Speed ​​Coordination and Collaborative Merging Control Methods for Intelligent Connected Dedicated Lane Scenarios

Abstract

This invention discloses a speed coordination and collaborative merging control method for intelligent connected dedicated lane scenarios. The method includes: establishing a combined control system; acquiring traffic flow data such as density and flow rate of upstream road segments, ramp segments, and bottleneck segments using traffic data acquisition equipment; deploying traffic signal dissemination equipment; issuing different speed limit and lane-changing combination control commands for dedicated lanes and general-purpose lanes; implementing a decision-making control agent based on a deep reinforcement learning algorithm; receiving traffic status from the data acquisition equipment; continuously iterating and training in training mode to generate the optimal control strategy; transitioning to execution mode; and outputting the optimal control command to the signal dissemination equipment. This invention is applicable to complex traffic flow scenarios with intelligent connected dedicated lanes, minimizing the negative effects of mixed traffic flow in dedicated lane scenarios while ensuring optimal control performance for each strategy.

Description

Technical Field

[0001] This invention belongs to the field of intelligent transportation and traffic control technology, specifically relating to a speed coordination and collaborative merging control method for intelligent connected dedicated lane scenarios. Background Technology

[0002] To address vehicle interference issues in intelligent connected mixed traffic flows, dedicated intelligent connected vehicle lanes can improve road capacity by creating homogeneous traffic flow through temporal or spatial separation. However, dedicated lane management strategies cannot completely eliminate vehicle interference. When intelligent connected vehicles changing lanes from general-purpose lanes into dedicated lanes, they obstruct the normal driving of vehicles in adjacent lanes and also disturb convoys traveling at high speeds in the dedicated lanes. Therefore, designing and developing control strategies such as speed coordination and cooperative merging in this scenario is of great significance in further reducing vehicle interaction.

[0003] Speed ​​coordination strategies can smooth out vehicle speed differences longitudinally. Based on real-time traffic data, they can determine traffic conditions and dynamically adjust the speed limits or recommended speeds on highways, creating gaps to prevent the propagation of merging bottlenecks and replacing fixed bottlenecks with moving ones. Coordinated merging strategies can coordinate vehicle movement laterally, guiding vehicles in the rightmost lane to pre-change into the left lane, ensuring that there is usable merging space on the main line.

[0004] Currently, the combined application of dedicated lane management with other control strategies is rarely considered. Various control strategies mainly exist independently, ignoring the potential changes that other strategies may cause when considering traffic conditions. In reality, different strategies may interfere with each other, thus affecting the implementation effect. Summary of the Invention

[0005] Purpose of the invention: To address the above problems, this invention proposes a speed coordination and collaborative merging combined control method for intelligent connected dedicated lane scenarios. This method ensures smooth and efficient vehicle movement in both longitudinal and lateral directions, maximizes traffic flow efficiency, minimizes the negative effects of mixed traffic flow in dedicated lane scenarios, and ensures that each strategy achieves optimal control performance.

[0006] Technical solution: To achieve the purpose of this invention, the technical solution adopted by this invention is: a speed coordination and collaborative merging control method for intelligent connected dedicated lane scenarios, comprising the following steps:

[0007] Step 1: Establish a speed coordination and collaborative merging control system for intelligent connected dedicated lane scenarios, including: traffic data acquisition equipment, traffic signal dissemination equipment, decision control intelligent agent, and vehicles;

[0008] Step 2: Deploy traffic data acquisition devices along the highway at preset intervals to obtain traffic flow data at specific detection sections and set the update frequency;

[0009] Step 3: Set up a control zone upstream of the bottleneck section and deploy variable signal signs and roadside signal units as traffic signal dissemination equipment. The variable signal signs will issue different speed limit instructions, and the roadside signal units will issue lane change instructions. Set the control cycle and control section spacing.

[0010] Step 4: Implement a decision-making and control agent based on a deep reinforcement learning network. The decision-making and control agent has a training mode and an execution mode, and receives traffic status data from the data acquisition device. t If the current mode is training, proceed to step 5; if the current mode is execution, proceed to step 6.

[0011] Step 5: In training mode, set a probability threshold ε based on the current number of training iterations. The decision-making control agent has a probability ε of taking a random action a. t There is a probability of 1-ε of taking the optimal action a, which yields the highest future average cumulative reward, i.e., the Q value. t The signal is sent to the traffic signal display equipment, converted into corresponding control commands, and transmitted to the controlled vehicles, recording the status during the interaction process. t Action a t The rewards obtained are stored in the experience base, and the agent is iteratively trained until the corresponding threshold conditions are met, at which point it enters the execution mode.

