charging at low valley, and can quickly respond to sudden power impact, thereby improving the suppression amplitude of power grid fluctuation, and balancing capacity and response speed. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 The overall system and method flowchart in the application;

[0053] Figure 2 The multi-source data acquisition and vehicle network state perception module flowchart in the application;

[0054] Figure 3 The vehicle network interaction strategy optimization module flowchart in the application;

[0055] Figure 4 The power dynamic allocation module flowchart in the application;

[0056] Figure 5 The energy storage collaborative regulation module flowchart in the application;

[0057] Figure 6 The real-time response and feedback module flowchart in the application;

[0058] Fig.:

[0059] 101, multi-source data acquisition and vehicle network state perception module; 102, vehicle network interaction strategy optimization module; 103, power dynamic allocation module; 104, energy storage collaborative control module; 105, real-time response and feedback module; 106, safety and protocol adaptation module. DETAILED DESCRIPTION

[0061] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0062] As Figures 1 to 6 shown, the present application provides an adaptive control system for large-scale ultra-fast charging facility vehicle network interaction, characterized in that it comprises a multi-source data acquisition and vehicle network state perception module 101, a vehicle network interaction strategy optimization module 102, a power dynamic allocation module 103, an energy storage collaborative control module 104, a real-time response and feedback module 105, and a safety and protocol adaptation module 106, wherein:

[0063] The multi-source data acquisition and vehicle network state perception module 101: through distributed sensing network and heterogeneous interface technology, a three-level data acquisition architecture of end-edge-cloud is constructed to collect real-time charging pile operation data, electric vehicle battery state, power distribution network operation parameters and user charging demand information;

[0064] The vehicle network interaction strategy optimization module 102: based on real-time vehicle network state and user demand prediction, a multi-agent collaborative optimization model is constructed to generate vehicle network interaction strategies through multi-agent reinforcement learning algorithm. This module breaks through the limitations of traditional single-objective optimization, realizes the multi-objective balance of user charging efficiency, power grid peak shaving contribution and operator revenue under the premise of meeting the safety constraints of the power distribution network;

[0065] The power dynamic allocation module 103: for the concurrent super-charging demand of large-scale ultra-fast charging pile group, based on vehicle network interaction strategy and real-time capacity constraints of power distribution network, power intelligent allocation is realized through dynamic priority scheduling algorithm, through the power pool of shared transformers and energy storage systems, it supports multiple piles to run at full power at the same time, solves the contradiction between transformer capacity limitation and user super-charging demand;

[0066] The energy storage coordination and regulation module 104: links the lithium iron phosphate battery energy storage and the capacitor energy storage, constructs a double-layer energy storage system with long-term energy storage and short-term buffering, and optimizes the charging and discharging plan through a model predictive control algorithm. The module releases energy storage power to support overcharging during the peak load of the power grid, avoids the investment in capacity expansion of the power grid, uses valley electricity for charging during the load valley to reduce the electricity purchase cost of the operator, and quickly responds through the super capacitor to suppress the impact of power fluctuation on the power grid during a sudden power impact;

[0067] The real-time response and feedback module 105: realizes the rapid response of the power grid instruction through the edge computing node, and forms a closed-loop control combined with the terminal sensing data;

[0068] The safety and protocol adaptation module 106: constructs a three-layer safety protection system of data-communication-device to ensure the data privacy and system safety in the process of vehicle-grid interaction; at the same time, a multi-protocol conversion interface is built in to support seamless docking with power grid platforms, vehicle enterprise systems and government supervision platforms at all levels, meet the cross-subject collaboration and policy compliance requirements.

[0069] The multi-source data acquisition and vehicle-grid state sensing module 101 includes a multi-source heterogeneous data acquisition sub-module and a vehicle-grid state fusion sensing sub-module;

[0070] The multi-source heterogeneous data acquisition sub-module acquires real-time charging power, voltage and current through the Hall sensor deployed on the direct current side of the charging pile; acquires the battery SOC, SOH, real-time temperature and charging and discharging cutoff voltage through the OBD-II interface and the electric vehicle BMS system communication; acquires the side voltage, current, power factor and frequency through the power distribution network FTU; at the same time, the charging reservation time, target SOC and payment mode submitted by the user through the APP are accessed, and CAN bus, 5G, LoRa and other multi-protocol parallel transmission are supported;

[0071] The vehicle-grid state fusion sensing sub-module adopts a weighted Bayesian fusion algorithm for data calibration and fusion in view of the time lag and noise interference of multi-source data. First, the various types of data are preprocessed, and then a dynamic weight distribution formula is used to generate a vehicle-grid state vector. The vehicle-grid state vector is:

[0072]

[0073] wherein, is the credibility weight of the ith data source, is the original data value, is the historical sliding mean value of the data, =3 is the dynamic correction coefficient, and the fused state vector contains the power grid safety margin, the charging demand intensity and the device health degree, which provides accurate state input for subsequent regulation.

