CN-121999608-A - Multi-intelligent vehicle coupling optimal control method and system based on V2X communication

CN121999608ACN 121999608 ACN121999608 ACN 121999608ACN-121999608-A

Abstract

The application relates to the field of intelligent traffic systems, discloses a multi-intelligent vehicle coupling optimization control method and system based on V2X communication, and aims to solve the problems of information island, local optimization and global target mismatch, response lag, control conflict and the like in multi-vehicle coordination in the prior art. The method comprises the steps of receiving dynamic information of adjacent vehicles and road sides through V2X, fusing high-precision state data of the vehicles, constructing a multi-agent state coupling diagram containing space-time constraints, generating local context embedded vectors through graphic neural network coding, inputting a distributed model prediction controller to solve a quadratic programming problem with hard constraints, enabling objective functions to integrate track tracking, control smoothness and dynamic safety distance, introducing consistency constraints to ensure unified multi-vehicle cooperative logic, and finally outputting a first-step control instruction to drive a linear control executing mechanism. The application realizes safe, efficient and energy-saving centerless multi-vehicle cooperative control under the high-density dynamic traffic scene.

Inventors

LI KENING
SONG ZHUMEI
ZHOU SHIQIONG
XIE XINGXIANG
GUAN MINGXIANG

Assignees

深圳信息职业技术大学

Dates

Publication Date: 20260508
Application Date: 20260126

Claims (10)

