A virtual vehicle fleet-based mixed traffic ramp cooperative merging control method

Abstract

The application discloses a kind of based on virtual car team's mixed traffic ramp cooperative merging control method, and multiple virtual car teams are respectively dynamically organized on main lane and ramp, and the virtual car team of ramp is mapped to main lane to form initial merging car team sequence, supposing adjusting the traffic order of some virtual car teams in initial merging car team sequence, form new traffic sequence scheme as candidate traffic scheme, with the total deviation of the target longitudinal distance required by the virtual car team needing adjustment and the actual longitudinal distance between two adjacent virtual car teams in initial merging car team sequence minimization as the target to construct target model, and the optimal merging car team sequence is obtained by solving, with the position deviation, speed deviation and instantaneous fuel consumption rate of the leading vehicle of any virtual car team minimum as the target to construct multi-objective optimization model, and the acceleration information of the leading vehicle of corresponding virtual car team is obtained by solving, combined with the longitudinal dynamics model of virtual car team, control virtual car team to complete ramp merging according to optimal merging sequence.

Description

Technical Field

[0001] This invention belongs to the technical field of intelligent transportation, specifically relating to a method for cooperative merging control of mixed traffic ramps based on virtual vehicle fleets. Background Technology

[0002] With the acceleration of urbanization and the continuous growth of motor vehicle ownership, highway ramp merging areas, as key nodes for traffic flow conversion, have become high-incidence areas of traffic congestion and accidents. Ramp vehicles need to find gaps in the main lane traffic flow within a limited space to merge, often forcing vehicles in the main lane to slow down, change lanes, or give way, which can easily create queues and congestion shock waves that travel upstream, resulting in decreased traffic efficiency, increased fuel consumption, and enhanced safety risks.

[0003] Currently, common ramp control methods include ramp signal control based on gap acceptance theory, speed limit control, and local longitudinal control strategies. These methods mostly rely on macroscopic traffic flow models or empirical thresholds, and can only adjust ramp release rates and main lane speeds to a certain extent. They do not consider the spatiotemporal relationships between individual vehicles precisely enough, and cannot fully leverage the collaborative advantages of connected autonomous vehicles in mixed traffic scenarios.

[0004] With the development of vehicle-to-everything (V2X) and autonomous driving technologies, some studies have explored coordinated control of vehicles on main lanes and ramps using methods such as fleet control, virtual lane mapping, and model predictive control. However, most existing methods assume homogeneous vehicle types or high CAV penetration, making them ill-suited for real-world traffic environments where CAVs and HDVs coexist and HDVs exhibit strong randomness in driving behavior. Furthermore, in high-traffic and complex merging scenarios, control methods based on single-vehicle trajectory optimization or global coordination often face challenges such as high decision-making dimensionality, heavy computational burden, and insufficient real-time performance, making them difficult to implement in engineering.

[0005] Against the backdrop of the parallel development of electrification, connectivity, and intelligence, ramp merging control not only needs to focus on traditional traffic efficiency and safety constraints, but also needs to consider fuel economy and emission control. Given the limited penetration rate of commercial vehicles (CAVs) and the still relatively high proportion of high-density vehicles (HDVs), there is currently a lack of a ramp merging control method that can simultaneously address heterogeneous driving characteristics, real-time solvability, and multi-objective coordination. Virtual vehicle fleet technology simplifies the complex coordination problem into fleet following control by grouping dispersed vehicles, providing a new approach to ramp merging in mixed traffic flows. However, existing virtual vehicle fleet research largely focuses on fleet formation, lacking a systematic design for merging sequence optimization and multi-objective trajectory planning, making it difficult to cope with dynamically changing traffic scenarios. Therefore, there is an urgent need for a control method that integrates virtual vehicle fleets, multi-objective optimization, and dynamic coordination to solve the challenges of ramp merging in mixed traffic flows on highways. Summary of the Invention

[0006] This invention provides a hybrid traffic ramp cooperative merging control method based on virtual vehicle fleets. By dynamically constructing "1+n" virtual vehicle fleets on the main lanes and ramps, and using virtual lane mapping technology to transform multi-lane spatial conflicts into a one-dimensional vehicle fleet sequence problem, the method optimizes multi-objective merging sequence at the strategy level and plans vehicle fleet trajectory at the tactical level, thereby achieving unified coordinated control of CAV and HDV. This improves ramp merging efficiency and reduces overall fuel consumption under different traffic demands and demand allocation conditions.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A hybrid traffic ramp cooperative merging control method based on virtual vehicle fleets includes the following steps: Step 1: Dynamically assemble multiple virtual convoys on the main lanes and ramps of the highway, each led by a networked autonomous vehicle and followed by multiple manually driven vehicles. Step 2: Map the virtual convoys on the ramps onto the main lanes to form an initial merging convoy sequence. Assume that adjusting the passage order of some virtual convoys in the initial merging convoy sequence forms a new passage sequence scheme as a candidate passage scheme. Construct a target model with the goal of minimizing the total deviation between the target longitudinal spacing required for the virtual convoys to be adjusted in any candidate passage scheme and the actual longitudinal spacing between two adjacent virtual convoys in the initial merging convoy sequence, and solve for the optimal merging convoy sequence. Step 3: For the optimal merging vehicle sequence, construct a multi-objective optimization model with the goal of minimizing the position deviation, speed deviation, and instantaneous fuel consumption rate of the leading vehicle of any virtual vehicle, and solve for the acceleration information of the leading vehicle of the corresponding virtual vehicle. Step 4: Based on the longitudinal dynamics model of the virtual convoy, control the corresponding virtual convoy to complete the ramp merging according to the optimal merging sequence using the acceleration information of each leading vehicle.

[0008] Furthermore, in step two, the initial merging vehicle sequence is recorded. The target longitudinal spacing between adjacent virtual vehicle fleets in the same lane is That is, the desired following distance, which is the target longitudinal spacing between adjacent virtual convoys in different lanes. That is, the expected merging distance. First, use the following formula to calculate the actual longitudinal distance between two adjacent virtual vehicle convoys. Longitudinal distance from the target , in, For virtual fleets The position of the lead vehicle, For virtual fleets The total length; in, Lane identifiers represent virtual convoys; convoys in the main lane are assigned a value of 1, while convoys in the ramps are assigned a value of 0. This indicates the index of the virtual vehicle fleet within the initial merging vehicle sequence. Indicates being in the first The sequential index of a virtual car fleet following a virtual car fleet. This represents the virtual vehicle fleet that needs to be adjusted in any candidate traffic scheme. Required target longitudinal spacing; Introducing binary decision variables The definition is as follows: ， Reconstruct the target model The feasibility constraints are: Then, the non-dominated sorting genetic algorithm NSGA-II is used to search for an approximate Pareto optimal solution, and the scheme with the minimum overall cost is selected as the optimal merging vehicle sequence.