[0074] Specifically,

[0075] Data source Original SOC error (%) Post-fusion SOC error (%) Time lag (ms) BMS direct acquisition ±5.2 120-180 Charging pile Hall sensor ±3.8 80-120 Weighted Bayesian fusion ±1.5 50

[0076] Specifically, in the vehicle network state perception, through the end-edge-cloud three-level collection architecture and the weighted Bayesian fusion algorithm, multi-dimensional data is integrated, time delay error is reduced, SOC measurement accuracy is improved, high-fidelity data foundation is provided for regulation and control, and the problem of vehicle network state evaluation distortion is effectively solved.

[0077] The vehicle network interaction strategy optimization module 102 includes a multi-agent game optimization submodule and an intelligent charging and discharging prediction submodule.

[0078] The multi-agent game optimization submodule constructs a non-cooperative game model of the power grid, the user and the operator, and clearly defines the benefit functions and constraint conditions of each party. The power grid party takes the minimization of network loss and frequency stability as the target, and its benefit function is:

[0079]

[0080] wherein, is the target load of the power grid, is the actual load, is the line loss caused by V2G discharging, =0.6 is the peak regulation weight, =0.4 is the network loss weight.

[0081] The user party takes the shortest charging time and the minimum battery loss as the target, and the benefit function is:

[0082]

[0083] wherein, =0.5 is the SOC target weight, =0.3 is the time weight, =0.2 is the battery loss weight, is the percentage of the current remaining battery capacity to the total capacity, is the percentage of the battery capacity reached after the user expects to complete charging, is the cumulative charging time from the start of charging to the current time, is the loss amount of the battery health state, which is estimated by a cycle number model.

[0084] The operator party takes the maximization of operation benefit as the target, and the benefit function is:

[0085]

[0086] wherein, =0.7 is the power sales benefit weight, =0.3 is the V2G subsidy weight, is a charging power, is a charging price, is a power purchase power from the power grid, is a power purchase price, is a V2G discharge power, is a V2G subsidy standard.

[0087] The three are balanced through Nash equilibrium, and meet:

[0088]

[0089] The intelligent charging and discharging prediction sub-module adopts an LSTM algorithm to predict future charging load and V2G available capacity. Model input features include historical charging and discharging data, user travel regularity, meteorological data and date type. The prediction accuracy is evaluated by the average relative error:

[0090]

[0091] Wherein, T is the number of prediction periods, is the actual charging and discharging power at time t, is the predicted power.

[0092] Specifically, the multi-objective collaborative optimization mechanism of the three-party game in the application solves the three-party benefit balance point through Nash equilibrium, realizes the collaborative improvement of user charging efficiency and operator revenue while ensuring the stability of the power grid frequency, and solves the problem of multi-party interest conflict.

[0093] Specifically,

[0094] Subject Core evaluation index Pre-optimization level Post-optimization level (Nash equilibrium point) Improvement range Power grid side Daily average network loss rate (%) 5.2~5.8 3.1~3.5 -38%~40% Daily peak shaving compliance rate (%) 68~75 92~96 +35%~28% User side Average charging duration (min / 100 km) 45~55 30~35 -33%~36% Monthly average battery SOH loss rate (%) 0.8~1 0.3~0.4 -62%~60% Average cost of single charging (yuan) 35~40 28~32 -20%~20% Operator side Daily average income per pile (yuan) 1200~1500 1800~2100 +50%~40% Monthly average V2G subsidy income (ten thousand yuan) 8~12 25~30 +212%~150%

[0095] The power dynamic allocation module 103 further includes a priority evaluation sub-module and a dynamic power allocation sub-module.

[0096] The priority evaluation sub-module constructs a three-dimensional priority evaluation system, and dynamically adjusts the priority of user charging urgency, battery health status and power grid load rate. The user urgency is calculated by the difference between the reservation time and the current time; the battery health status is obtained by BMS data, and the battery with SOH≤80% reduces the priority to reduce the cycle loss; when the power grid load rate exceeds 0.8, the priority of the power grid friendly user is improved. The priority calculation formula is:

[0097]

[0098] Wherein, is the user urgency, the user charging within 10 minutes is 0.9, the user charging after 1 hour is 0.3, battery state of health, grid load rate.