1. A V2X communication-based multi-intelligent vehicle coupling optimization control method is characterized by comprising the following steps: Receiving dynamic traffic information from adjacent intelligent vehicles and road side units in real time through a vehicle-mounted V2X communication unit; Synchronously acquiring high-precision motion state data of the vehicle; Constructing a multi-agent state coupling diagram containing space-time constraint based on the dynamic traffic information and the high-precision motion state data of the vehicle, wherein nodes of the multi-agent state coupling diagram represent intelligent vehicles, and the weight of edges is determined based on the relative distance, the relative speed, the lane relation and the communication link quality between the vehicles; Performing graph neural network coding processing on the multi-agent state coupling graph to generate a local context embedded vector of each node; Inputting the local context embedded vector to a distributed model prediction controller, wherein the distributed model prediction controller is used for solving the optimization problem with hard constraint on line based on a vehicle kinematic model, a preset objective function and multiple constraint conditions including dynamic safety distance maintenance constraint, consistency constraint and vehicle executor physical constraint to obtain an optimal control instruction sequence of the vehicle; And outputting and executing the first-step control quantity in the optimal control instruction sequence to drive a drive-by-wire executing mechanism of the vehicle, and entering the next control period based on the updated vehicle state and environment information.
2. The V2X communication-based multi-agent state coupling optimization control method according to claim 1, wherein constructing the multi-agent state coupling map including space-time constraints specifically comprises: Selecting neighborhood vehicle nodes according to a preset communication radius by taking the vehicle as a center; calculating the relative distance, the relative speed, the lane offset and the communication signal-to-noise ratio between each neighborhood vehicle node and the vehicle; Based on the calculated relative distance, relative speed, lane offset and communication signal to noise ratio, calculating the weight of the edge pointing to each neighborhood vehicle node from the vehicle node in a weighted fusion mode; based on the received time stamp of the state information of the neighborhood vehicle, calculating communication delay, and performing time delay compensation on the state information of the neighborhood vehicle by adopting a state extrapolation method to obtain corrected state estimation at the current moment.
3. The V2X communication-based multi-agent vehicle coupling optimization control method according to claim 1, wherein performing graph neural network encoding processing on the multi-agent state coupling graph specifically comprises: Constructing an initial feature vector containing corrected state data for each node in the multi-agent state coupling diagram; Processing the initial feature vector through a multi-layer graph convolution neural network, wherein each layer of graph convolution passes through the features of aggregation nodes and neighborhood nodes thereof, and updates the feature representation of the nodes through nonlinear transformation; And taking the characteristic vector output by the neural network of the convolution of the last layer of graph as the local context embedding vector of the node.
4. The V2X communication-based multi-intelligent vehicle coupling optimization control method according to claim 1, wherein the objective function is constructed specifically as follows: The track tracking error item is used for punishing the position and speed deviation between the predicted track and the expected reference track of the host vehicle; the control smoothness cost item is used for punishing the size and the change rate of the control instruction; The dynamic safety distance punishment item is used for punishing the situation that the predicted distance between the vehicle and the neighborhood vehicle in the prediction time domain is smaller than the safety distance calculated dynamically aiming at the neighborhood vehicle judged to be in potential conflict.
5. The V2X communication-based multi-intelligent vehicle coupling optimization control method according to claim 4, wherein the dynamic safety distance is dynamically calculated according to the current speed of the vehicle and the collision vehicle, a preset total reaction time of the system and an adhesion coefficient of the current road surface.
6. The V2X communication-based multi-intelligent vehicle coupling optimization control method according to claim 1, wherein the consistency constraint is used for coordinating cooperative logic between multiple vehicles in a specific traffic scene; For the intersection passing scene, the consistency constraint is based on the passing sequence of the road side unit arbitration, and is converted into a constraint condition for limiting the time sequence of the vehicle passing through the conflict point; For a formation driving scene, the consistency constraint requires that the position error of the following vehicle relative to the formation datum point is kept within a preset tolerance range; For an emergency braking scenario, the compliance constraint requires that the following vehicle meet a cooperative requirement in response time and deceleration value after receiving an emergency braking signal from the preceding vehicle.
7. The V2X communication-based multi-intelligent vehicle coupling optimization control method according to claim 6, wherein the distributed model predictive controller iteratively solves and coordinates the distributed optimization problem coupled with the consistency constraint by using an alternate direction multiplier method until the common variable of each related vehicle converges.
8. The V2X communication-based multi-intelligent vehicle coupling optimization control method according to claim 2, further comprising: When the dynamic traffic information is received, checking the received data packet; and counting the communication packet loss rate based on the verification result, and dynamically adjusting the trust weight of the V2X information from the corresponding remote vehicle according to the condition that the communication packet loss rate exceeds a preset threshold.
9. The V2X communication-based multi-intelligent vehicle coupling optimization control method according to claim 1, wherein outputting and executing the first-step control amount in the optimal control instruction sequence specifically comprises: converting the longitudinal acceleration instruction in the first-step control amount into a driving motor torque instruction or an electronic braking pressure instruction, and sending the driving motor torque instruction or the electronic braking pressure instruction to a corresponding driving or braking controller; And directly sending the front wheel steering angle instruction in the first-step control quantity to a controller of the drive-by-wire steering executing mechanism.
10. A V2X communication-based multi-intelligent vehicle coupling optimization control system applied to the V2X communication-based multi-intelligent vehicle coupling optimization control method as set forth in claims 1 to 9, comprising: The V2X communication data receiving module is used for receiving dynamic traffic information from the adjacent intelligent vehicles and the road side units in real time through the vehicle-mounted V2X communication unit; the host vehicle state sensing module is used for acquiring high-precision motion state data of the host vehicle; The multi-agent state coupling diagram construction module is used for constructing a multi-agent state coupling diagram containing space-time constraint based on the dynamic traffic information and the high-precision motion state data of the vehicle, wherein nodes of the multi-agent state coupling diagram represent intelligent vehicles, and the weight of edges is determined based on the relative distance, the relative speed, the lane relation and the communication link quality between the vehicles; The graph neural network coding module is used for performing graph neural network coding processing on the multi-agent state coupling graph to generate local context embedded vectors of each node; The distributed model prediction control module is used for inputting the local context embedded vector to a distributed model prediction controller, wherein the distributed model prediction controller is used for solving the optimization problem with hard constraint on line based on a vehicle kinematic model, a preset objective function and multiple constraint conditions including dynamic safety distance maintenance constraint, consistency constraint and vehicle executor physical constraint and outputting an optimal control instruction sequence of the vehicle; And the vehicle execution control module is used for intercepting the first step control quantity in the optimal control instruction sequence and transmitting the first step control quantity to a wire control driving, wire control steering and wire control actuating mechanism of the vehicle.