[0012] Step 6: In execution mode, the decision-making control agent directly selects the optimal action a, which has the highest average cumulative reward in the future. t The signal is sent to the traffic signal broadcasting equipment, converted into corresponding control commands, and then transmitted to the controlled vehicles.

[0013] Furthermore, in step 2, the collected traffic flow data includes upstream flow rate of the bottleneck, upstream density of the bottleneck, ramp flow rate, ramp density, bottleneck density, and traffic flow unevenness coefficient of the bottleneck section.

[0014] Furthermore, in step 3, the speed limit instruction is a multiple of 10 between the maximum and minimum speeds; the lane change instruction includes two items: normal driving and changing lanes to the left.

[0015] Furthermore, in step 5, the training process of the decision-making control agent includes the following steps:

[0016] Step 5.1: Initialize the experience base and set the sample size; initialize the deep learning neural network and establish an online network and a target network with the same structure. The online network continuously interacts with the environment to update the network parameters, while the target network is used to output the optimal action that will obtain the highest average cumulative reward in the future.

[0017] Step 5.2: The decision control agent receives the traffic status, inputs it into the network to obtain the future average cumulative reward for different actions, i.e., the Q value, takes an action randomly with a probability of ε, and takes the action with the maximum Q value with a probability of 1-ε.

[0018] Step 5.3: Convert the actions in Step 5.2 into corresponding control commands and issue them through the traffic signal equipment. After one control cycle, calculate the reward value R of the previous control cycle t. t :

[0019]

[0020] In the formula, k b For the bottleneck density, k c For critical density, k g Congestion density;

[0021] Step 5.4: The control cycle ends, and the agent receives the traffic state s for the next control cycle t+1. t+1 The sequence of interactive information within the control cycle (s) t ,a t ,R t ,s t+1 Store it in the experience base and denoted as sample. t ;

[0022] Step 5.5: Extract k batches of empirical samples from the experience base, and define the loss function as follows:

[0023]

[0024] Where θ represents the online network parameters, Q(s) t ,a t ;θ) represents the state s during the control period t. t Action is a t At that time, the state Q value, Y, is calculated by the online network. t The target value is expressed by the following formula:

[0025]

[0026] Where R t+1 Q(s) represents the reward value obtained during control period t+1, where γ is the discount factor; t+1 ,a t ;θ) represents the state s within the control period t+1. t+1 Action is a t At that time, the Q value obtained by online network calculation; θ - For the target network parameters, Indicates that in state st+1 The online network selects the action 'a' with the largest Q value based on parameter θ; This indicates that based on the action 'a' selected above, in state s... t+1 Below, the Q value calculated by the target network;

[0027] Step 5.6: Assign weights to different empirical samples, calculated using the following formula:

[0028]

[0029] Where ω t For sample t The weight values ​​are N, where N is the total number of samples in the experience base, and β and π are constants; P t Indicates sample t The probability of being selected, p t Indicates sample t The priority, where α is the priority weight, is used to correct p. t ; This represents the sum of the priorities of all samples in batch k after correction;

[0030] Step 5.7: Update the online network parameters according to the gradient descent direction, as shown in the following formula:

[0031]

[0032] Where θ represents the online network parameters, θ' represents the updated online network parameters, and κ is the learning rate. This represents the gradient value of the loss function. Indicates the online network in Q(s) t ,a t The gradient value at θ);

[0033] Step 5.8: When the preset control cycle number is reached, copy the online network parameter θ to the target network parameter θ. - ;

[0034] Step 5.9: Repeat steps 5.2 to 5.8 until the set stopping condition is met.

[0035] Furthermore, based on the existing reward R t Additional reward / penalty values ​​are set on top of the existing values, when the bottleneck density k b With key density k c Satisfying k c -2 <k b <k c +2, the reward value increases by an additional 1;

[0036] When the bottleneck density k b Greater than congestion density k gWhen the upstream density k is [a certain value], the reward value is reduced by an additional 1; m Less than the critical density k c If any speed limit or lane change order is issued, the bonus value will be reduced by an additional 0.5.