[0099] The dynamic power distribution submodule distributes the total available power of the grid and the energy storage system based on the priority results obtained by the priority evaluation submodule using a proportional fair algorithm. According to the priority proportion of each charging pile, the total available power is distributed to each charging pile. In the distribution process, it is strictly ensured that the allocated power of each charging pile does not exceed its own demand power, and also does not exceed the total available power of the system, so as to realize the reasonable distribution of power. When the system detects that the grid load rate is overloaded, the power reduction mechanism is triggered immediately, and the charging piles with lower priority are preferentially reduced in power. The reduction range is determined according to the extent of the overload of the grid load rate.

[0100] Specifically,

[0101] Index Traditional mode Dynamic allocation mode Improvement range Number of piles with simultaneous full power 6 10 67％ Transformer load rate 93.75％ 85.3％ -8.4％ User waiting time 15 min 0 min 100％

[0102] Specifically, the dynamic allocation strategy of the power pool sharing in the application combines the priority evaluation system and the proportional fair algorithm, supports the simultaneous operation of multiple super charging piles, greatly improves the utilization rate of the charging piles, and solves the problems of resource waste and user waiting caused by single pile monopoly.

[0103] The energy storage cooperative control module 104 further comprises an energy storage state monitoring submodule and an energy storage and energy storage and charging cooperative optimization submodule;

[0104] The energy storage state monitoring submodule collects key parameters of the energy storage system in real time, including lithium battery SOC, charging and discharging power, cycle number and temperature; voltage, instantaneous power and health status of super capacitor. The available capacity formula of the energy storage is calculated as:

[0105]

[0106] Among them = 20%, which is a protection threshold to ensure that the charging and discharging depth of the energy storage system does not exceed 80%, refers to the actual available capacity of the energy storage system, refers to the rated capacity of the energy storage system.

[0107] The energy storage and charging cooperative optimization submodule uses the MPC algorithm to perform rolling optimization on the energy storage charging and discharging plan in the future period. In the optimization process, the target is to make the grid interaction power as close as possible to the target flattening power and the energy storage power change rate as small as possible. Through such an optimization mode, the stable operation of the grid is ensured, the influence caused by power fluctuation is reduced, the loss of the energy storage equipment caused by frequent charging and discharging is reduced, and the service life is prolonged.

[0108] Specifically, the double-layer architecture of the lithium iron phosphate battery and the super capacitor in the application optimizes the charging and discharging plan through the MPC algorithm, releases the energy storage at the peak to reduce the capacity expansion cost of the power grid, charges at the trough to reduce the power purchase cost, and can quickly respond to sudden power impact, thereby increasing the suppression amplitude of the power grid fluctuation, and balancing the capacity and response speed.

[0109] The real-time response and feedback module 105 further comprises a power grid instruction analysis submodule and a closed-loop correction submodule.

[0110] The power grid instruction analysis submodule receives power grid control instructions, including frequency modulation signals, reactive power compensation requirements and emergency load shedding commands, through a power distribution network dispatching interface, and converts the frequency modulation signals into power regulation quantities.

[0111]

[0112] wherein, = 100kW / Hz is the frequency response coefficient, is the frequency modulation signal.

[0113] The closed-loop correction submodule compares the actual response power with the target value of the power grid instruction, calculates the deviation between the two, and records the reasons for the deviation in detail, and optimizes and adjusts the strategy in the next control period according to the reasons.

[0114] Specifically, the edge-driven closed-loop response technology realizes fast response to power grid instructions based on edge computing nodes, greatly improves the speed compared with traditional cloud control, combines the frequency response formula and the closed-loop correction mechanism, meets the requirements of power grid primary frequency modulation, and improves the control accuracy.

[0115] Working principle and use process of the application: when the application works, the multi-source data acquisition and vehicle network state perception module 101 acquires and fuses multiple types of data, and the vehicle network interaction strategy optimization module 102 generates a strategy accordingly. The power dynamic distribution module 103 distributes power according to the strategy, and the energy storage collaborative control module 104 optimizes the charging and discharging of the energy storage. The real-time response and feedback module 105 quickly responds to the power grid instruction and corrects the strategy, and the safety and protocol adaptation module 106 guarantees safety and docking. The use process is: after the terminal is started, data is automatically acquired and fused, control strategies are generated and executed, instructions are responded to and corrected in real time, and safety and cross-platform adaptation are guaranteed throughout the process.

