Cyber-physical system based cooperative lane-changing decision method for intelligent vehicles

Abstract

The application relates to an intelligent automobile cooperative lane-changing decision method and system based on an information physical system, and belongs to the technical field of intelligent traffic and automatic driving control. The method collects lane-changing vehicle, surrounding vehicle, lane traffic flow and communication topology data, and constructs a CPS closed-loop model. Physical states of the vehicle are mapped into information space digital twin states to generate target lane candidate gaps. A safe feasible region is established according to a forward safety distance, a backward safety distance and a predicted acceleration after lane changing. A cooperative benefit function is established according to vehicle benefits, neighboring vehicle disturbance, lane load balancing, risk punishment and control smoothing cost. A double-layer planning model is formed by upper-layer road section traffic efficiency optimization and lower-layer single-vehicle travel cost optimization, and lane flow adjustment coefficients and single-vehicle lane-changing actions are updated by using execution feedback rolling. The application can improve traffic efficiency, reduce lane-changing disturbance, reduce sudden deceleration and improve traffic flow stability in a mixed traffic environment.

Description

Technical Field

[0001] This invention belongs to the technical fields of intelligent transportation, autonomous driving control, and cyber-physical systems, and relates to an intelligent vehicle cooperative lane-changing decision-making method based on cyber-physical systems. Background Technology In urban expressways, highways, and multi-lane arterial roads, lane-changing behavior directly impacts vehicle safety, traffic flow stability, and road efficiency. In the mixed traffic phase where human-driven vehicles and intelligent vehicles coexist, lane-changing vehicles not only need to determine if there is a gap to insert into the target lane, but also need to evaluate the continuous impact of the lane-changing action on vehicles following in the target lane, vehicles following in the original lane, and the overall traffic flow of the road segment. If lane-changing decisions are made solely based on local gaps for a single vehicle or fixed thresholds, problems such as misjudgment of target gaps, sudden deceleration of following vehicles, frequent lane changes, and local optima can easily occur.

[0002] Existing lane-changing models typically handle safety judgments, benefit evaluations, and traffic flow optimization in a decentralized manner, lacking continuous feedback between physical and information spaces. Although some models introduce MOBIL, game theory, or learning algorithms, there is a lack of clear coupling between the efficiency objectives of upper-level road segments and the lane-changing actions of individual vehicles in the lower level. This makes it difficult for the models to explain how the global traffic state constrains lane-changing and to correct the generated lane-changing decisions in a timely manner based on execution errors.

[0003] Therefore, how to unify vehicle local lane-changing safety, surrounding vehicle disturbance, overall road segment efficiency, and vehicle execution feedback into a single computable framework in a mixed traffic environment, so that collaborative lane-changing decisions can not only meet safety and feasibility requirements but also generate stable benefits at the traffic flow level, is an urgent problem to be solved in this field. Summary of the Invention

[0004] In view of this, the purpose of this invention is to address the problems that easily arise in mixed traffic environments when lane-changing vehicles make decisions based solely on local gaps or individual vehicle benefits, leading to sudden deceleration of following vehicles in the target lane, lane load imbalance, frequent lane changes, and amplified traffic flow disturbances. This invention provides an intelligent vehicle cooperative lane-changing decision-making method based on a cyber-physical system. This invention integrates lane-changing vehicles, neighboring vehicles, lane traffic flow, communication topology, and vehicle execution feedback into a unified cyber-physical system closed loop. Through state mapping from physical space to information space, cooperative optimization decision-making within the information space, and feedback correction after control execution, it achieves safe, feasible, explainable, and globally coordinate intelligent vehicle cooperative lane-changing decisions.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A cooperative lane-changing decision-making method for intelligent vehicles based on cyber-physical systems includes the following steps: S1: Collect data on lane-changing vehicles, surrounding vehicles, road lanes, traffic flow status, and vehicle communication topology, and preprocess the data to obtain fused status data for collaborative lane-changing decisions. S2: Construct a CPS closed-loop model that includes a physical traffic layer, a perception and communication layer, an information mapping layer, a collaborative decision-making layer, and a control execution layer, and map the vehicle operation status, lane traffic status, and communication topology status in the physical space to a digital twin status in the information space. S3: Based on the digital twin state, identify the candidate target lane and candidate target gap of the lane-changing vehicle, generate a set of candidate lane-changing actions, and predict the motion response of the lane-changing vehicle, the vehicle in front of the target lane, the vehicle behind the target lane, and the adjacent vehicles within the communication range after each candidate lane-changing action is executed. S4: Based on the actual forward distance between the lane-changing vehicle and the vehicle in front of the target lane, the actual backward distance between the lane-changing vehicle and the vehicle behind the target lane, the minimum forward safe distance, the minimum backward safe distance, the predicted acceleration of the lane-changing vehicle, and the predicted acceleration of the vehicle behind the target lane, a safety constraint vector for the candidate lane-changing action is constructed, and the safety constraint vector being less than or equal to zero is used as the criterion for determining the safe feasible region. S5: Within the safe and feasible domain, a cooperative lane-changing benefit function is constructed based on the lane-changing vehicle passage benefit, the acceleration disturbance cost of adjacent vehicles within the communication range, the improvement of lane flow variance before and after lane changing, the target gap risk penalty, and the trajectory control smoothing cost. Candidate lane-changing actions that meet the benefit threshold are then selected based on the cooperative lane-changing benefit function. S6: Establish an upper-level road segment traffic efficiency optimization model and a lower-level single-vehicle travel cost optimization model. The upper-level road segment traffic efficiency optimization model outputs a lane flow adjustment coefficient. The lower-level single-vehicle travel cost optimization model selects single-vehicle lane-changing actions under the constraints of the lane flow adjustment coefficient, the safe and feasible domain, and the collaborative benefit threshold. The coupling between the road segment efficiency target and the single-vehicle lane-changing actions is achieved through rolling iteration. S7: Generate longitudinal speed planning and lateral lane-changing trajectory based on the single vehicle lane-changing action, and send the longitudinal speed planning and lateral lane-changing trajectory to the vehicle control system; after the vehicle executes, it sends the physical feedback state back to the information space, updates the CPS closed-loop model, and enters the next decision cycle.