[0009] Furthermore, let any two adjacent virtual convoys in the optimal merging convoy sequence be denoted as convoys. convoy And the convoy Located in the team Ahead of them, the longitudinal distance between the two convoys is... ,in, , The actual longitudinal distance between the two convoys can be calculated using the following formula. , in, Indicates the team The leading vehicle in the middle of the time Location, Indicates the team The leading vehicle in the middle of the time Location, Indicates the team The total length; like If so, then the two teams will remain as they are. like Then the convoy merged into the team They are merged into a single virtual fleet, which controls the fleet. The lead vehicle gradually moved towards the convoy They move closer together until the actual longitudinal distance between them reaches the target longitudinal distance.

[0010] Furthermore, the distances from the lead vehicle of each virtual convoy on the main lane and ramp to the merging point are arranged in ascending order to map the virtual convoys on the ramp onto the main lane, forming the initial merging convoy sequence. .

[0011] Furthermore, in step three, a multi-objective optimization model is first constructed using the following formula: in, Indicates the length of the prediction time domain. To predict the time step in the time domain, Indexing the fleet, This is a time discount factor used to suppress uncertainty in long-term forecasts. These are the loss coefficients, used to balance the relative weights of position and velocity deviations in the overall target. For position and velocity error tolerances, used to normalize deviations of different dimensions, weighting coefficients. These correspond to efficiency and fuel economy, respectively, to meet the requirements. , ,and , Indicates instantaneous fuel consumption rate. , They respectively represent the position of the first The actual position and speed of the virtual convoy. , They represent the first The ideal position and ideal speed of the virtual convoy passing through the area are determined by the optimal merging convoy sequence. The determined platoon topology following relationship is obtained by recursive calculation based on the motion state and safety threshold of the preceding virtual platoon; The non-dominated sorting genetic algorithm NSGA-II is then used to solve the multi-objective optimization model, using the acceleration sequence of the leading vehicle in the prediction time domain as the decision variable, and searching for an approximate Pareto optimal solution set in the feasible solution space.

[0012] Furthermore, in the process of forming a virtual fleet in step one, for any vehicle entering the formation area PZ... First, calculate its relationship with the preceding virtual convoy. The tail car Initial spacing between , in, Indicates vehicle Time of arrival at formation zone PZ and They represent The last car of the virtual convoy before the timeline and vehicles Location, For vehicles Length; Define minimum safe merge distance : in, , These represent the preceding virtual convoys. Nakao car and vehicles exist The speed of time, Indicates the preceding virtual convoy Nakao car Distance to the end of the formation zone (PZ) For maximum comfort acceleration, Maximum speed; Then, the rules for forming the organization are established; When the vehicle When it is HDV, if Then the vehicle Join the preceding virtual convoy; otherwise, remain as a standalone HDV (ungrouped vehicle). When the vehicle When it is CAV, if Then the vehicle As the lead vehicle of the new virtual fleet, it is used to form the virtual fleet; if Within the longest platoon length behind it, there are multiple HDVs, and the preceding virtual platoon... The maximum convoy length has been reached; move the vehicles... As the lead vehicle of the new virtual fleet; otherwise, join the preceding virtual fleet. As a member.

[0013] Furthermore, the main lanes and ramps are divided sequentially from upstream to the merging point into a formation zone PZ, a cooperative control zone CZ, and a merging zone MZ. Virtual fleet formation is completed in the formation zone PZ, and the calculations for steps two and three are completed in the cooperative control zone CZ.

[0014] Compared with the prior art, the beneficial effects of the present invention are: (1) Improve the adaptability of mixed traffic. Through the “1+n” virtual fleet model, CAVs lead HDVs, reducing the interference of random behavior of HDVs on traffic flow. It can operate stably in mixed scenarios with CAV penetration rate of 30%-80%. Compared with uncontrolled scenarios, the total travel time is reduced by up to about 33.3%, which significantly alleviates queuing and congestion at ramp merging points.

[0015] (2) Simplify the complexity of collaborative control. Virtual lane mapping technology transforms two-dimensional multi-lane coordination into a one-dimensional fleet following problem, reducing the dimension of decision variables and the number of merging conflicts. Combined with the NSGA-II algorithm to optimize the merging order, it ensures global optimality and significantly improves merging efficiency.

[0016] (3) Achieve multi-objective optimization balance. With total driving time and fuel consumption as optimization objectives, improve fuel economy while improving traffic efficiency by quantifying fuel consumption model parameters, thus balancing efficiency and environmental protection.

[0017] (4) Enhance driving safety by using dynamic safety distance threshold and real-time trajectory adjustment to avoid vehicles from accelerating, decelerating and cutting in line, thereby reducing the speed difference and spacing fluctuation of adjacent vehicle platoons; at the same time, through the rolling optimization mechanism, the control framework can maintain a stable improvement effect under various traffic demand combinations and random initial conditions. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the overall system flow of the present invention; Figure 2 This is a schematic diagram of the control area division of the present invention, showing the location and range of the formation area PZ, the cooperative control area CZ, and the merging area MZ, as well as the virtual lane mapping process; Figure 3 This is a schematic diagram of the merging sequence optimization process of the present invention, wherein stage T1 shows the process based on the distance to the merging point. The initial merging convoy sequence is shown in local configurations. Phase T2 presents the globally optimized traffic flow scheme, which involves recombining conflicting or inefficient local arrangements in the initial merging convoy sequence to obtain the globally optimal merging convoy sequence that minimizes the deviation between the actual longitudinal spacing of the overall convoys and the target longitudinal spacing. ; Figure 4 This is a spatiotemporal trajectory diagram of vehicles with a traffic flow of 2200veh / h, illustrating the spatiotemporal trajectories of traffic under different strategies. Figure 5 This is a comparison chart showing the fuel consumption of the control method of this invention and the traditional strategy at flow rates of 1400veh / h, 1800veh / h, and 2200veh / h. Detailed Implementation

[0019] To make the technical means, creative features, objectives and effects of the present invention easier to understand, the following embodiments, in conjunction with the accompanying drawings, specifically illustrate the hybrid traffic ramp cooperative merging control method based on virtual fleets of the present invention. It should be noted that the description of these embodiments is for the purpose of helping to understand the present invention, but does not constitute a limitation of the present invention.

[0020] like Figure 1 As shown, this invention provides a hybrid traffic ramp cooperative merging control method based on virtual vehicle fleets. Through a virtual lane mapping mechanism, the complex two-dimensional coordination problem of multiple vehicle fleets is transformed into a one-dimensional fleet following problem. A 1+n type virtual vehicle fleet is dynamically constructed based on real-time vehicle status, where one CAV leads multiple HDVs. A central coordinator optimizes the fleet merging order and generates speed trajectories. HDV car-following control is achieved by combining a full-speed difference FVD model. Multi-objective optimization ensures a smooth and efficient merging process, significantly improving ramp merging efficiency, reducing fuel consumption, and alleviating congestion in highway merging areas. It is suitable for complex traffic scenarios such as peak hours. Furthermore, the method runs in real-time at a frequency of 1Hz on an RSU / edge computing platform and can be seamlessly integrated into simulation environments such as SUMO or real-vehicle V2X control units, exhibiting deployability and high robustness.