Description

Multi-intelligent vehicle coupling optimal control method and system based on V2X communication Technical Field The invention belongs to the field of intelligent traffic systems, and particularly relates to a multi-intelligent vehicle coupling optimization control method and system based on V2X communication. Background With the development of intelligent networking technology, the cooperative control of multiple vehicles based on V2X communication becomes a key for improving traffic efficiency and safety. V2X technology enables vehicles to exchange information with surrounding vehicles, infrastructure and networks, providing the potential for collaborative decisions beyond single vehicle perception. In complex dynamic scenes such as intersections, high-speed formation and the like, multiple vehicle movements are tightly coupled, and a control method capable of unifying global coordination and local optimization is needed. The existing multi-vehicle cooperative scheme mainly has two types of limitations that a centralized method depends on a central node and is challenged in scale expansibility, communication reliability and real-time performance, and a common distributed method generally carries out independent decision based on local simple rules, and dynamic coupling relation among vehicles is not fully considered, so that suboptimal system performance, control conflict or vehicle team instability are easily caused. In addition, the existing method often fails to fully fuse heterogeneous environment information in V2X, lacks an effective processing mechanism for uncertainty such as actual communication delay and packet loss, and has insufficient robustness in an actual high-dynamic traffic scene. Therefore, a distributed cooperative control scheme which can deeply fuse multi-source information and explicitly model dynamic coupling relations among vehicles and has good expandability and communication robustness is needed in the field so as to realize safe, efficient and energy-saving multi-vehicle cooperative operation under centerless scheduling. Disclosure of Invention The invention provides a V2X communication-based multi-intelligent vehicle coupling optimization control method and system, and aims to solve the structural technical problems of information island, local optimization and global target mismatch, dynamic environment response lag, control instruction conflict and the like in the multi-vehicle cooperative control process in the existing intelligent traffic system. The prior art generally relies on a bicycle intelligent sensing and decision mechanism, even if a bicycle-road cooperation or bicycle-bicycle communication is introduced, the data interaction is still mainly broadcasted in a static state, and the unified modeling capability of a dynamic feedback mechanism between a multi-bicycle dynamics coupling relation, a macroscopic evolution trend of traffic flow and microscopic individual behaviors is lacking, so that the safe, efficient and energy-saving global cooperative control is difficult to realize under a high-density and high-dynamic traffic scene. As one implementation mode of the invention, the multi-intelligent vehicle coupling optimization control method based on V2X communication comprises the following steps: Receiving dynamic traffic information from adjacent intelligent vehicles and road side units in real time through a vehicle-mounted V2X communication unit; Synchronously acquiring high-precision motion state data of the vehicle; Constructing a multi-agent state coupling diagram containing space-time constraint based on the dynamic traffic information and the high-precision motion state data of the vehicle, wherein nodes of the multi-agent state coupling diagram represent intelligent vehicles, and the weight of edges is determined based on the relative distance, the relative speed, the lane relation and the communication link quality between the vehicles; Performing graph neural network coding processing on the multi-agent state coupling graph to generate a local context embedded vector of each node; Inputting the local context embedded vector to a distributed model prediction controller, wherein the distributed model prediction controller is used for solving the optimization problem with hard constraint on line based on a vehicle kinematic model, a preset objective function and multiple constraint conditions including dynamic safety distance maintenance constraint, consistency constraint and vehicle executor physical constraint to obtain an optimal control instruction sequence of the vehicle; And outputting and executing the first-step control quantity in the optimal control instruction sequence to drive a drive-by-wire executing mechanism of the vehicle, and entering the next control period based on the updated vehicle state and environment information. Further, constructing a multi-agent state coupling diagram including space-time constraints specifically includes: Selec