[0006] Furthermore, in step S2, the digital twin state in the information space is represented as follows:

[0007] in, For the first The car in Vehicle status during each decision cycle For vehicle assembly, and The first Lane density and traffic flow, For lane collection, This is the adjacency matrix of the vehicle communication topology. This represents the residual state of the model. The first The car in Vehicle status during each decision cycle Represented as:

[0008] in, The longitudinal position of the vehicle's center of gravity. The vehicle's lateral position. For vehicle speed, To accelerate the vehicle, This refers to the vehicle's heading angle; The model residual state The value is determined based on the difference between the physical space feedback state and the information space prediction state, and is expressed as:

[0009] in, This is the physical space feedback state. Predicting the state of information space, For the normalized weight matrix, To prevent positive numbers with a denominator of zero; Vehicle longitudinal state prediction satisfies:

[0010]

[0011] in This represents the sampling time interval.

[0012] Furthermore, in step S3, the first The set of candidate lane-changing actions for a smart car is represented as:

[0013] in, The vehicle's current lane, For candidate target lanes, The candidate target gap is defined by the ALV of the vehicle ahead in the target lane and the AFV of the vehicle behind in the target lane. For lane change duration, This represents the number of candidate lane-changing maneuvers.

[0014] Furthermore, in step S4, let the lane-changing vehicle be SV, the vehicle in front of the target lane be ALV, and the vehicle behind the target lane be AFV, and let the actual forward distance between the lane-changing vehicle and the vehicle in front of the target lane be... And the actual rearward distance between the vehicle changing lanes and the vehicle behind in the target lane. They are respectively:

[0015]

[0016] in, , , These represent the longitudinal positions of the centroids of the vehicle changing lanes, the vehicle in front of the vehicle in the target lane, and the vehicle behind the vehicle in the target lane, respectively. , and These are the vehicle lengths of the vehicle changing lanes, the vehicle in front of the target lane, and the vehicle behind the target lane, respectively. Forward minimum safe distance and minimum safe distance backward They are defined as follows:

[0017]

[0018] in, For stationary safe distance, For reaction time, , , These are the available decelerations for the corresponding vehicles. ; Candidate lane change action The safety constraint vector is:

[0019] in, and These are the forward and backward safety margins, respectively. Predicted acceleration for vehicles performing candidate actions after changing lanes. The predicted acceleration of the vehicle following in the target lane after being affected by the lane change. For comfortable deceleration threshold, For safe deceleration threshold; when the following conditions are met At that time, the candidate lane change action falls within the safe and feasible domain.

[0020] Furthermore, the cooperative lane-changing benefit function described in step S5 is:

[0021] in, For the benefit of vehicles changing lanes. This represents the improvement in the variance of lane flow before and after a lane change. The cost of acceleration disturbances of nearby vehicles within the communication range. Penalty for target gap risk, The cost of trajectory control smoothing , , , , All are non-negative weighting coefficients; The traffic revenue of lane-changing vehicles The calculation is as follows:

[0022] in:

[0023] in, For the first A number of vehicles performing lane-changing maneuvers The predicted average speed during the lane change duration, For the first The predicted average speed of vehicles changing lanes within the same time range while maintaining their current lane. For the first A number of vehicles performing lane-changing maneuvers Later in time Prediction speed, For the first When a vehicle changing lanes maintains its current lane at a certain time Prediction speed, For the first The maximum permissible speed for vehicles changing lanes. To prevent positive numbers with a denominator of zero; when When, it indicates that a candidate lane change action is being executed. It can increase the average travel speed of vehicles changing lanes during the lane-changing duration; when When this occurs, it indicates that executing the candidate lane-changing action cannot generate positive traffic benefits; The improvement in lane load balancing before and after lane changing The calculation is as follows:

[0024] in, The first lane change action after the candidate lane change action is executed Predicted traffic flow for each lane; The cost of disturbance from nearby vehicles The calculation is as follows:

[0025] in For the communication range of lane-changing vehicles, For nearby vehicles The perturbation weights, Acceleration for adjacent vehicles before they change lanes The predicted acceleration of adjacent vehicles after a candidate lane change is executed; The target gap risk penalty The calculation is as follows:

[0026] in, and This is the risk attenuation coefficient. Weighting of penalties for vehicles behind in the target lane that suddenly decelerate; and To quantify the safety margin of the risk level of candidate lane-changing maneuvers, it is expressed as:

[0027] The trajectory control smoothing cost The calculation is as follows:

[0028] in, and They are longitudinal and lateral jerk, respectively. and These are the vertical and horizontal smoothing weights, respectively.

[0029] Furthermore, the establishment of the upper-level road segment traffic efficiency optimization model and the lower-level single-vehicle travel cost optimization model in step S6 includes: The travel time function for each lane is defined as:

[0030] in, For the first Lane flow regulation coefficient for each lane For the first Free-flow travel time for each lane For lane capacity, and Let be the impedance function parameter, and ; The traffic efficiency optimization model for the upper-level road segment uses the lane flow adjustment coefficient. Let be the decision variable, and its objective function be:

[0031] in, , This is the set of vehicle lane-changing actions output from the lower layer. To execute a set of actions The later Predicted traffic flow for each lane The average predicted traffic flow for each lane. This refers to the lane flow adjustment coefficient from the previous decision-making cycle. , , These are non-negative weighting coefficients; The lower-level bicycle travel cost optimization model uses candidate lane-changing actions. Let be the decision variable, and its objective function be:

[0032] in, For the time cost of a single vehicle, For comfort costs, The global traffic cost on the road segment caused by the candidate action. , , , These are non-negative weighting coefficients; The optimal lane-changing maneuver for a single bicycle is determined using the following formula:

[0033] And it satisfies:

[0034] in, The threshold for synergistic benefits; The upper and lower layers together constitute a two-layer programming model:

[0035]

[0036]

[0037] in A collection of intelligent vehicles participating in collaborative decision-making; The upper-level road segment traffic efficiency optimization model influences the selection of candidate lane-changing actions in the lower-level single-vehicle travel cost optimization model through the lane flow adjustment coefficient. The single-vehicle lane-changing actions output by the lower-level single-vehicle travel cost optimization model change the lane flow, lane density, and vehicle speed distribution, and feed the changes back to the upper-level road segment traffic efficiency optimization model.

[0038] Furthermore, the coupling between the road segment efficiency target and the single-vehicle lane-changing action achieved through rolling iteration in step S6 includes: the partial derivative of the travel time function with respect to the lane flow regulation coefficient is:

[0039] In the In each rolling iteration, the lane flow adjustment coefficient is updated according to the following formula:

[0040] in, For projection operators, To optimize step size, For feedback correction coefficient, For the first The feedback deviation term for each lane is defined as follows:

[0041] in, and Lane flow and density are obtained from feedback in physical space. and Lane flow and density are predicted from the information space. and To prevent positive numbers with a denominator of zero, For density feedback weights; The bilevel programming rolling iteration stops when the following conditions are met:

[0042] in, The convergence threshold of the upper-level objective function. The threshold for lane-changing actions. This is a difference measurement function between candidate lane-changing actions.