[0037] Furthermore, in step 6, the action selection strategy formula for the agent in the execution mode is as follows:

[0038]

[0039] Where π(a|s) represents the policy function, a represents the action, s represents the state, and Q represents the policy function. π (s,a) represents the Q-value calculated by the online network based on action a in state s, γ represents the discount factor, and Rs t This represents the reward value at control period t. This represents the average cumulative reward earned starting from the first control period.

[0040] Furthermore, the traffic flow unevenness coefficient of the bottleneck section is used to measure the utilization rate of different lanes. The formula for calculating the unevenness coefficient is as follows:

[0041]

[0042] Among them, CV b The variable represents the traffic flow unevenness coefficient at the bottleneck section, where L represents the number of lanes and k represents the number of lanes. b (n) represents the density of the nth lane from the inside of the merging zone, k b This indicates the average lane density within the merging zone.

[0043] Furthermore, different speed limit instructions are issued for different types of lanes. For intelligent connected vehicle lanes, the issued speed is the recommended speed for intelligent connected vehicles; for general lanes, the issued speed is the maximum speed limit for vehicles.

[0044] Furthermore, different lane-changing instructions are issued for different types of vehicles: intelligent connected vehicles are guided to change lanes to the left and enter the dedicated lane; manually driven vehicles are guided to change lanes to the left and enter the general lane.

[0045] Beneficial effects: Compared with the prior art, the technical solution of the present invention has the following beneficial technical effects:

[0046] 1. Compared to existing hybrid traffic control strategies, this invention enables the combined application of different control strategies. The speed coordination strategy smooths out vehicle speed differences longitudinally, while the coordinated merging strategy coordinates vehicle movement laterally. By designing combined control commands, the intelligent agent can avoid outputting actions that might cause mutual interference, thus ensuring that each strategy achieves optimal control performance and avoiding negative effects.

[0047] 2. It possesses the ability to update the optimal control strategy in real time. The control agent can switch freely between training mode and execution mode. In training mode, it continuously learns the optimal strategy using historical experience data obtained from interaction with the traffic environment, thus entering execution mode. At the same time, it can also collect new data based on changes in traffic flow status to achieve continuous updates of the control strategy, avoid control delays, and have strong stability.

[0048] 3. Applicable to complex traffic flow scenarios with dedicated intelligent connected lanes. Existing inventions are mainly applied to basic sections of highways. This invention, based on highways with dedicated intelligent connected lanes, uses the merging bottleneck at the entrance ramp as the key control area, enabling overall highway control and maximizing overall traffic efficiency. Attached Figure Description

[0049] Figure 1 This is a schematic diagram of the combined control system structure of the present invention.

[0050] Figure 2 This is a flowchart of the training mode of the decision control agent in this invention.

[0051] Figure 3 This is a schematic diagram of the control scenario and control equipment layout in an embodiment of the present invention. Detailed Implementation

[0052] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. It should be noted that the following examples are for illustrative purposes only and are only some embodiments of the present invention, not all embodiments.

[0053] The speed coordination and cooperative merging control method for intelligent connected dedicated lane scenarios described in this invention specifically includes the following steps:

[0054] Step 1: Establish a speed coordination and collaborative merging control system for intelligent connected dedicated lane scenarios, such as... Figure 1 As shown, it includes: traffic data acquisition equipment, traffic signal dissemination equipment, decision control intelligent agents, and vehicles.

[0055] In this example, a control strategy application scenario is built based on SUMO simulation software, such as... Figure 3 As shown in the diagram. The study scenario includes a 2km three-lane highway and a 500m single-lane entrance ramp. The speed limit on the main line is 100km / h, and the speed limit on the ramp is 40km / h. A dedicated CAV lane is provided along the inner side of the main line, and vehicles on the ramp merge into the main line via a 200m acceleration lane. The scenario is divided into three zones: a 1000m control zone, a 200m merging zone, and free-traffic zones at the beginning and end of the main line, each 400m long. The lengths of the zones in the diagram do not represent specific proportions and are for illustrative purposes only.

[0056] Step 2: Deploy traffic data collection devices along the highway at 500-meter intervals, with an update frequency of 30 seconds. The collected traffic flow data includes the upstream flow rate (d) of the bottleneck. m upstream density k of the bottleneck m Ramp flow rate d r Ramp density k r Bottleneck density k b Traffic flow unevenness coefficient (CV) on bottleneck sections b The formula for calculating the non-uniformity coefficient is as follows.