[0021] Specifically as follows: Step 1: Dynamically assemble multiple virtual convoys on the main lanes and ramps of the highway, each led by a networked autonomous vehicle (CAV) and followed by multiple manually driven vehicles (HDVs).

[0022] S11, Data Acquisition like Figure 2 As shown, around the merging area of ​​the ramps, based on road geometry and traffic characteristics, the main lanes and ramps are divided into a formation zone (PZ), a coordinated control zone (CZ), and a merging zone (MZ) from upstream to the merging point. Preferably, the length of the PZ is 200–800 m, the length of the CZ is 500–2000 m, and the length of the MZ is 100–300 m.

[0023] For data collection, detectors and V2X communication modules are deployed in the formation zone (PZ) and cooperative control zone (CZ) of highway ramps and main lanes to collect traffic flow data and vehicle status information at predetermined time steps (e.g., 1 second). Traffic flow data acquisition: Traffic flow of each lane on the main road and ramps (veh / h), density (veh / km), average speed (km / h); Vehicle status information acquisition: CAVs upload real-time location via V2X communication. (m) speed (km / h), acceleration (m / s²); HDVs acquire their status via laser detectors, simultaneously collecting data on vehicle type (CAV / HDV) and length. (m) and mass m (kg), etc.

[0024] S12, Data Preprocessing Kalman filtering algorithm is used to remove noise and outliers from traffic flow data; vehicle status information is synchronized in time to ensure the timeliness of V2X communication data; Calculate the average speed of all vehicles within the control area. With average density Where N is the total number of vehicles in the control area and M is the number of lanes, the real-time traffic status of the control area is obtained, providing input for the subsequent construction and optimization of virtual fleets.

[0025] S13, Establish a "1+n" virtual fleet Within the formation zone (PZ), vehicles entering the control zone are divided into virtual platoons based on vehicle type and motion status. CAVs are prioritized as potential lead vehicles, while HDVs are determined to join the preceding platoon based on a safe distance threshold.

[0026] When vehicles enter the formation zone (PZ), none of the vehicles are pre-assigned to a platoon. For any vehicle entering the PZ area... First, calculate the vehicle With the preceding virtual convoy tail car initial spacing : in, Indicates vehicle Time of arrival at formation zone PZ; and They represent Preceding Virtual Fleet Nakao car and vehicles Location; For vehicles The length of the vehicle is assumed to be the same for all vehicles in this invention. Equation (1) is used to characterize the available longitudinal space between the candidate vehicle and the preceding convoy.

[0027] Secondly, determine the vehicle Can it achieve maximum comfort acceleration? Under constraints, they caught up with the preceding convoy.

[0028] Minimum safe merge distance based on vehicle kinematics characteristics The following vehicle must be able to smoothly reduce its distance before entering the merging zone (MZ), and its terminal speed must not exceed [the specified value]. The value is determined by the vehicle's speed, acceleration, and length, and the specific calculation formula is as follows: in, , These are the preceding virtual convoys. Nakao car and vehicles exist The speed of time, Indicates the preceding virtual convoy Nakao car Distance to the end of the formation zone (PZ) For maximum comfort acceleration, The maximum speed is given by equation (2). Equation (2) is used to estimate the maximum relative distance that a candidate vehicle can catch up to without exceeding the comfort acceleration.

[0029] Secondly, the rules for forming a racing team are designed, specifically including the following points: To avoid excessively long queues reducing traffic flow efficiency, a maximum queue length is defined. in, and These represent the maximum platoon length for the main lane and the ramp, respectively. The maximum size of a single platoon is limited to "1+5", which is 1 CAV + 5 HDVs, to avoid excessively long platoons affecting traffic flexibility.

[0030] When the vehicle When it is HDV, if Then the vehicle Join the preceding convoy; otherwise, remain as a solo rider as an ungrouped HDV. When the vehicle When it is CAV, if Then the vehicle As the lead vehicle of the new virtual fleet, establish a "1+n" virtual fleet; if The longest convoy behind it The memory contains multiple HDVs (not less than n=5 vehicles), and the preceding virtual convoy... The maximum fleet length has been reached. Put the vehicle As the lead vehicle of the new virtual fleet, it provides a new CAV lead vehicle for subsequent HDVs; otherwise, it joins the preceding virtual fleet. As a member, the fleet formation mechanism ensures that vehicles on both the main lanes and ramps can dynamically generate a 1+n type hybrid fleet structure.

[0031] S14. Longitudinal dynamics modeling of the "1+n" virtual vehicle fleet, specifically including the following points: To characterize the longitudinal evolution of a “1+n” virtual fleet under mixed traffic conditions during the collaborative merging process, this invention models the fleet as a linear state-space system based on the FVD car-following model.

[0032] Each virtual convoy consists of one CAV (Central Autonomous Vehicle) as the lead vehicle and n HDV (High-Density Vehicle) as followers, numbered sequentially from 1 to n+1. Let the longitudinal position and speed of the lead vehicle be denoted as follows: , The vehicle length is In this invention, it is assumed that vehicles within the same fleet have the same length. The speed of the following vehicle is ( ), No. The distance between the rear of the following vehicle and the rear of the vehicle in front is : (4) In a mixed vehicle convoy, the longitudinal acceleration of the HDV is given by the FVD model: (5) in, For the first The longitudinal acceleration of an HDV; , This is the reaction coefficient, and its value range is based on the actual vehicle following calibration results; Based on the distance between the front and rear of the vehicle The given desired speed function is in piecewise form based on the distance between the vehicle heads: (6) in, This is the interval between parking / strong deceleration thresholds. The spacing for transitioning to a free-flow state; This represents the maximum permissible operating speed for the road. (Middle section function) A linear approximation can be used, specifically represented as: (7) Thus ensuring when As the speed increases within a safe range, the desired speed increases monotonically and... Smooth transition to .

[0033] For CAV lead vehicle Its acceleration is determined by the control input given by the central coordinator: (8) in, The control input for the virtual convoy represents the expected longitudinal acceleration of the CAV lead vehicle during the current control cycle. The CAV broadcasts its own status and control commands to the HDV of its convoy via V2X communication at 1 s intervals to achieve speed guidance from 1 to n; the HDV then automatically adjusts according to the FVD model in equation (5).

[0034] The states of the lead vehicle and each following vehicle are aggregated into a platoon state vector: (9) Based on this, the longitudinal dynamics model of the virtual vehicle fleet can be represented in the following state-space form: (10) in, For the control input (desired longitudinal acceleration) of the CAV navigator, matrices M and N are represented in block matrix form as follows: (11) Each of them , , , , Each block is a 2×2 or 2×1 submatrix, specifically in the form of: (12) in, , , The vehicle following response coefficient comprehensively reflects the driver's sensitivity to the relative distance and speed difference of the vehicle in front. Its value can be obtained through measurement data or simulation data calibration.