[0043] Furthermore, when multiple intelligent vehicles request to switch to the same target gap within the same decision cycle, the candidate lane-changing actions are ranked according to their safety margin, the value of the collaborative lane-changing benefit function, and the available time of the target gap. The candidate lane-changing action with the highest ranking is retained, while the remaining candidate lane-changing actions are canceled or recalculated in the next decision cycle.

[0044] Furthermore, the lateral lane-changing trajectory described in step S7 is represented as follows:

[0045] And satisfy the following boundary conditions:

[0046]

[0047] in, This indicates the current lateral position of the lane centerline. The lateral position of the center line of the target lane.

[0048] On the other hand, the present invention provides an intelligent vehicle cooperative lane-changing decision-making system based on cyber-physical systems, comprising: The acquisition module is used to collect vehicle operating status, road lane status, traffic flow status and vehicle communication topology data, and output fused status data to the CPS mapping module; The CPS mapping module is used to generate a digital twin state in the information space based on the fused state data, and to calculate the model residual state. The candidate gap generation module is used to generate a set of candidate target lanes, candidate target gaps, and candidate lane-changing actions based on the digital twin state. The safety feasibility evaluation module is used to filter candidate lane-changing actions that belong to the safe and feasible domain based on the actual forward spacing, actual backward spacing, minimum forward safety distance, minimum backward safety distance, predicted acceleration of lane-changing vehicles, and predicted acceleration of vehicles following in the target lane. The collaborative benefit evaluation module is used to calculate the collaborative lane-changing benefits of candidate lane-changing actions based on the lane-changing vehicle passage benefits, the acceleration disturbance cost of adjacent vehicles, the lane flow variance improvement, the target gap risk penalty, and the trajectory control smoothing cost. The two-level programming solution module is used to solve the traffic efficiency optimization model of the upper-level road segment and the single-vehicle travel cost optimization model of the lower-level road segment, and outputs the lane flow adjustment coefficient and single-vehicle lane-changing action. The trajectory generation module is used to generate longitudinal speed planning and lateral lane-changing trajectory based on the single vehicle lane-changing action, and output vehicle control commands. The feedback update module is used to correct the digital twin state in the CPS mapping module based on the vehicle execution status, lane flow, lane density and communication status, and trigger the next decision cycle.

[0049] This invention also provides an intelligent vehicle cooperative lane-changing decision-making system based on cyber-physical systems, including a data acquisition module, a CPS mapping module, a candidate gap generation module, a safety feasibility evaluation module, a cooperative benefit evaluation module, a bi-level programming solution module, a trajectory generation module, and a feedback update module. The data acquisition module is used to collect vehicle, road, traffic flow, and communication topology data; the CPS mapping module is used to construct a state mapping between physical space and information space; the candidate gap generation module is used to determine candidate target lanes and candidate target gaps; the safety feasibility evaluation module is used to screen candidate lane-changing actions that satisfy the safety feasible domain constraint; the cooperative benefit evaluation module is used to calculate the cooperative benefit of candidate lane-changing actions; the bi-level programming solution module is used to solve for the lane flow regulation coefficient and single-vehicle lane-changing actions; the trajectory generation module is used to generate lane-changing trajectories; and the feedback update module is used to correct the CPS closed-loop model based on the vehicle execution results.

[0050] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the above-described intelligent vehicle cooperative lane-changing decision-making method based on cyber-physical systems is implemented.

[0051] Compared with the prior art, the present invention has at least the following beneficial effects: First, this invention uses a CPS closed-loop unified description of vehicle physical state acquisition, information space collaborative decision-making, and vehicle control execution feedback, avoiding lane-changing decisions from remaining in a single cycle, localized, and static judgment.

[0052] Second, the present invention constructs a safe and feasible domain by combining forward safe distance, backward safe distance, predicted acceleration of lane-changing vehicles, and predicted acceleration of vehicles following in the target lane. This allows for the explicit elimination of candidate actions that may lead to sudden deceleration of vehicles following in the target lane or insufficient safe distance before generating a lane-changing action.

[0053] Third, this invention constructs a collaborative benefit function that simultaneously includes the vehicle's own travel benefits, adjacent vehicle disturbances, lane load balancing, collision risk, and control smoothness, so that lane-changing decisions no longer only pursue the increase of individual vehicle speed, but also take into account local safety and overall traffic flow stability.

[0054] Fourth, this invention establishes a two-layer coupling relationship through an upper-level road segment traffic efficiency optimization model and a lower-level vehicle travel cost optimization model, so that the upper-level lane flow regulation coefficient can constrain the lane-changing choices of lower-level vehicles, and the execution results of lower-level vehicles can feed back and update the upper-level traffic efficiency target, thereby achieving coordinated optimization of road segment efficiency and vehicle behavior.

[0055] Fifth, this invention uses CPS state residuals and lane feedback bias terms to perform rolling corrections on the model, which can improve the method's adaptability to communication delays, vehicle heterogeneity, execution biases, and traffic flow disturbances, and has good engineering feasibility and scalability.

[0056] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0057] To make the objectives, technical solutions, and advantages of the present invention clearer, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, wherein: Figure 1 This is a diagram of the CPS closed-loop architecture of the present invention; Figure 2 This is a flowchart of the collaborative lane-changing decision-making method of the present invention; Figure 3 This is a schematic diagram illustrating the relationship and safe distance between vehicles in the coordinated lane-changing mechanism according to the present invention. Figure 4 This is a coupling diagram of the two-layer planning model of the present invention; Figure 5 This is a block diagram of the collaborative lane-changing decision-making system of the present invention; Figure 6 The figures show (a) a comparison of average velocity and (b) a comparison of equivalent travel time under different permeability rates according to the present invention. Figure 7 (a) shows the number of vehicles passing through the road segment, and (b) shows a comparison of energy consumption proxy indicators. Figure 8 This is a graph showing the improvement in multiple indicators at a penetration rate of 60% according to the present invention; Figure 9 This is a time-series diagram of the average speed of a road section under 60% penetration rate according to the present invention. Detailed Implementation

[0058] The following specific examples illustrate the implementation of the present invention. 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 be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0059] 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.

[0060] 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.