[0116] It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting; it is not intended to exclude myriad other embodiments of the present application that other inventors can develop based on the same general inventive concepts embodied by the described embodiments. That is, although the present application is described in terms of particular embodiments and illustrative figures, it should be apparent that the scope of the present application is not limited to these specific embodiments.

[0117] While the embodiments of the application have been shown and described herein, it is to be understood that the scope of the application, jointly pointed out in the appended claims, is not to be limited to the above-described embodiments but can be otherwise variously changed, modified, replaced, and altered within the principles and spirit of the present application.

Claims

1. A self-adaptive regulation system for large-scale ultra-fast charging facility vehicle network interaction, characterized in that, The system comprises a multi-source data acquisition and vehicle network state perception module (101), a vehicle network interaction strategy optimization module (102), a power dynamic allocation module (103), a energy storage collaborative control module (104), a real-time response and feedback module (105), and a safety and protocol adaptation module (106), wherein: The multi-source data acquisition and vehicle network state perception module (101) comprises a multi-source heterogeneous data acquisition sub-module and a vehicle network state fusion perception sub-module. The multi-source heterogeneous data acquisition sub-module acquires real-time charging power, voltage, and current through Hall sensors deployed on the direct current side of the charging pile; acquires battery SOC, SOH, real-time temperature, and charge / discharge cutoff voltage through OBD-II interface communication with the electric vehicle BMS system; acquires side voltage, current, power factor, and frequency through the power distribution network FTU; and simultaneously acquires charging reservation time, target SOC, and payment mode submitted by users through the APP, supporting multi-protocol parallel transmission such as CAN bus, 5G, and LoRa. The vehicle network interaction strategy optimization module (102) comprises a multi-agent reinforcement learning algorithm based on real-time vehicle network state and user demand prediction to generate vehicle network interaction strategies. The power dynamic allocation module (103) comprises a dynamic priority scheduling algorithm based on vehicle network interaction strategies and real-time capacity constraints of the power distribution network to achieve intelligent power allocation, support multiple piles operating at full power at the same time through the power pool of shared transformers and energy storage systems, and solve the contradiction between transformer capacity limitations and user supercharging needs. The energy storage collaborative control module (104) comprises a long-term energy storage system with a short-term buffer system, and optimizes charge / discharge plans through model predictive control algorithms. The real-time response and feedback module (105) comprises an edge computing node to achieve fast response to grid commands and form a closed-loop control with terminal sensing data. 2.The adaptive control system for large-scale ultra-fast charging facility vehicle network interaction according to claim 1, wherein, The safety and protocol adaptation module (106) comprises a data-communication-device three-layer security protection system to ensure data privacy and system security during vehicle network interaction, and a multi-protocol conversion interface to support seamless connection with power grid platforms, vehicle enterprise systems, and government regulatory platforms, meeting cross-subject collaboration and policy compliance requirements. The multi-source data acquisition and vehicle network state perception module (101) comprises a multi-source heterogeneous data acquisition sub-module and a vehicle network state fusion perception sub-module. The multi-source heterogeneous data acquisition sub-module acquires real-time charging power, voltage, and current through Hall sensors deployed on the direct current side of the charging pile; acquires battery SOC, SOH, real-time temperature, and charge / discharge cutoff voltage through OBD-II interface communication with the electric vehicle BMS system; acquires side voltage, current, power factor, and frequency through the power distribution network FTU; and simultaneously acquires charging reservation time, target SOC, and payment mode submitted by users through the APP, supporting multi-protocol parallel transmission such as CAN bus, 5G, and LoRa. The vehicle-network state fusion perception sub-module adopts a weighted Bayesian fusion algorithm for data calibration and fusion in view of time lag and noise interference of multi-source data. First, various types of data are preprocessed, and then a vehicle-network state vector is generated through a dynamic weight distribution formula. The vehicle-network state vector is: ， wherein, is the credibility weight of the ith data source, is the original data value, is the historical moving average of the data, = 3 is the dynamic correction coefficient, and the fused state vector contains the power grid safety margin, charging demand intensity, and equipment health degree, providing accurate state input for subsequent regulation and control. 3.The adaptive control system for large-scale ultra-fast charging facility vehicle network interaction according to claim 1, wherein, The vehicle-network interaction strategy optimization module (102) comprises a multi-agent game optimization sub-module and an intelligent charging and discharging prediction sub-module. The multi-agent game optimization sub-module constructs a non-cooperative game model among the power grid, users and operators, and clearly defines the benefit functions and constraint conditions of each party. The power grid party takes the minimization of network loss and frequency stability as the target, and its benefit function is: ， wherein, is the grid target load, is the actual load, is the line loss due to V2G discharging, = 0.6 is the peak shaving weight, = 0.4 is the line loss weight; The user party takes the minimization of charging time and battery loss as the target, and the benefit function is: ， wherein, = 0.5, is a SOC target weight, = 0.3, is a time weight, = 0.2, is a battery loss weight, is a current state of charge percentage, is a user desired state of charge percentage, is a cumulative charging time from the start of charging to the current time, is a battery state of health loss, estimated by a cycle number model; The operator party takes the maximization of operation benefit as the target, and the benefit function is: ， wherein, = 0.7 is the weight of the electricity sales revenue, = 0.3 is the weight of the V2G subsidy, is the charging power, is the charging tariff, is the power purchased from the grid, is the purchase tariff, is the V2G discharging power, is the V2G subsidy standard; The three parties reach a balance point of benefits through Nash equilibrium, satisfying: ， The intelligent charging and discharging prediction sub-module predicts future charging load and V2G available capacity using the LSTM algorithm. The model input features include historical charging and discharging data, user travel habits, weather data and date type. The prediction accuracy is evaluated by the average relative error: ， Wherein, T is the number of prediction periods, denotes the actual charge and discharge power at time t, is the predicted power. 4.The adaptive control system for large-scale ultra-fast charging facility vehicle network interaction according to claim 1, wherein, The power dynamic allocation module (103) further comprises a priority evaluation sub-module and a dynamic power allocation sub-module. The priority evaluation sub-module constructs a three-dimensional priority evaluation system to dynamically adjust the priority of users according to the charging urgency, battery health status and grid load rate. The user urgency is calculated by the difference between the reservation time and the current time; the battery health status is obtained through BMS data, and the priority of the battery with SOH ≤ 80% is reduced to reduce the cycle loss; when the grid load rate exceeds 0.8, the priority of the grid-friendly user is increased. The priority calculation formula is: ， wherein, is the user urgency, 10 minutes before the appointment of the user is 0.9, 1 hour after the user is charged 0.3, is the battery health state, is the grid load rate; The dynamic power allocation sub-module allocates the total available power of the grid and the energy storage system based on the priority results obtained by the priority evaluation sub-module using the proportional fairness algorithm. According to the priority proportion of each charging pile, the total available power is allocated to each charging pile. In the allocation process, it will strictly ensure that the allocated power of each charging pile does not exceed its own demand power, and also does not exceed the total available power of the system, so as to realize the reasonable allocation of power. When the system detects that the grid load rate is overloaded, the power reduction mechanism will be triggered immediately, and the charging piles with lower priority will be preferentially reduced in power. The reduction range is determined according to the extent of the overload of the grid load rate.