[0035] The aforementioned 1+n virtual vehicle longitudinal dynamics model provides unified state evolution constraints, as well as vehicle position, speed, and acceleration updates, for subsequent virtual lane mapping, merging order optimization, and multi-objective speed trajectory planning.

[0036] Step 2: Map the virtual convoys on the ramps onto the main lanes to form an initial merging convoy sequence. Assume that adjusting the passage order of some virtual convoys in the initial merging convoy sequence to form a new passage sequence is a candidate passage scheme. Construct a target model with the goal of minimizing the total deviation between the target longitudinal spacing required by the virtual convoys to be adjusted and the actual longitudinal spacing between two adjacent virtual convoys in the initial merging convoy sequence, and solve for the optimal merging convoy sequence.

[0037] S21. Construct virtual lanes, which includes the following: Collect the distance from the lead vehicle of all virtual convoys to the MZ entrance of the merging zone. It is uploaded to the central coordinator via V2X communication; the central coordinator then... The virtual vehicle fleets on the main lanes and ramps are arranged in ascending order, thereby mapping the virtual vehicle fleets on the ramps onto the main lanes, constructing virtual lanes, and generating virtual vehicle fleet sequences. That is, the initial merging vehicle sequence The ideal target longitudinal spacing between two adjacent virtual convoys is set as: the convoy spacing within the same lane. Spacing between vehicles in different lanes .

[0038] The smaller the index of the virtual convoy sequence, the closer the location is to the merging point. This virtual lane is only used for dimensionality reduction, and the specific merging order still needs to be determined through subsequent multi-objective optimization.

[0039] S22. Optimize the merging order of platoons on the virtual lane, specifically including the following points: In a merging scenario where a single-lane mainline merges with a ramp, the initial merging vehicle sequence Merging must be completed sequentially through the merging point. If merging is strictly followed in the order of arrival at the merging area entrance, the mainline convoy often needs to frequently slow down to yield to the ramp convoy, thus compressing the available longitudinal space near the merging area entrance and increasing the overall travel time. This is due to the initial merging convoy sequence. As an inseparable whole participating in traffic, the order in which different convoys pass through the merging point directly affects the utilization of longitudinal space and merging efficiency. Therefore, by rationally adjusting the convoy passage order near the merging zone entrance, so that convoys in the same lane pass through the merging zone as continuously as possible (e.g., changing from "mainline-ramp-mainline-ramp" to "mainline-mainline-ramp-ramp"), unnecessary deceleration and waiting can be effectively reduced, longitudinal space utilization efficiency can be improved, and overall travel time can be shortened. This is the core idea of ​​convoy coordination strategies to improve merging efficiency.

[0040] Based on the above considerations, the virtual lane only provides an initial passage order based on the distance between the convoy and the merging zone entrance, describing the relative spatial relationships of the convoy near the merging zone entrance, and does not represent the final merging passage order. For example... Figure 3 As shown, in the initial merging vehicle sequence Based on this, through feasibility analysis and optimized allocation of the longitudinal space near the merging zone entrance, an optimal merging vehicle sequence is determined for the virtual vehicle sequence, which satisfies safety constraints and is more stable, efficient and energy-saving overall.

[0041] S221. Construct the objective function: In the initial merging vehicle sequence In the middle, two adjacent convoys form a longitudinal "slot" gap near the entrance to the merging zone. Let the convoy be located in the sequential position. and If two adjacent convoys are the preceding convoy and the following convoy, then the actual longitudinal spacing they form near the merging zone entrance is defined as: in, and These represent the longitudinal positions of the lead vehicles in the preceding and following virtual convoys near the merging zone entrance, respectively. This represents the total length of the preceding virtual convoy. This slot gap indicates the available longitudinal space between the rear of the preceding convoy and the front of the following virtual convoy, which can be used to support adjustments to the passage order.

[0042] Construct candidate passage schemes for passage order adjustment: In the initial merging vehicle sequence To reduce the number of times the mainline convoy needs to slow down, the passage sequence will be adjusted to ensure that convoys in the same lane pass through the merging point consecutively, thus forming a candidate passage sequence. There are N candidate routes, from which the optimal route is selected.

[0043] Team Adjust the target longitudinal spacing to the desired sequence of passes: When the initial merging vehicle sequence Central Plains First One of the teams behind the merging teams In the adjusted candidate passage sequence, it is the [number]. The merging convoys will proceed in two phases, with the preceding and following convoys forming the initial merging convoy sequence. The preceding convoy (located in) (position) and the following convoy (originally the first) The adjusted position Bit).

[0044] To ensure safe merging, sufficient longitudinal space must be reserved for the convoy under this traffic sequence. In any candidate traffic route, the convoy... The ideal target longitudinal spacing that needs to be satisfied is defined as: in, Lane markings indicate the number of lanes used for convoys: 1 for mainline convoys and 0 for ramp convoys. This indicates the index of the virtual vehicle fleet within the initial merging vehicle sequence. Indicates being in the first The sequential index of the fleet following the virtual fleet; The expected following distance under the same lane conditions. This is the expected merging distance between the mainline and the ramp. The squared term is used to determine whether adjacent convoys originate from the same lane, thereby automatically selecting the corresponding safe distance threshold.

[0045] Insertion position selection variable and order uniqueness constraint: To select a unique and realizable ordinal outcome from all possible candidate travel routes, a binary decision variable is introduced. If and only if the team Arranged as the first When a convoy passes through the merging zone entrance, Otherwise, it is 0. This variable is used to indicate the fleet size. Was it selected as the first The location is a merging zone, used to activate the objective function term and feasibility constraints associated with that location, where the convoy... As one of the existing fleets in the initial merging fleet sequence, the insertion operation only indicates that it has been reordered near the merging zone entrance, and does not introduce a new fleet entity.

[0046] To ensure the validity and uniqueness of the merging order, each convoy occupies only one passing position, and each passing position corresponds to only one convoy. A binary decision variable is introduced. : Spatial feasibility constraints and merging pattern identification: When the car team Arranged as the first in the passage order When a convoy passes through, the initial merging convoy sequence Central Plains First The actual available longitudinal gap in the merging zone is used for passage. The following spatial feasibility constraints must be met: When the above conditions are met, it indicates that the longitudinal space near the merging zone entrance is sufficient to support the convoy. As an independent team passing through the merging point, at this time the team The order of passage shall be adjusted as described above.

[0047] Merging order optimization objective function: Under the above definition and satisfying constraint formula 16, the merging order optimization aims to minimize the deviation between the actual longitudinal spacing of adjacent convoys near the merging zone entrance and the ideal target longitudinal spacing. The objective function is constructed as follows: By using this objective function, under the premise of satisfying spatial feasibility and safety constraints, a merging sequence that makes more reasonable use of longitudinal space is selected, thereby reducing unnecessary deceleration and waiting and improving overall merging efficiency.