[0061] Example 1: Please see Figure 1 The intelligent vehicle cooperative lane-changing decision-making process of this invention operates within a CPS closed-loop system. In the physical traffic layer, the lane-changing vehicle (SV), the preceding vehicle (LV) in the original lane, the following vehicle (FV) in the original lane, the preceding vehicle (ALV) in the target lane, and the following vehicle (AFV) in the target lane enter the information space via a P2C link. The information space completes state mapping, candidate gap generation, safety feasibility evaluation, cooperative benefit evaluation, and bi-layer programming solution. Subsequently, the target lane, target gap, and lane-changing trajectory are sent to the vehicle control system via a C2P link. The physical feedback state after vehicle execution is again transmitted back to the information space to update the CPS state for the next decision cycle.

[0062] Please see Figure 2 The method of this invention is executed cyclically in the following order: data acquisition, CPS state update, candidate gap generation, safety evaluation, benefit evaluation, bi-level programming solution, trajectory control, and feedback update. Please refer to [link / reference]. Figure 3 Candidate lane-change maneuvers must simultaneously satisfy the forward safety constraints between SV and ALV, the backward safety constraints between SV and AFV, and the predicted vehicle acceleration constraints after the lane change. Please refer to [link / reference]. Figure 4 The upper-level road segment model and the lower-level vehicle model form a closed-loop coupling through lane flow regulation coefficients, candidate lane-changing actions, and execution feedback. Please refer to [link / reference]. Figure 5 The system of the present invention can be composed of a data acquisition module, a CPS mapping module, a candidate gap generation module, a safety feasibility evaluation module, a collaborative benefit evaluation module, a two-layer programming solution module, a trajectory generation module, and a feedback update module. The above modules can be deployed on an on-board computing unit, a roadside computing unit, or a cloud-edge collaborative computing platform.

[0063] Example 1: This embodiment provides a cooperative lane-changing decision-making method for intelligent vehicles based on cyber-physical systems, including the following steps: S1: Collect data and perform time synchronization, coordinate unification, anomaly removal, and missing data completion on the collected data to obtain fused state data for collaborative lane-changing decisions. In this embodiment, the collected data includes vehicle number, vehicle type, lane number, longitudinal position, lateral position, speed, acceleration, heading angle, vehicle length, target lane preceding vehicle status, target lane following vehicle status, original lane preceding vehicle status, original lane following vehicle status, lane flow rate, lane density, average speed, queue length, communication adjacency relationship, communication latency, and vehicle control execution status.

[0064] S2: Construct a CPS closed-loop model that includes a physical traffic layer, a perception and communication layer, an information mapping layer, a collaborative decision-making layer, and a control execution layer. Map the vehicle operation status, lane traffic status, and communication topology status in the physical space to digital twin status in the information space, and determine the model residual status based on the difference between the feedback status in the physical space and the predicted status in the information space.

[0065] Specifically, let the decision-making cycle be... The sampling time interval is The vehicle group is The collection of intelligent vehicles participating in collaborative decision-making is The lane set is . No. The car in The vehicle state for each decision cycle is represented as follows:

[0066] in, The longitudinal position of the vehicle's center of gravity. The vehicle's lateral position. For vehicle speed, To accelerate the vehicle, This refers to the vehicle's heading angle.

[0067] The state of a CPS digital twin in the information space is represented as follows:

[0068] in, and The first Lane density and traffic flow, This is the adjacency matrix for vehicle communication topology. This represents the normalized residual between the physical feedback state and the information space predicted state. The vehicle longitudinal state prediction satisfies:

[0069]

[0070] S3: Based on the digital twin state, identify candidate target lanes and candidate target gaps for lane-changing vehicles, generate a set of candidate lane-changing actions, and predict the motion response of the lane-changing vehicle, the vehicle in front of the target lane, the vehicle behind the target lane, and neighboring vehicles within the communication range after each candidate lane-changing action is executed. Specifically, the... The set of candidate lane-changing actions for a smart car is represented as:

[0071] in, The vehicle's current lane, For candidate target lanes, The candidate target gap is defined by the ALV of the vehicle ahead in the target lane and the AFV of the vehicle behind in the target lane. For lane change duration, This represents the number of candidate lane-changing maneuvers.

[0072] S4: Based on the actual forward distance between the lane-changing vehicle and the vehicle in front in the target lane, the actual backward distance between the lane-changing vehicle and the vehicle behind in the target lane, the minimum forward safe distance, the minimum backward safe distance, the predicted acceleration of the lane-changing vehicle, and the predicted acceleration of the vehicle behind in the target lane, a safety constraint vector for the candidate lane-changing action is constructed. The condition for determining the safe feasible region is that the safety constraint vector is less than or equal to zero. Specifically, let the lane-changing vehicle be SV, the vehicle in front in the target lane be ALV, and the vehicle behind in the target lane be AFV. Since... This indicates the longitudinal position of the vehicle's center of gravity and the actual forward distance between the vehicle changing lanes and the vehicle in front in the target lane. And the actual rearward distance between the vehicle changing lanes and the vehicle behind in the target lane. They are respectively:

[0073]

[0074] in, , and These are the vehicle lengths of the vehicle changing lanes, the vehicle in front of the target lane, and the vehicle behind the target lane, respectively.

[0075] The minimum forward safe distance and the minimum backward safe distance are defined as follows:

[0076]

[0077] in, For stationary safe distance, For reaction time, , , These are the available decelerations for the corresponding vehicles. .

[0078] Candidate lane change action Its security constraint vector is defined as:

[0079] in, and These are the forward and backward safety margins, respectively. Predicted acceleration for lane-changing vehicles after performing candidate actions. The predicted acceleration of the vehicle following in the target lane after being affected by lane changing. For the comfortable deceleration threshold, The safe deceleration threshold. When the following condition is met:

[0080] At the same time, the candidate lane-changing action simultaneously satisfies the constraints of forward safety, rearward safety, comfort of the lane-changing vehicle, and safety of the vehicle behind in the target lane.

[0081] S5: Within the safe and feasible domain, construct a cooperative lane-changing benefit function based on the lane-changing vehicle passage benefit, the acceleration disturbance cost of nearby vehicles within the communication range, the improvement of lane flow variance before and after lane changing, the target gap risk penalty, and the trajectory control smoothing cost, and then screen candidate lane-changing actions that meet the benefit threshold based on the cooperative lane-changing benefit function.