[0057]

[0058] Where L represents the number of lanes, which is 4 in this example, k b (n) represents the density of the nth lane from the inside of the merging zone. This represents the average lane density within the merging zone. Based on the above traffic data, the spatial representation of the state information is s = (d m ,d r ,k m ,k r ,k b ,CV b ).

[0059] Step 3: Establish a control zone upstream of the bottleneck section, and deploy variable signal signs and roadside signal units as traffic signal dissemination equipment. The variable signal signs will disseminate different speed limit commands, with different speed limits between a maximum speed of 100 km / h and a minimum speed of 60 km / h. The action set of the speed limit command is 'a'. SH ={60,80,100} (unit: km / h); The roadside signal equipment issues a lane-changing instruction, including two options: "normal driving" and "change lanes to the left". The action set of the lane-changing instruction is a. RM = {0, 1}, where 0 indicates no lane change guidance and 1 indicates lane change guidance. The control cycle is set to 30 seconds, and the control section spacing is set to 500 meters.

[0060] Combining the two control commands, the system outputs the following combined control commands, totaling 18 actions.

[0061]

[0062] Speed ​​limit instructions differentiate between different types of lanes. For dedicated lanes for intelligent connected vehicles, the published speed is the recommended speed for intelligent connected vehicles; for general lanes, the published speed is the maximum speed limit for vehicles.

[0063] The lane-changing instructions differentiate between different types of vehicles: for intelligent connected vehicles, they are guided to change lanes to the left and enter the dedicated lane; for manually driven vehicles, they are guided to change lanes to the left and enter the general lane.

[0064] The effectiveness of control commands will be affected by driver compliance. The compliance rate for manually driven vehicles is 0.9, while that for intelligent connected vehicles is 1.

[0065] Step 4: Implement the decision control agent based on the Deep Reinforcement Learning (DQN) algorithm. The decision control agent has a training mode and an execution mode, such as... Figure 2 As shown, the intelligent agent receives traffic status s from the data acquisition device. t If the current mode is training, proceed to step 5; if the current mode is execution, proceed to step 6.

[0066] Step 5, in training mode, the training process of the decision control agent includes the following steps:

[0067] Step 5.1: Initialize the deep learning neural network. Establish the online network and the target network. The two networks have the same structure, each containing an input layer, a hidden layer, and an output layer. The input layer has 6 neurons, the output layer has 18 neurons, and the hidden layer contains 20 neurons. The activation function is the ReLU function. Initialize the experience base with a sample size of 10,000.

[0068] Step 5.2, the decision-making and control agent receives the traffic state s t The input network obtains the Q-values ​​of different actions. The action is randomly selected with probability ε, and the action with the maximum Q-value is selected with probability 1-ε. ε is initially set to 0.9, and decreases by 0.00003 in each training session, and remains unchanged after decreasing to 0.1.

[0069] Step 5.3: Convert the actions in Step 5.2 into corresponding control commands and issue them through the traffic signal equipment. After one control cycle of 30 seconds, calculate the reward value R of the previous control cycle t. t :

[0070]

[0071] Where k b For the bottleneck density, k c For the critical density, in this example, the value is taken as 27 veh / km, k g For congestion density, a value of 45veh / km is used in this example;

[0072] To accelerate training convergence, the original reward R... t Additional reward / penalty values ​​are set on top of the existing values, when the bottleneck density k b When located near the critical density (k c-2 <k b <k c +2), the reward value increases by an additional 1; when the bottleneck density k b When greater than the congestion density (k b >k g The reward value is reduced by an additional 1 when the upstream density k m When less than the critical density (k m <k c If any speed limit or lane change order is issued, the bonus value will be reduced by an additional 0.5.

[0073] Step 5.4: The control cycle ends, and the agent receives the traffic state s for the next control cycle t+1. t+1 The sequence of interactive information within the control cycle (s) t ,a t ,R t ,s t+1 Store it in the experience base and denoted as sample. t ;

[0074] Step 5.5: Extract k batches of empirical samples from the experience base, and define the loss function as follows:

[0075]

[0076] Where θ represents the online network parameters, Q(s) t ,a t ;θ) represents the state s during the control period t. t Action is a t At that time, the state Q value, Y, is calculated by the online network. t The target value is expressed by the following formula:

[0077]

[0078] Where R t+1 This represents the reward value obtained during the control period t+1, where γ is the discount factor, and in this example, γ is 0.9; Q(s t+1 ,a t ;θ) represents the state s within the control period t+1. t+1 Action is a t At that time, the Q value obtained by online network calculation; θ - For the target network parameters, Indicates that in state s t+1 The online network selects the action 'a' with the largest Q value based on parameter θ; This indicates that based on the action 'a' selected above, in state s... t+1 Below, the Q value calculated by the target network;