[0048] The above process can be repeated for different slot positions and different order adjustment schemes. All candidate traffic order schemes are comprehensively evaluated through the objective function, and then input into the S222 optimal merging vehicle sequence solution process. The merging sequence that makes the most reasonable use of longitudinal space and has the highest overall traffic efficiency is selected as the output result.

[0049] S222, Solving for the optimal merging vehicle sequence Since the merging order optimization problem on virtual lanes is a permutation-based combinatorial optimization, when the number of virtual vehicles in the initial merging vehicle sequence is small (e.g., the total number of vehicles...), the optimization is less problematic. The exhaustive method can be used to enumerate all possible merging sequences, calculate the objective function value, and select the scheme with the smallest deviation as the global optimal solution. For scenarios with a large number of convoys and a large number of permutations and combinations, in order to balance the solution speed and solution quality, this invention can use heuristic combinatorial optimization algorithms such as Genetic Algorithm (GA), Simulated Annealing, or Non-Dominated Sorting Genetic Algorithm (NSGA-II) to search for the merging sequence. Preferably, the NSGA-II algorithm is used to search for an approximate Pareto optimal solution. This algorithm uses the initial merging convoy sequence as a starting point. As a chromosome encoding, diverse solution sets are generated through crossover and mutation, and the scheme with the minimum overall cost is finally selected as the optimal merging vehicle sequence. It is then output to the subsequent collaborative connection and trajectory planning module.

[0050] S223, Optimal Merging Team Sequence Team Integration Based on the optimal merging vehicle sequence This determines the cooperative following relationship of each virtual fleet in physical space, as well as the corresponding ideal position and speed trajectory.

[0051] Suppose that the optimal merging vehicle sequence is... In the middle, the first The virtual fleet is the fleet Its preceding position (the first) The virtual fleet (position) is the fleet In physical space, these two adjacent car convoys and The actual longitudinal spacing between them is defined as: ( ) in, For a moment Team The position of the lead vehicle. For a moment Team The position of the lead vehicle. For the team The total length.

[0052] Corresponding target longitudinal spacing The selection logic is: if the team and the team If they are in the same lane, then (Expected following distance); if in different lanes, then (Expected merging distance).

[0053] when This indicates that there is still usable redundancy between the two teams. Considered an independent lead vehicle fleet, its CAV lead vehicle is part of the fleet. Leading car speed For reference, coordinated deceleration or acceleration is performed according to FVD following constraints and safety distance constraints, so that the actual speed gradually approaches the reference speed, so that the two teams form a stable serial running relationship, while maintaining their respective internal formation structure unchanged; when This indicates that the distance between the two convoys is insufficient to maintain independent operation. As a team to be merged into, its lead CAV gradually moved towards the front of the team under safety constraints. By slightly adjusting acceleration and deceleration, the longitudinal spacing is brought closer to the target longitudinal spacing, i.e., the ideal following distance or the ideal merging distance. The HDVs inside the convoy then passively respond according to the FVD model, achieving a compact formation of the entire virtual convoy within the cooperative control area.

[0054] To facilitate trajectory planning in step three, it is necessary to determine the ideal motion state of each virtual vehicle fleet during the collaborative connection process. This ideal position... and ideal speed It is not a fixed set value, but rather based on the optimal merging sequence. The determined fleet topology following relationship is used to calculate the recursive expected value based on the motion state and safety threshold of the preceding virtual fleet.

[0055] From the above process of merging fleets into the optimal fleet sequence, it can be seen that the fleet... (No. (through the merging zone) The target motion state at any given time is obtained recursively using the following formula: = = ( ) ( ) in, and For a moment Team The actual speed and position of the lead vehicle For the team Total length, The ideal target longitudinal spacing selected based on the relationship between the two lanes .

[0056] Introducing error tolerance parameters (Position error tolerance) and (Speed ​​error tolerance) is used to determine the speed of the vehicle fleet. With its preceding convoy Whether the coordination and connection between them has been completed. Only if the convoy... actual location Compared with actual speed The following conditions must be met for a team to be considered a valid team. With the previous team The coordination and integration between them has been completed: ( ) ( ) Step 3: For the optimal merging convoy sequence, construct a multi-objective optimization model with the goal of minimizing the position deviation, speed deviation, and instantaneous fuel consumption rate of the leading vehicle of any virtual convoy, and solve for the acceleration information of the leading vehicle of the corresponding virtual convoy.

[0057] The optimal merging fleet sequence Z after fleet integration * Based on the target distance, a smooth, feasible, and fuel-efficient speed-acceleration trajectory is generated for the lead vehicle of each virtual convoy. This ensures that the convoy can accurately execute the optimal merging order, allowing its actual operating state to approach the ideal state given by the strategy layer as closely as possible to safely pass through the merging zone. Specifically: S31. Constructing a multi-objective function Set up a convoy The actual position and velocity are respectively , Based on the recursive calculation theory defined in S223, the corresponding ideal position and ideal velocity are obtained as follows: , Trajectory optimization aims to determine how to execute previous decisions at the lowest cost, i.e., to minimize the deviation between the actual state and the desired state while satisfying a series of constraints such as vehicle dynamics, safe distance, and fuel consumption.

[0058] The formula for calculating the corresponding function of the multi-objective optimization model is as follows: (twenty four) in, To predict the time step in the time domain, Indexing the fleet, This is a time discount factor used to suppress uncertainty in long-term forecasts. These are the loss coefficients, used to balance the relative weights of position and velocity deviations in the overall target. For position and velocity error tolerances, used to normalize deviations of different dimensions, weighting coefficients. These correspond to efficiency and fuel economy, respectively, to meet the requirements. , , ; This represents the prediction time domain length, i.e., the number of prediction steps considered in trajectory optimization, used to describe the look-ahead range of the control strategy for future merging processes. Its value is determined by comprehensively considering control performance and computational complexity.

[0059] To ensure that the weights of each item in the multi-objective function can balance operational efficiency and fuel economy, this invention conducts orthogonal experiments under medium flow scenarios to adjust the parameters. and discount factor The sensitivity of the parameters was analyzed, and the set of parameters that yielded the best trade-off between total driving time and fuel consumption was selected as the preferred configuration.

[0060] The first term of the multi-objective function is the driving state deviation term, which quantifies the difference between the actual state and the ideal state. The second term is the fuel economy term, which quantifies the total fuel consumption in the prediction time domain. for The instantaneous fuel consumption rate at any given moment is closely related to the vehicle's driving conditions, especially propulsion power and acceleration. To accurately describe this relationship, this invention calculates the instantaneous fuel consumption rate based on the Akcelik fuel model. .

[0061] To ensure that the planned trajectory is physically feasible, provides a comfortable driving experience, and is safe to operate, the above optimization process must meet a series of strict constraints, including the following: At any time convoy speed Vehicles must be kept between the legally mandated minimum and maximum speed limits to avoid excessive deceleration in the merging zone, which could cause localized congestion.