[0082] Specifically, to quantify the risk level of candidate lane-changing actions, a safety margin is defined:

[0083] The risk penalty function is then:

[0084] in, and This is the risk attenuation coefficient. This is the weight for the penalty of sudden deceleration of the vehicle following in the target lane. This function is non-negative, and the smaller the safety margin between the front and rear vehicles and the greater the deceleration of the vehicle following in the target lane, the greater the risk penalty.

[0085] set up The set of neighboring vehicles within the communication range of the lane-changing vehicle is defined as follows:

[0086] in, For nearby vehicles The perturbation weights, Acceleration for adjacent vehicles before they change lanes This is the predicted acceleration of nearby vehicles after the candidate lane change is executed. This parameter is used to evaluate the impact of the candidate lane change on the operational stability of nearby vehicles within the communication range.

[0087] The improvement in lane load balance before and after lane changing is defined as follows:

[0088] in, The first lane change action after the candidate lane change action is executed Predicted traffic flow for each lane. When When this is the case, it indicates that the lane-changing action helps reduce the degree of uneven traffic flow between lanes.

[0089] The trajectory control smoothing cost is defined as:

[0090] in, and They are longitudinal and lateral jerk, respectively. and These are the vertical and horizontal smoothing weights, respectively.

[0091] Lane-changing vehicle traffic revenue The calculation is as follows:

[0092] in:

[0093] in, For the first A number of vehicles performing lane-changing maneuvers The predicted average speed during the lane change duration, For the first The predicted average speed of vehicles changing lanes within the same time range while maintaining their current lane. For the first A number of vehicles performing lane-changing maneuvers Later in time Prediction speed, For the first When a vehicle changing lanes maintains its current lane at a certain time Prediction speed, For the first The maximum permissible speed for vehicles changing lanes. To prevent positive numbers with a denominator of zero.

[0094] when When, it indicates that a candidate lane change action is being executed. It can increase the average travel speed of vehicles changing lanes during the lane-changing duration; when When this occurs, it indicates that executing the candidate lane-changing action cannot generate positive traffic benefits.

[0095] Based on the above quantities, the collaborative benefit function of the candidate lane-changing action is defined as:

[0096] in, For the benefit of vehicles changing lanes. , , , , All are non-negative weighting coefficients. As can be seen from the following formula, the greater the synergistic benefit, the better the overall effect of the candidate lane-changing action in improving the vehicle's own traffic efficiency, improving lane load balance, reducing disturbances from adjacent vehicles, reducing risks, and improving control smoothness.

[0097] S6: Establish an upper-level road segment traffic efficiency optimization model and a lower-level single-vehicle travel cost optimization model. The upper-level road segment traffic efficiency optimization model uses the lane flow adjustment coefficient as the decision variable, while the lower-level single-vehicle travel cost optimization model uses candidate lane-changing actions as the decision variable. The upper-level model outputs the lane flow adjustment coefficient, and the lower-level model outputs single-vehicle lane-changing actions under the constraints of the lane flow adjustment coefficient, the safe feasible region, and the collaborative benefit threshold. The coupling between the road segment efficiency target and the single-vehicle lane-changing actions is achieved through rolling iteration.

[0098] The traffic efficiency optimization model for upper-level road sections uses lane flow adjustment coefficients. For decision variables. The first... The travel time function for each lane is defined as:

[0099] in, For the first Free-flow travel time for each lane For lane capacity, and Let be the impedance function parameter, and .

[0100] The upper-level objective function is defined as:

[0101] in, , This is the set of vehicle lane-changing actions output from the lower layer. To execute a set of actions The later Predicted traffic flow for each lane The average predicted traffic flow for each lane. This refers to the lane flow adjustment coefficient from the previous decision-making cycle. , , These are non-negative weighting coefficients.

[0102] The lower-level bicycle travel cost optimization model uses candidate lane-changing actions. Let be the decision variable, and its objective function be:

[0103] in, For the time cost of a single vehicle, For comfort costs, The global traffic cost on the road segment caused by the candidate action. , , , These are non-negative weighting coefficients.

[0104] The solution for the lower-level model is as follows:

[0105] And satisfy:

[0106] in, The threshold for collaborative benefits is defined as follows: Therefore, only candidate actions that simultaneously satisfy both safety constraints and collaborative benefit constraints can be considered as executable lane-changing actions.

[0107] The upper and lower layers together constitute a two-layer programming model:

[0108]

[0109]

[0110] Among them, the upper-level model uses lane flow adjustment coefficients. The global traffic cost of each lane is changed, and the lower-level model selects a single-vehicle lane-changing action based on the adjustment coefficient. After the lower-level lane-changing action is executed, the lane flow, lane density and vehicle speed distribution are changed, and the changes are fed back to the upper-level model to achieve closed-loop coupling between the road segment efficiency target and the single-vehicle lane-changing action.

[0111] The partial derivative of the travel time function with respect to the lane flow regulation coefficient is:

[0112] In the In each rolling iteration, the lane flow adjustment coefficient is updated according to the following formula:

[0113] in, For projection operators, To optimize step size, For feedback correction coefficient, For the first The feedback deviation term for each lane is defined as follows:

[0114] in, and Lane flow and density are obtained from feedback in physical space. and Lane flow and density are predicted from the information space. and To prevent positive numbers with a denominator of zero, For density feedback weights.

[0115] CPS state residuals are defined as:

[0116] in, This is the physical space feedback state. Predicting the state of information space, For the normalized weight matrix, It is a positive integer. When the preset threshold is exceeded, data fusion, candidate gap generation, safety feasibility assessment and collaborative benefit calculation are re-executed to correct decision deviations caused by communication delays, perception errors or vehicle execution deviations.

[0117] The bilevel programming rolling iteration stops when the following conditions are met:

[0118] in, The convergence threshold of the upper-level objective function. The threshold for lane-changing actions. This is a difference measurement function between candidate lane-changing actions.

[0119] S7: Generates longitudinal speed planning and lateral lane-changing trajectory based on the lane-changing action of a single vehicle, and sends the longitudinal speed planning and lateral lane-changing trajectory to the vehicle control system; after the vehicle executes, it sends the vehicle execution status, lane flow, lane density and communication status back to the information space as physical feedback status, updates the CPS closed-loop model based on the physical feedback status, and enters the next decision cycle.

[0120] The horizontal trajectory is represented as:

[0121] And satisfy the following boundary conditions:

[0122]

[0123] in, This indicates the current lateral position of the lane centerline. The lateral position is defined by the centerline of the target lane. These boundary conditions ensure that the lateral velocity and lateral acceleration of the lane-changing trajectory are zero at the beginning and end of the lane change, thus improving vehicle smoothness.