[0079] Step 5.6: Assign weights to different empirical samples, calculated using the following formula:

[0080]

[0081] Where ω t For sample t The weight values ​​are N, where N is the total number of samples in the experience base, and β and π are constants; P t Indicates sample t The probability of being selected, p t Indicates sample t The priority, where α is the priority weight, is used to correct p. t ; This represents the sum of the priorities of all samples in batch k after correction; in this example, α is 0.6, β is initially 0.4, the increment rate is 0.4, it remains unchanged after increasing to 1, and π is set to 0.01.

[0082] Step 5.7: Update the online network parameters according to the gradient descent direction, as shown in the following formula:

[0083]

[0084] Where θ represents the online network parameters, θ' represents the updated online network parameters, and κ is the learning rate, which is set to 0.005 in this example. This represents the gradient value of the loss function. Indicates the online network in Q(s) t ,a t The gradient value at θ);

[0085] Step 5.8: When the preset control cycle number is reached, copy the online network parameter θ to the target network parameter θ. - ;

[0086] Step 5.9: Repeat steps 5.2 to 5.8 until the maximum number of training iterations is reached. The training mode ends and the system switches to execution mode.

[0087] Step 6: In execution mode, the decision-making control agent directly selects the optimal action a. t The formula for the optimal action selection strategy of the agent in the execution mode is as follows:

[0088]

[0089] Where π(a|s) represents the policy function, a represents the action, s represents the state, and Q represents the policy function. π (s,a) represents the Q-value calculated by the online network based on action a in state s, γ represents the discount factor, and Rs t This represents the reward value at control period t. This represents the average cumulative reward earned starting from the first control period.

Claims

1. A speed coordination and collaborative merging control method for intelligent connected dedicated lane scenarios, characterized in that, The method includes the following steps: Step 1: Establish a speed coordination and collaborative merging control system for intelligent connected dedicated lane scenarios, including: traffic data acquisition equipment, traffic signal dissemination equipment, decision control intelligent agent, and vehicles; Step 2: Deploy traffic data acquisition devices along the highway at preset intervals to obtain traffic flow data at specific detection sections and set the update frequency; Step 3: Set up a control zone upstream of the bottleneck section and deploy variable signal signs and roadside signal units as traffic signal dissemination equipment. The variable signal signs will issue different speed limit instructions, and the roadside signal units will issue lane change instructions. Set the control cycle and control section spacing. Step 4: Implement a decision-making and control agent based on a deep reinforcement learning network. The decision-making and control agent has a training mode and an execution mode, and receives traffic status data from the data acquisition device. t If the current mode is training, proceed to step 5; if the current mode is execution, proceed to step 6. Step 5: In training mode, set a probability threshold ε based on the current number of training iterations. The decision-making control agent has a probability ε of taking a random action a. t There is a probability of 1-ε of taking the optimal action a, which yields the highest future average cumulative reward, i.e., the Q value. t The signal is sent to the traffic signal display equipment, converted into corresponding control commands, and transmitted to the controlled vehicles, recording the status during the interaction process. t Action a t The rewards obtained are stored in the experience base, and the agent is iteratively trained until the corresponding threshold conditions are met, at which point it enters the execution mode. The training process of the decision-making control agent includes the following steps: Step 5.1: Initialize the experience base and set the sample size; initialize the deep learning neural network and establish an online network and a target network with the same structure. The online network continuously interacts with the environment to update the network parameters, while the target network is used to output the optimal action that will obtain the highest average cumulative reward in the future. Step 5.2: The decision control agent receives the traffic status, inputs it into the network to obtain the future average cumulative reward for different actions, i.e., the Q value, takes an action randomly with a probability of ε, and takes the action with the maximum Q value with a probability of 1-ε. Step 5.3: Convert the actions in Step 5.2 into corresponding control commands and issue them through the traffic signal equipment. After one control cycle, calculate the reward value R of the previous control cycle t. t : ， In the formula, k b For the bottleneck density, k c For critical density, k g Congestion density; Step 5.4: The control cycle ends, and the agent receives the traffic state s for the next control cycle t+1. t+1 The sequence of interactive information within the control cycle (s) t , a t , R t , s t+1 Store it in the experience base and denoted as sample. t ; Step 5.5: Extract k batches of empirical samples from the experience base, and define the loss function as follows: ， in Indicates online network parameters, This indicates that the state is within the control period t. Actions At that time, the state Q value calculated by the online network, The target value is expressed by the following formula: ， in Indicates control cycle The reward value obtained, Discount factor; Indicates during the control cycle Inside, the state is Actions The Q value obtained by online network calculation at that time; For the target network parameters, Indicates the state Online network based on parameters Choose the action 'a' with the highest Q value; This indicates that based on the action 'a' selected above, in the state... Below, the Q value calculated by the target network; Step 5.6: Assign weights to different empirical samples, calculated using the following formula: ， Where ω t For sample t The weight values ​​are given by N, where N is the total number of samples in the experience base, and β and π are constants. Indicates sample t The probability of being selected. Indicates sample t The priority, where α is the priority weight, is used to correct. ; This represents the sum of the priorities of all samples in batch k after correction; Step 5.7: Update the online network parameters according to the gradient descent direction, as shown in the following formula: ， in Indicates online network parameters, This represents the updated online network parameters, where κ is the learning rate. This represents the gradient value of the loss function. Indicates online network The gradient value at that point; Step 5.8: When the preset control cycle number is reached, the online network parameters will be... Copy to target network parameters ; Step 5.9: Repeat steps 5.2 to 5.8 until the set stopping condition is met; Step 6: In execution mode, the decision-making control agent directly selects the optimal action a, which has the highest average cumulative reward in the future. t The signal is sent to the traffic signal broadcasting equipment, converted into corresponding control commands, and then transmitted to the controlled vehicles.