5. The adaptive control system for vehicle-to-grid interaction of large-scale ultra-fast charging facilities according to claim 1, characterized in that, The energy storage collaborative regulation module (104) further comprises an energy storage state monitoring sub-module and an energy storage and energy storage and charging collaborative optimization sub-module. The energy storage state monitoring sub-module collects key parameters of the energy storage system in real time, including lithium battery SOC, charging and discharging power, cycle number and temperature; voltage, instantaneous power and health status of supercapacitor. The available capacity of energy storage is calculated as: ， wherein = 20%, is a protection threshold, ensuring that the energy storage system charge and discharge depth does not exceed 80%, refers to the current available capacity of the energy storage system, refers to the rated capacity of the energy storage system; The storage and charging collaborative optimization submodule adopts an MPC algorithm to perform rolling optimization on the energy storage charging and discharging plan of a future period. In the optimization process, the target is to make the grid interaction power as close as possible to the target power and the energy storage power change rate as small as possible. Through such an optimization mode, the stable operation of the grid is ensured, the impact caused by power fluctuation is reduced, the loss of the energy storage equipment caused by frequent charging and discharging is reduced, and the service life of the energy storage equipment is prolonged.

6. The adaptive control system for vehicle-to-grid interaction of large-scale ultra-fast charging facilities according to claim 1, characterized in that, The real-time response and feedback module (105) further comprises a grid instruction analysis submodule and a closed-loop correction submodule. The grid instruction analysis submodule receives grid control instructions, including frequency modulation signals, reactive power compensation requirements and emergency load shedding commands, through a power distribution grid dispatching interface, and converts the frequency modulation signals into power regulation quantities. ， wherein, = 100 kW / Hz is the frequency response coefficient, is the frequency modulated signal; The closed-loop correction submodule compares the actual response power with the grid instruction target value, calculates the deviation between the two, and records the reasons for the deviation in detail. In the next control period, the strategy is optimized and adjusted in view of these reasons.

Images