[0124] Example 2: In this embodiment, the system uses a fixed decision cycle. run, The timeframe can be between 0.2s and 1.0s. At the start of each decision cycle, the system collects vehicle operating status, lane traffic status, and communication topology status to form the information space CPS state. For intelligent vehicles with collaborative lane-changing capabilities, the system generates a set of candidate lane-changing actions based on the availability of the current lane, adjacent lanes, and vehicle distribution in the target lane. .

[0125] After candidate actions are generated, the system first performs a safety feasibility screening. For any candidate action... The safety constraint vector is calculated based on the forward safety distance, the backward safety distance, the predicted acceleration of vehicles changing lanes, and the predicted acceleration of vehicles following in the target lane. .when If the desired lane change is made based solely on the target clearance size, the candidate action is included in the collaborative benefit evaluation; otherwise, the candidate action is eliminated. This avoids the problem of ignoring the risk of sudden deceleration of vehicles behind in the target lane when lane change decisions are made solely based on the target clearance size.

[0126] For candidate actions that satisfy the safety and feasibility domain, the system further calculates the collaborative benefits. .like If the candidate action is safe and feasible, it indicates that its overall contribution to the vehicle's traffic efficiency, adjacent vehicle disturbance suppression, and lane load balancing is insufficient, and the vehicle will continue to travel in the current lane; if one or more candidate actions satisfy... Then, proceed to the next level to optimize the cost of bicycle travel.

[0127] In the lower-level model, vehicles use candidate lane-changing actions as decision variables, comprehensively considering individual vehicle time cost, comfort cost, global traffic cost, and collaborative benefits to determine the current lane adjustment coefficient. Optimal lane-changing maneuver under given conditions When multiple intelligent vehicles simultaneously request to switch into the same target gap, the system resolves the conflict based on safety margin, cooperation benefits, and the available time of the target gap. For vehicles in conflict, the action with the higher overall ranking is retained, while the remaining vehicles either stay in their lane or enter the next cycle to regenerate candidate actions.

[0128] In the upper-level model, the system uses the lane flow regulation coefficient. The decision variables are total travel time of the road segment, lane load imbalance, and the magnitude of change in the adjustment coefficient, which are the optimization objects. The set of lane-changing actions output by the lower layer will change the predicted traffic flow of each lane. ,density And speed distribution, the upper-level model is updated according to the changed traffic conditions. This will guide the next round of lane-changing actions at the lower level.

[0129] In the In each rolling iteration, the system follows the principle of "given by the upper layer". —Lower layer selection —Updated traffic forecasts—Feedback and corrections The solution is performed in the order of "...". The iteration stops when the change in the upper-level objective function is less than a preset threshold, and the change in lane-changing actions of each intelligent vehicle is less than a preset action difference threshold, and the optimal set of lane-changing actions for the current period is output. If the CPS state residual... If the preset threshold is exceeded, it indicates that communication latency, perception error, vehicle execution deviation, or external traffic disturbance has caused a significant deviation between information space prediction and physical space feedback. The system then re-executes data fusion, candidate gap generation, security evaluation, and collaborative benefit evaluation.

[0130] Once the optimal lane-changing maneuver is determined, the system generates longitudinal speed planning and lateral lane-changing trajectory, and sends them to the vehicle control system for execution. After execution, the physical space feedback status is transmitted back to the information space to update the vehicle status, lane flow, lane density, and CPS state residuals, before entering the next decision cycle. Through this process, the upper-level road segment efficiency objective and the lower-level individual vehicle safety and comfort objective are collaboratively optimized within the same CPS closed loop.

[0131] Example 3: To verify the technical effectiveness of the method of this invention in a mixed traffic environment, a micro-simulation experiment of a two-lane mixed traffic system was constructed. The experimental road was designed with a slow-vehicle bottleneck and different intelligent vehicle penetration rates. The lane baseline maintenance method, the ordinary lane-changing method (GLC), and the CPS-CLC method of this invention were compared. For each penetration rate, five sets of random seeds were used for repeated simulations, and the average value was taken as the experimental result. The experimental parameters are shown in Table 1.

[0132] Table 1

[0133] With a smart car penetration rate of 60%, the main improvements of the CPS-CLC method of this invention compared to the ordinary lane-changing GLC method are shown in Table 2.

[0134] Table 2

[0135] Please see Figures 6 to 9 Experimental results show that the method of the present invention can improve the average speed of road segments, reduce the equivalent travel time, and increase the number of vehicles passing through road segments under different intelligent vehicle penetration rates. At a 60% intelligent vehicle penetration rate, the method of the present invention also improves speed variance, sudden deceleration event rate, and energy consumption proxy index, indicating that it can not only improve traffic efficiency but also reduce lane-changing disturbances and improve traffic flow stability.

[0136] Compared to conventional lane-changing methods, the CPS-CLC method of this invention has the following advantages: Conventional lane-changing methods mainly rely on local clearance and individual vehicle benefits for decision-making, which can easily cause deceleration of following vehicles in the target lane and local traffic disturbances; the method of this invention corrects the prediction bias in the information space through CPS closed-loop feedback, eliminates high-risk lane-changing actions through a safe and feasible region, evaluates the vehicle's benefits, adjacent vehicle disturbances, and lane load changes through a collaborative benefit function, and coordinates road segment efficiency and individual vehicle behavior through a two-level programming model. Therefore, this invention can achieve more stable, safer, and more collaborative lane-changing decisions in mixed traffic environments.

[0137] Example 4: An electronic device, comprising a memory and a processor; The memory is used to store computer programs; The processor is configured to implement the method described in Embodiment 1 when executing the computer program.

[0138] Example 5: A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in Embodiment 1.

[0139] Example 6: A computer program product includes a computer program that, when executed by a processor, implements the method described in Example 1.

[0140] In the above embodiments, the reference to "this embodiment" in the specification indicates that a specific feature, structure, or characteristic described in connection with the embodiment is included in at least some embodiments, but not necessarily all embodiments. Multiple appearances of "this embodiment" do not necessarily refer to the same embodiment.

[0141] In the above embodiments, although the invention has been described in conjunction with specific embodiments thereof, many substitutions, modifications, and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory structures (e.g., dynamic RAM (DRAM)) may be used with the embodiments discussed. The embodiments of the invention are intended to cover all such substitutions, modifications, and variations falling within the broad scope of the appended claims.