2. The speed coordination and collaborative merging control method for intelligent connected dedicated lane scenarios according to claim 1, characterized in that, In step 2, the collected traffic flow data includes upstream flow rate of the bottleneck, upstream density of the bottleneck, ramp flow rate, ramp density, bottleneck density, and traffic flow unevenness coefficient of the bottleneck section.

3. The speed coordination and collaborative merging control method for intelligent connected dedicated lane scenarios according to claim 1, characterized in that, In step 3, the speed limit instruction is a different speed limit value, which is an integer multiple of 10 between the maximum speed and the minimum speed; the lane change instruction includes two items: normal driving and changing lanes to the left.

4. The speed coordination and cooperative merging control method for intelligent connected dedicated lane scenarios according to claim 1, characterized in that, In the original reward R t Additional reward / penalty values ​​are set on top of the existing values, when the bottleneck density k b With key density k c Satisfying k c -2 < k b < k c +2, the reward value increases by an additional 1; When the bottleneck density k b Greater than congestion density k g When the upstream density k is [a certain value], the reward value is reduced by an additional 1; m Less than the critical density k c If any speed limit or lane change order is issued, the bonus value will be reduced by an additional 0.

5.

5. The speed coordination and collaborative merging control method for intelligent connected dedicated lane scenarios according to claim 1, characterized in that, In step 6, the action selection strategy formula for the agent in execution mode is as follows: ， in Represents the policy function. Indicates an action, Indicates state, This represents the Q-value calculated by the online network based on action a in state s. Indicates the discount factor. This represents the reward value at control period t. This represents the average cumulative reward earned starting from the first control period.

6. The speed coordination and collaborative merging control method for intelligent connected dedicated lane scenarios according to claim 2, characterized in that, The traffic flow unevenness coefficient for bottleneck sections is used to measure the utilization rate of different lanes. The formula for calculating the unevenness coefficient is as follows: ， in The variable represents the traffic flow unevenness coefficient at the bottleneck section, where L represents the number of lanes and k represents the number of lanes. b (n) represents the density of the nth lane from the inside of the merging zone. This indicates the average lane density within the merging zone.

7. The speed coordination and cooperative merging control method for intelligent connected dedicated lane scenarios according to claim 3, characterized in that, Different speed limits are issued for different types of lanes. For dedicated lanes for intelligent connected vehicles, the issued speed is the recommended speed for intelligent connected vehicles; for general lanes, the issued speed is the maximum speed limit for vehicles.

8. The speed coordination and cooperative merging control method for intelligent connected dedicated lane scenarios according to claim 3, characterized in that, Different lane-changing instructions are issued for different types of vehicles. Intelligent connected vehicles are guided to change lanes to the left and enter the dedicated lane; manually driven vehicles are guided to change lanes to the left and enter the general lane.

Images