[0062] ( ) At any time convoy acceleration The acceleration or deceleration must be kept within the limits allowed by driving comfort and vehicle power performance to avoid sudden acceleration or deceleration. ( ) in, These are the maximum deceleration and maximum acceleration, respectively, used to limit the longitudinal acceleration and deceleration intensity of the vehicle to avoid sudden braking and acceleration behaviors that result in poor ride comfort or are prone to rear-end collisions. If the throttle and brake signals are explicitly modeled as control inputs Then there is That is, by modeling the vehicle's dynamics, the acceleration input of the lead vehicle can be controlled. Controlling the acceleration of the convoy .

[0063] By solving this constrained multi-objective optimization problem, an optimal control sequence consisting of a series of acceleration commands can be generated for each virtual fleet's CAV navigator, guiding its driving operations in subsequent time periods.

[0064] S32. A multi-objective evolutionary algorithm is used to solve the trajectory optimization problem.

[0065] Based on the aforementioned objective function and constraints, this invention preferably employs the non-dominated sorting genetic algorithm NSGA-II, using the acceleration sequence of the CAV leader vehicle in the prediction time domain as the decision variable, and searching for an approximate Pareto optimal solution set in the feasible solution space. In each generation of the population, the algorithm performs rapid non-dominated sorting of individuals based on the objective function value and constraint feasibility, and maintains solution set diversity using crowding distance; new candidate solutions are generated through crossover and mutation operations, iterating until the Pareto front change is less than a preset threshold or the maximum number of generations is reached. Finally, based on pre-selected weights or "inflection point" principles, a compromise solution is selected from the Pareto front as the optimal acceleration trajectory for each convoy leader vehicle, and the acceleration-velocity reference curve for the most recent period is sent down to the vehicle execution layer.

[0066] Step 4: Based on the longitudinal dynamics model of the virtual convoy, control the corresponding virtual convoy to complete the ramp merging according to the optimal merging sequence using the acceleration information of each leading vehicle.

[0067] S41. Control Command Issuance and Execution. After completing the trajectory optimization in step three, the central coordinator transmits the optimal merging vehicle sequence via the V2X communication unit. The corresponding acceleration-speed trajectory of the CAV lead vehicle is sent to each convoy. For CAV vehicles, the vehicle control module directly adjusts the longitudinal acceleration based on the received reference trajectory to achieve automatic tracking; for HDV vehicles, the variable message display (VMS) unit prompts the driver with instructions such as "follow the convoy, maintain distance, prepare to merge" to assist the driver in completing the coordinated merging according to the predetermined formation.

[0068] S42. Rolling Monitoring and Iterative Updates. This invention uses a 1-second control cycle to collect the actual position, speed, and spacing information of the convoy in real time, and compares it with the planned trajectory obtained in step three. When the position error of any convoy's leading vehicle exceeds 5 meters or the speed error exceeds 5 km / h, it is considered that the current trajectory deviation has exceeded the allowable range. The central coordinator restarts the virtual lane mapping and merging order optimization process, generates a new merging order and trajectory, and distributes it to each convoy to ensure the robustness of the control effect.

[0069] S43. Control Flow Termination. When all virtual convoys pass through the merging zone MZ and exit the cooperative control zone CZ, the current batch cooperative merging control flow terminates. After detecting a new vehicle arriving at the formation zone PZ, the system restarts from step one, achieving periodic batch control of continuous traffic flow.

[0070] To verify the feasibility of this invention, we conducted the following experiment: This embodiment takes the merging area of ​​a single-lane main lane and a ramp on a highway as the research object, such as... Figure 2 As shown. The main lanes are 3500 m long, and the ramps are 1500 m long. The control area is divided into three zones along the driving direction: the formation zone (PZ), the cooperative control zone (CZ), and the merging zone (MZ). The formation zone (PZ) is located approximately 1500 m upstream of the main lanes and ramps, and is used to identify vehicle types, collect initial states, and form a "1+n" virtual convoy. The cooperative control zone (CZ) is 1000 m long and located downstream of the PZ, and is used to perform virtual lane mapping, merging order optimization, and cooperative connection. The merging zone (MZ) is 200 m long and is used to describe the geometric merging area of ​​the main lanes and ramps.

[0071] Traffic flow and vehicle status detection units are deployed within the PZ and CZ, including lane flow, density, and average speed detection equipment, as well as roadside units (RSUs) supporting DSRC or C-V2X communication. The detection units acquire information such as flow, density, and average speed for each lane with a 1-second sampling period. The CAVs periodically report their precise location, speed, and acceleration via V2X communication. Virtual fleet control units (VFCUs) are deployed on roadside or cloud servers, receiving data from the detection units and CAVs, and performing functions such as virtual fleet formation, virtual lane mapping, merging sequence optimization, collaborative connection, and speed trajectory planning. The communication unit is responsible for issuing control commands to the vehicle execution units. The CAVs execute acceleration and speed commands directly through the onboard controller, while the HDVs provide recommended speeds and merging opportunities to the driver via a variable message sign (VMS).

[0072] To verify the applicability of the control method of the present invention under different traffic loads, this embodiment sets up three total demand scenarios: low demand, medium demand, and high demand, corresponding to traffic flows of 1400veh / h, 1800veh / h, and 2200veh / h, respectively. Here, 1400 / 1800 / 2200 represents the one-way traffic flow of the main lanes, corresponding to total network demands of 2800 / 3600 / 4400. Under each total demand scenario, three demand allocation combinations for main lanes and ramps are further considered: 50:50, 65:35, and 80:20. For example, in the medium demand scenario, a 50:50 allocation corresponds to 1800veh / h for both main lanes and ramps; a 65:35 allocation corresponds to 2340veh / h for main lanes and 1260veh / h for ramps. Vehicle types include CAV and HDV, with a CAV penetration rate of 50% and an HDV share of 50%. Parameters such as vehicle length, mass, maximum speed, and maximum acceleration / deceleration are shown in the table below.

[0073] Based on the above scenario and parameter settings, the specific execution flow of the method of the present invention during the process of a batch of vehicles passing through the control zone is described, and a set of numerical examples at typical moments are given, taking a 50:50 demand allocation scenario as an example. The entire control flow of the control method of the present invention is executed in a rolling manner with a control cycle of 1 second.

[0074] In this embodiment, the longitudinal motion of all virtual convoys in the main lanes and ramps satisfies the 1+n virtual convoy dynamics model. The longitudinal acceleration of the CAV leader in each control cycle is given by the virtual convoy control unit, and the HDV members automatically adjust their speed and distance according to the FVD car-following model.

[0075] Step 1: Create a virtual fleet During a certain control cycle, the detectors measured an average speed of 80 km / h and a density of 18.75 veh / km on the main lane, and an average speed of 70 km / h and a density of 17 veh / km on the ramp. The control unit processes the raw data system as input.