[0142] As will be understood by those skilled in the art, the computer-readable storage medium described in this embodiment allows for the implementation of all or part of the steps in the above method embodiments by computer program-related hardware. The aforementioned computer program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0143] The electronic terminal provided in this embodiment includes a processor, a memory, a transceiver, and a communication interface. The memory and the communication interface are connected to the processor and the transceiver and complete communication between them. The memory is used to store computer programs, the communication interface is used to perform communication, and the processor and the transceiver are used to run the computer programs, so that the electronic terminal performs the steps of the above method.

[0144] In this embodiment, the memory may include random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device.

[0145] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0146] This invention can be used in a wide range of general-purpose or special-purpose computing system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices, etc.

[0147] This invention can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This invention can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0148] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A cooperative lane-changing decision-making method for intelligent vehicles based on cyber-physical systems, characterized in that: Includes the following steps: S1: Collect data on lane-changing vehicles, surrounding vehicles, road lanes, traffic flow status, and vehicle communication topology, and preprocess the data to obtain fused status data for collaborative lane-changing decisions. S2: Construct a CPS closed-loop model that includes a physical traffic layer, a perception and communication layer, an information mapping layer, a collaborative decision-making layer, and a control execution layer, and map the vehicle operation status, lane traffic status, and communication topology status in the physical space to a digital twin status in the information space. S3: Based on the digital twin state, identify the candidate target lane and candidate target gap of the lane-changing vehicle, generate a set of candidate lane-changing actions, and predict the motion response of the lane-changing vehicle, the vehicle in front of the target lane, the vehicle behind the target lane, and the adjacent vehicles within the communication range after each candidate lane-changing action is executed. S4: Based on the actual forward distance between the lane-changing vehicle and the vehicle in front of the target lane, the actual backward distance between the lane-changing vehicle and the vehicle behind the target lane, the minimum forward safe distance, the minimum backward safe distance, the predicted acceleration of the lane-changing vehicle, and the predicted acceleration of the vehicle behind the target lane, a safety constraint vector for the candidate lane-changing action is constructed, and the safety constraint vector being less than or equal to zero is used as the criterion for determining the safe feasible region. S5: Within the safe and feasible domain, a cooperative lane-changing benefit function is constructed based on the lane-changing vehicle passage benefit, the acceleration disturbance cost of adjacent vehicles within the communication range, the improvement of lane flow variance before and after lane changing, the target gap risk penalty, and the trajectory control smoothing cost. Candidate lane-changing actions that meet the benefit threshold are then selected based on the cooperative lane-changing benefit function. S6: Establish an upper-level road segment traffic efficiency optimization model and a lower-level single-vehicle travel cost optimization model. The upper-level road segment traffic efficiency optimization model outputs a lane flow adjustment coefficient. The lower-level single-vehicle travel cost optimization model selects single-vehicle lane-changing actions under the constraints of the lane flow adjustment coefficient, the safe and feasible domain, and the collaborative benefit threshold. The coupling between the road segment efficiency target and the single-vehicle lane-changing actions is achieved through rolling iteration. S7: Generate longitudinal speed planning and lateral lane-changing trajectory based on the single vehicle lane-changing action, and send the longitudinal speed planning and lateral lane-changing trajectory to the vehicle control system; after the vehicle executes, it sends the physical feedback state back to the information space, updates the CPS closed-loop model, and enters the next decision cycle.

2. The intelligent vehicle cooperative lane-changing decision-making method based on cyber-physical systems according to claim 1, characterized in that: In step S2, the digital twin state in the information space is represented as follows: in, For the first The car in Vehicle status during each decision cycle For vehicle assembly, and The first Lane density and traffic flow, For lane collection, This is the adjacency matrix of the vehicle communication topology. This represents the residual state of the model. The first The car in Vehicle status during each decision cycle Represented as: in, The longitudinal position of the vehicle's center of gravity. The vehicle's lateral position. For vehicle speed, To accelerate the vehicle, This refers to the vehicle's heading angle; The model residual state The value is determined based on the difference between the physical space feedback state and the information space prediction state, and is expressed as: in, This is the physical space feedback state. For predicting the state of information space, For the normalized weight matrix, To prevent positive numbers with a denominator of zero; Vehicle longitudinal state prediction satisfies: in This represents the sampling time interval.

3. The intelligent vehicle cooperative lane-changing decision-making method based on cyber-physical systems according to claim 2, characterized in that: In step S3, the first The set of candidate lane-changing actions for a smart car is represented as: in, The vehicle's current lane, For candidate target lanes, The candidate target gap is defined by the ALV of the vehicle ahead in the target lane and the AFV of the vehicle behind in the target lane. For lane change duration, This represents the number of candidate lane-changing maneuvers.

4. The intelligent vehicle cooperative lane-changing decision-making method based on cyber-physical systems according to claim 3, characterized in that: In step S4, let the lane-changing vehicle be SV, the vehicle in front of the target lane be ALV, and the vehicle behind the target lane be AFV. The actual forward distance between the lane-changing vehicle and the vehicle in front of the target lane is... And the actual rearward distance between the vehicle changing lanes and the vehicle behind in the target lane. They are respectively: in, , , These represent the longitudinal positions of the centroids of the vehicle changing lanes, the vehicle in front of the vehicle in the target lane, and the vehicle behind the vehicle in the target lane, respectively. , and These are the vehicle lengths of the vehicle changing lanes, the vehicle in front of the target lane, and the vehicle behind the target lane, respectively. Forward minimum safe distance and minimum safe distance backward They are defined as follows: in, For stationary safe distance, For reaction time, , , These are the available decelerations for the corresponding vehicles. ; Candidate lane change action The safety constraint vector is: in, and These are the forward and backward safety margins, respectively. Predicted acceleration for vehicles performing candidate actions after changing lanes. The predicted acceleration of the vehicle following in the target lane after being affected by the lane change. For the comfortable deceleration threshold, For safe deceleration threshold; when the following conditions are met At that time, the candidate lane change action falls within the safe and feasible domain.