[0076] Step 2: Optimize the merging order based on virtual mapping For ease of understanding, let's assume the initial virtual convoy sequence is as follows: in, Main line convoy For ramp convoys, the corresponding lane markings are: In this initial sequence, the second slot is formed between vehicle platoons P1 and P2. Based on the aforementioned virtual lane mapping and longitudinal space discrimination model, the corresponding available longitudinal gap is: In the aforementioned initial virtual queue In the middle, the convoy passes through the merging point alternately in the manner of "mainline-ramp-mainline-ramp". It is possible to consider allowing the mainline convoy P3, which was originally behind the lane, to pass earlier, forming a "mainline-mainline-ramp-ramp" pattern, thus adjusting the traffic order to candidate scheme A. This adjustment refers to changing the order in which convoys arrive at the merging point near the entrance of the merging zone through longitudinal coordinated control, so that the convoys... Prior to the convoy in virtual order Through the merging point. This process does not involve physical cutting in or overtaking in the middle of the road, but rather establishes a new traffic order near the entrance to the merging zone through longitudinal speed regulation.

[0077] When the passage order is adjusted to candidate scheme A: convoy Located in slot 2 of the original initial sequence, its preceding and following convoys are respectively and That is, at this time ,but It means "in order to make the convoy follow the instructions" – – – The sequence passes through the merging point, where originally... and "The formed slots must reserve at least the total longitudinal space," which includes the fleet space. Forward safety clearance + platoon length + rearward safety clearance: When the passage order is adjusted to candidate plan A: convoy ,at this time = If satisfied Then the convoy As an independent team, they passed through the merging point; Under the condition Below, with the objective of minimizing the deviation between the actual longitudinal spacing and the ideal target longitudinal spacing, an objective function is constructed: Then, the optimal merging order is obtained by comprehensively evaluating all candidate passage orders through the objective function. Then, guided by the fleet integration of the optimal merging fleet sequence of S223, the coordinated following relationship is determined.

[0078] Step 3: In the above optimal merging vehicle sequence Based on this, a multi-objective optimization function is constructed and solved based on position, speed and fuel consumption to obtain the CAV acceleration signal of the lead vehicle of the virtual convoy, acquire multi-objective trajectory planning information, and then control the entire convoy to better complete the merging.

[0079] Step 4: After multi-objective trajectory planning, the virtual fleet control unit transmits the optimal merging fleet sequence via the V2X communication unit. The reference acceleration-velocity trajectory of each convoy's lead vehicle is sent to the vehicle's execution unit.

[0080] To verify the comprehensive effectiveness of the method of the present invention in terms of traffic efficiency, fuel economy, and ramp queue length, this embodiment sets up three control strategies for comparison on the simulation platform of Embodiment 1: Strategy A is a no-control strategy, where vehicles operate autonomously based solely on the FVD car-following model; Strategy B is a single-vehicle cooperative control strategy, which only guides the speed of individual CAVs without considering the virtual fleet structure; Strategy C is the cooperative merging control strategy proposed in this invention based on the mapping of "1+n" virtual fleets and virtual lanes.

[0081] Taking a medium-demand 50:50 allocation scenario as an example, under the same simulation duration and random seed conditions, simulation results show that compared with the uncontrolled strategy A, strategy C of this invention can significantly shorten the total vehicle travel time, increase the average speed of the main lane and ramps, and effectively suppress the growth of ramp queue length. In low-demand and medium-demand scenarios, the fuel consumption per unit mileage of strategy C of this invention is significantly lower than that of strategies A and B, and can improve traffic efficiency while taking into account fuel economy. In high-demand scenarios, although strategy C of this invention sacrifices some fuel economy locally to maintain continuous traffic flow in the bottleneck area, it can still avoid severe queuing and traffic collapse, and ensure the stable operation of highway ramp merging areas under high-load mixed traffic conditions.

[0082] Simulations were conducted using SUMO software to compare the performance of the proposed strategy C with that of a controlless strategy A and a single-vehicle cooperative control strategy B. Experimental data under a 50:50 demand allocation model visually demonstrates the advantages of the proposed strategy C in terms of efficiency, energy consumption, and safety, as shown in the table below. Figure 4-5 : Figure 4 The simulation results show that the vehicle spatiotemporal trajectory diagram at a traffic flow of 2200 vehicles per hour illustrates a high-demand scenario, exceeding the theoretical maximum capacity of highway ramps. Under such dense traffic pressure, simulation results indicate that due to the extremely high traffic density and significantly shortened headway, the convoy begins to slow down in advance to ensure safe merging and reserve sufficient merging space. Despite the extremely high traffic volume, strategies B and C successfully prevented traffic collapse, with vehicle deceleration remaining within acceptable limits, effectively preventing further deterioration of traffic congestion. The simulation results confirm that the proposed ramp merging cooperative control algorithm can significantly improve ramp efficiency and alleviate congestion even under severe congestion conditions, fully demonstrating its strong adaptability and robust control performance under high-demand scenarios.

[0083] Figure 5 The chart comparing fuel consumption under different traffic demands reveals significant differences in fuel efficiency among the three strategies. In Strategy A, fuel consumption per unit distance decreases continuously as traffic volume increases. Specifically, fuel consumption is 90.07 ml / km at low traffic volume, drops to 87.90 ml / km at medium traffic volume, and further decreases to 86.62 ml / km at high traffic volume. This phenomenon is mainly attributed to the reduced vehicle spacing during high traffic volume, effectively shortening ramp waiting times and increasing the average driving speed of the entire network, thereby reducing fuel consumption. However, the improvement in fuel economy is limited by safety distance requirements and will eventually reach a critical value.

[0084] In summary, this invention effectively improves the efficiency, safety, and fuel economy of mixed-flow highway ramp merging by quantifying model parameters, introducing nonlinear safety constraints, and adapting to multiple scenarios, and has good practical application value.

[0085] It is important to note that the schemes and arrangements of this application shown in the exemplary embodiments are merely exemplary. Although only a few embodiments are described in detail in this disclosure, those who consult this disclosure will readily understand that many modifications are possible (e.g., variations in various parameter values ​​(temperature, power, humidity, etc.), installation arrangements, names, colors, logical orders, etc.) without substantially departing from the novel teachings and advantages of the subject matter described in this application. Therefore, all such modifications are also included within the scope of the invention, and the order or sequence of any process or method steps may be changed or rearranged according to alternative embodiments. In the claims, any "apparatus plus function" clause is intended to cover the structure described herein for performing the function, and not only structurally equivalent but also equivalent in structure. Other substitutions, modifications, alterations, and omissions may be made in the design, operation, and arrangement of the exemplary embodiments without departing from the scope of the invention. Therefore, the invention is not limited to the particular embodiments but extends to a variety of modifications that still fall within the scope of the appended claims.