5. The intelligent vehicle cooperative lane-changing decision-making method based on cyber-physical systems according to claim 4, characterized in that: The cooperative lane-changing benefit function mentioned in step S5 is: in, For the benefit of vehicles changing lanes. This represents the improvement in the variance of lane traffic flow before and after a lane change. The cost of acceleration disturbance of nearby vehicles within the communication range, Penalty for target gap risk, The cost of trajectory control smoothing , , , , All are non-negative weighting coefficients; The traffic revenue of lane-changing vehicles The calculation is as follows: in: in, For the first A number of vehicles performing lane-changing maneuvers The predicted average speed during the lane change duration, For the first The predicted average speed of vehicles changing lanes within the same time range while maintaining their current lane. For the first A number of vehicles performing lane-changing maneuvers Later in time Prediction speed, For the first When a vehicle changing lanes maintains its current lane at a certain time Prediction speed, For the first The maximum permissible speed for vehicles changing lanes. To prevent positive numbers with a denominator of zero; when When, it indicates that a candidate lane change action is being executed. It can increase the average travel speed of vehicles changing lanes during the lane-changing duration; when When this occurs, it indicates that executing the candidate lane-changing action cannot generate positive traffic benefits; The improvement in lane load balance before and after lane changing The calculation is as follows: in, The first lane change action after the candidate lane change action is executed Predicted traffic flow for each lane; The cost of disturbance from nearby vehicles The calculation is as follows: in For the communication range of lane-changing vehicles, For nearby vehicles The perturbation weights, Acceleration for adjacent vehicles before they change lanes. The predicted acceleration of adjacent vehicles after a candidate lane change is executed; The target gap risk penalty The calculation is as follows: in, and This is the risk attenuation coefficient. Weighting of penalties for vehicles behind in the target lane that suddenly decelerate; and To quantify the safety margin of the risk level of candidate lane-changing maneuvers, it is expressed as: The trajectory control smoothing cost The calculation is as follows: in, and They are longitudinal and lateral jerk, respectively. and These are the vertical and horizontal smoothing weights, respectively.

6. The intelligent vehicle cooperative lane-changing decision-making method based on cyber-physical systems according to claim 5, characterized in that: Step S6, which involves establishing an upper-level road segment traffic efficiency optimization model and a lower-level single-vehicle travel cost optimization model, includes: The travel time function for each lane is defined as: in, For the first Lane flow regulation coefficient for each lane For the first Free-flow travel time for each lane For lane capacity, and Let be the impedance function parameter, and ; The traffic efficiency optimization model for the upper-level road segment uses the lane flow adjustment coefficient. Let be the decision variable, and its objective function be: in, , This is the set of vehicle lane-changing actions output from the lower layer. To execute a set of actions The next Predicted traffic flow for each lane The average predicted traffic flow for each lane. This refers to the lane flow adjustment coefficient from the previous decision-making cycle. , , These are non-negative weighting coefficients; The lower-level bicycle travel cost optimization model uses candidate lane-changing actions. Let be the decision variable, and its objective function be: in, For the time cost of a single vehicle, For comfort costs, The global traffic cost on the road segment caused by the candidate action. , , , These are non-negative weighting coefficients; The optimal lane-changing maneuver for a single bicycle is determined using the following formula: And it satisfies: in, The threshold for synergistic benefits; The upper and lower layers together constitute a two-layer programming model: in A collection of intelligent vehicles participating in collaborative decision-making; The upper-level road segment traffic efficiency optimization model influences the selection of candidate lane-changing actions in the lower-level single-vehicle travel cost optimization model through the lane flow adjustment coefficient. The single-vehicle lane-changing actions output by the lower-level single-vehicle travel cost optimization model change the lane flow, lane density, and vehicle speed distribution, and feed the changes back to the upper-level road segment traffic efficiency optimization model.

7. The intelligent vehicle cooperative lane-changing decision-making method based on cyber-physical systems according to claim 6, characterized in that: Step S6, which describes coupling the road segment efficiency target with the single-vehicle lane-changing action through rolling iteration, includes: the partial derivative of the travel time function with respect to the lane flow regulation coefficient is: In the In each rolling iteration, the lane flow adjustment coefficient is updated according to the following formula: in, For projection operators, To optimize step size, For feedback correction coefficient, For the first The feedback deviation term for each lane is defined as follows: in, and Lane flow and density are obtained from feedback in physical space. and Lane flow and density are predicted from the information space. and To prevent positive numbers with a denominator of zero, For density feedback weights; The bilevel programming rolling iteration stops when the following conditions are met: in, The convergence threshold of the upper-level objective function. The threshold for lane-changing actions. This is a difference measurement function between candidate lane-changing actions.

8. The intelligent vehicle cooperative lane-changing decision-making method based on cyber-physical systems according to claim 7, characterized in that: When multiple intelligent vehicles request to switch to the same target gap within the same decision cycle, the candidate lane-changing actions are ranked according to their safety margin, the value of the cooperative lane-changing benefit function, and the available time of the target gap. The candidate lane-changing action with the highest ranking is retained, and the remaining candidate lane-changing actions are canceled or recalculated in the next decision cycle.

9. The intelligent vehicle cooperative lane-changing decision-making method based on cyber-physical systems according to claim 8, characterized in that: The lateral lane-changing trajectory described in step S7 is represented as follows: And satisfy the following boundary conditions: in, This indicates the current lateral position of the lane centerline. The lateral position of the center line of the target lane.

10. A cyber-physical system-based intelligent vehicle cooperative lane-changing decision-making system, characterized in that: include: The acquisition module is used to collect vehicle operating status, road lane status, traffic flow status and vehicle communication topology data, and output fused status data to the CPS mapping module; The CPS mapping module is used to generate a digital twin state in the information space based on the fused state data, and to calculate the model residual state. The candidate gap generation module is used to generate a set of candidate target lanes, candidate target gaps, and candidate lane-changing actions based on the digital twin state. The safety feasibility evaluation module is used to filter candidate lane-changing actions that belong to the safe and feasible domain based on the actual forward spacing, actual backward spacing, minimum forward safety distance, minimum backward safety distance, predicted acceleration of lane-changing vehicles, and predicted acceleration of vehicles following in the target lane. The collaborative benefit evaluation module is used to calculate the collaborative lane-changing benefits of candidate lane-changing actions based on the lane-changing vehicle passage benefits, the acceleration disturbance cost of adjacent vehicles, the lane flow variance improvement, the target gap risk penalty, and the trajectory control smoothing cost. The two-level programming solution module is used to solve the traffic efficiency optimization model of the upper-level road segment and the single-vehicle travel cost optimization model of the lower-level road segment, and outputs the lane flow adjustment coefficient and single-vehicle lane-changing action. The trajectory generation module is used to generate longitudinal speed planning and lateral lane-changing trajectory based on the single vehicle lane-changing action, and output vehicle control commands. The feedback update module is used to correct the digital twin state in the CPS mapping module based on the vehicle execution status, lane flow, lane density and communication status, and trigger the next decision cycle.

Images