[0086] Furthermore, in order to provide a concise description of exemplary embodiments, not all features of actual embodiments (i.e., those features that are not relevant to the best mode of carrying out the invention as currently considered, or those features that are not relevant to implementing the invention) may be omitted.

[0087] It should be understood that numerous specific implementation decisions can be made during the development of any practical implementation, such as in any engineering or design project. Such development efforts may be complex and time-consuming, but for those skilled in the art who benefit from this disclosure, the development effort will be a routine work of design, manufacturing, and production without requiring much experimentation.

[0088] 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 technical solutions 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 method for cooperative merging control of hybrid traffic ramps based on virtual vehicle fleets, characterized in that... Includes the following steps: Step 1: Dynamically assemble multiple virtual convoys on the main lanes and ramps of the highway, each led by a networked autonomous vehicle and followed by multiple manually driven vehicles. Step 2: Map the virtual convoys on the ramps onto the main lanes to form an initial merging convoy sequence. Assume that adjusting the passage order of some virtual convoys in the initial merging convoy sequence forms a new passage sequence scheme as a candidate passage scheme. Construct a target model with the goal of minimizing the total deviation between the target longitudinal spacing required for the virtual convoys to be adjusted in any candidate passage scheme and the actual longitudinal spacing between two adjacent virtual convoys in the initial merging convoy sequence, and solve for the optimal merging convoy sequence. Step 3: For the optimal merging convoy sequence, construct a multi-objective optimization model with the goal of minimizing the position deviation, speed deviation, and instantaneous fuel consumption rate of the leading vehicle of any virtual convoy, and solve for the acceleration information of the leading vehicle of the corresponding virtual convoy. Step 4: Based on the longitudinal dynamics model of the virtual convoy, control the corresponding virtual convoy to complete the ramp merging according to the optimal merging sequence using the acceleration information of each leading vehicle.

2. The method for cooperative merging control of mixed traffic ramps based on virtual vehicle fleets according to claim 1, characterized in that: In step two, the initial merging vehicle sequence is recorded. The target longitudinal spacing between adjacent virtual vehicle fleets in the same lane is That is, the desired following distance, which is the target longitudinal spacing between adjacent virtual convoys in different lanes. That is, the expected merging distance. First, use the following formula to calculate the actual longitudinal distance between two adjacent virtual vehicle convoys. Longitudinal distance from the target , in, For virtual fleets The position of the lead vehicle, For virtual fleets The total length; in, Lane identifiers represent virtual convoys; convoys in the main lane are assigned a value of 1, while convoys in the ramps are assigned a value of 0. This indicates the index of the virtual vehicle fleet within the initial merging vehicle sequence. Indicates being in the first The sequential index of a virtual car fleet following a virtual car fleet. This represents the virtual vehicle fleet that needs to be adjusted in any candidate traffic scheme. Required target longitudinal spacing; Introducing binary decision variables The definition is as follows: ， Reconstruct the target model The feasibility constraints are: Then, the non-dominated sorting genetic algorithm NSGA-II is used to search for an approximate Pareto optimal solution, and the scheme with the minimum overall cost is selected as the optimal merging vehicle sequence.

3. The method for cooperative merging control of mixed traffic ramps based on virtual vehicle fleets according to claim 2, characterized in that: Let any two adjacent virtual convoys in the optimal merging convoy sequence be denoted as convoys. convoy And the convoy Located in the team Ahead of them, the longitudinal distance between the two convoys is... ,in, , The actual longitudinal distance between the two convoys can be calculated using the following formula. , in, Indicates the team The leading vehicle in the middle of the time Location, Indicates the team The leading vehicle in the middle of the time Location, Indicates the team The total length; like If so, then the two teams will remain as they are. like Then the convoy merged into the team They are merged into a single virtual fleet, which controls the fleet. The lead vehicle gradually moved towards the convoy They move closer together until the actual longitudinal distance between them reaches the target longitudinal distance.

4. The method for cooperative merging control of mixed traffic ramps based on virtual vehicle fleets according to claim 2, characterized in that: The distances from the lead vehicle of each virtual convoy on the main lane and ramp to the merging point are arranged in ascending order to map the virtual convoys on the ramp onto the main lane, forming the initial merging convoy sequence. .

5. The method for cooperative merging control of mixed traffic ramps based on virtual vehicle fleets according to claim 1, characterized in that: In step three, a multi-objective optimization model is first constructed using the following formula. in, Indicates the length of the prediction time domain. To predict the time step in the time domain, Indexing the fleet, This is a time discount factor used to suppress uncertainty in long-term forecasts. These are the loss coefficients, used to balance the relative weights of position and velocity deviations in the overall target. For position and velocity error tolerances, used to normalize deviations of different dimensions, weighting coefficients. These correspond to efficiency and fuel economy, respectively, to meet the requirements. , ,and , Indicates instantaneous fuel consumption rate. , They respectively represent the position of the first The actual position and speed of the virtual convoy. , They represent the first The ideal position and ideal speed of the virtual convoy passing through the area are determined by the optimal merging convoy sequence. The determined platoon topology following relationship is obtained by recursive calculation based on the motion state and safety threshold of the preceding virtual platoon; The non-dominated sorting genetic algorithm NSGA-II is then used to solve the multi-objective optimization model, using the acceleration sequence of the leading vehicle in the prediction time domain as the decision variable, and searching for an approximate Pareto optimal solution set in the feasible solution space.

6. The method for cooperative merging control of mixed traffic ramps based on virtual vehicle fleets according to claim 1, characterized in that: In the process of forming a virtual convoy in step one, for any vehicle entering the formation zone PZ... First, calculate its relationship with the preceding virtual convoy. The last car Initial spacing between , in, Indicates vehicle Time of arrival at formation zone PZ and They represent The last car of the virtual convoy before the timeline and vehicles Location, For vehicles Length; Define minimum safe merge distance : in, , These represent the preceding virtual convoys. Nakao car and vehicles exist The speed of time, Indicates the preceding virtual convoy Nakao car Distance to the end of the formation zone (PZ) For maximum comfort acceleration, Maximum speed; Then, the rules for forming the organization are established; When the vehicle When it is HDV, if Then the vehicle Join the preceding virtual convoy; otherwise, remain as a standalone HDV. When the vehicle When it is CAV, if Then the vehicle As the lead vehicle of the new virtual fleet, it is used to form the virtual fleet; if Within the longest platoon length behind it, there are multiple HDVs, and the preceding virtual platoon... The maximum convoy length has been reached; move the vehicles... As the lead vehicle of the new virtual fleet; otherwise, join the preceding virtual fleet. As a member.

7. The method for cooperative merging control of mixed traffic ramps based on virtual vehicle fleets according to claim 1, characterized in that: The main lanes and ramps are divided into three zones from upstream to the merging point: formation zone PZ, cooperative control zone CZ, and merging zone MZ. Virtual fleet formation is completed in formation zone PZ, and calculations for steps two and three are completed in cooperative control zone CZ.

Images