CPS control strategy for expressway high-flow passing efficiency optimization

Abstract

The invention belongs to the field of information physical system control, and discloses a CPS control strategy for expressway high-flow passing efficiency optimization, which comprises the following steps: S1, determining a system-level objective function by taking communication overhead, calculation overhead and cluster correlation degree required by a road section cluster structure as optimization targets; s2, solving an optimal road section cluster structure of the expressway in a current traffic state; s3, establishing five subsystem-level objective functions, and establishing a multi-objective optimization problem; s4, solving a multi-objective optimization problem to obtain an optimal control decision; s5, judging whether active control measures are started or not and whether a control decision is updated or not; s6, calculating the expected speed of the vehicle; and S7, the main road ICV of the road section travels according to the expected speed, and the ramp ICV travels according to the ramp control signal. According to the strategy provided by the invention, the network connection relationship between the roadside equipment can be dynamically adjusted, targeted information sharing and cooperative control are realized, the communication resource overhead is saved, and the distributed cooperative control of the expressway under the limited constraint of the cyber resources is realized.

Description

Technical Field

[0001] The present invention belongs to the field of cyber-physical system control, and specifically relates to a CPS control strategy for optimizing the traffic efficiency of highways with large traffic volumes. Background Art

[0002] Due to the limited road capacity, sudden large traffic flows can easily cause traffic congestion, especially on closed highways, where congestion is difficult to dissipate, which not only brings inconvenience to travel, but also causes economic losses and environmental pollution. In addition, cyber resources (computing and communication resources) are limited, so how to optimize the efficiency of highway traffic under the constraints of cyber resources has become a problem worthy of attention.

[0003] The existing research on control methods includes variable speed limit (VSL), ramp metering (RM), etc., and three types of model predictive control (MPC) are mostly used to solve the optimal control signal: centralized MPC, local MPC and fixed-structure distributed MPC. Each of the three has its own advantages, but it is difficult to ensure both control effect and real-time requirements under the constraint of limited cyber resources. Among them, centralized MPC has better control effect, but it is difficult to meet the real-time requirements under the constraint of limited computing resources; local MPC can reduce computing overhead, but due to the lack of information sharing between road sections, the control effect on the overall traffic efficiency of the highway is poor; fixed-structure distributed MPC can guarantee real-time performance, but with the increase of intelligent connected vehicles (ICV) and traffic flow, the demand for cyber resources has gradually increased, and this method has been difficult to adapt to the needs of highway traffic control under the constraint of cyber resources. In addition, the highway system presents increasingly complex distributed and networked characteristics, which not only involve the physical relationship between roads and vehicles, but also the cyber relationship between nodes such as ICV and roadside equipment and the information processing and interaction between them, forming a complex cyber-physical system (CPS). Therefore, it is necessary to further explore the problem of distributed collaborative control of highways under the constraint of limited cyber resources from the perspective of CPS. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a CPS control strategy for optimizing the traffic efficiency of highways with large traffic flows. The present invention aims to solve the problem of low traffic efficiency caused by large traffic flows under the constraints of limited computing and communication resources on various sections of highways.

[0005] The technical solution adopted by the present invention to solve its technical problem is:

[0006] A three-layer cyber-physical system architecture based on ICV includes three levels: system level, subsystem level and vehicle level. At the system level, according to the traffic density of each section of the highway, the sections are grouped by dynamically adjusting the network connection to form a section cluster, i.e., a subsystem; at the subsystem level, each subsystem triggers the optimized speed recommendation for the main road vehicles and the control signal for the ramp based on the internally shared traffic information and the traffic flow situation; at the vehicle level, based on the recommended speed, the speed guidance of the ICV on the main road is realized, and the ICV on the ramp is controlled to drive according to the ramp flow control signal.

[0007] Based on the above system architecture, the present invention provides a CPS control strategy for optimizing the traffic efficiency of large-volume expressways, including a system-level dynamic grouping strategy, a subsystem-level collaborative control strategy, and a vehicle-level speed guidance strategy. The implementation steps of the strategy are as follows:

[0008] S1. Taking the communication overhead, computational overhead and cluster correlation required by the road segment cluster structure as the optimization target, determine the system-level objective function;

[0009] S2. Use the optimal clustering algorithm to solve the optimal section cluster structure of the highway under the current traffic conditions;

[0010] S3. Based on the optimal road segment cluster structure obtained in step S2, the total driving time, total driving distance, entrance ramp queue length, speed limit mutation within the subsystem, and ramp release rate mutation are used as optimization targets, and five subsystem-level objective functions are established. The upper and lower bounds of the road segment speed limit and the upper and lower bounds of the ramp release rate are further used as constraints to establish a multi-objective optimization problem;

[0011] S4. Using a subsystem-level multi-objective optimization algorithm, solve the multi-objective optimization problem established in step S3 to obtain the optimal control decision, that is, the speed recommendation for the main road of each section and the ramp control signal;

[0012] The subsystem-level multi-objective optimization algorithm uses a non-dominated sorting evolutionary algorithm to solve the multi-objective optimization problem established in step S3, obtains a Pareto optimal solution set under the multi-objective function, and then selects an optimal control decision in the Pareto optimal solution set based on the approximate ideal solution sorting method;

[0013] S5. Design a congestion index for a road section, and use the congestion index to calculate the congestion level of each road section in the subsystem. After that, each road section determines whether to initiate active control measures and whether to update the control decision based on its own traffic status and the congestion level of the adjacent road sections in the subsystem to which it belongs;

[0014] S6. Calculate the expected speed of the vehicle based on the recommended speed of the road section and the distance between the vehicle and the vehicle in front;

[0015] S7. The ICV on the main road of the road section travels at the desired speed, while the ICV on the ramp travels according to the ramp control signal.

[0016] Further, the step S1 includes the following sub-steps:

[0017] S1.1 Calculate the communication overhead E of the system at time kT c (k);

[0018]

[0019] In the formula, C i (k) is the i-th subsystem; C(k) is the subsystem set; m represents C i (k) the mth road segment; c m The amount of data sent for the road segment m within the cluster; E elec The energy consumed for each bit of data sent or received; ε amp The energy consumption per square meter for sending each bit of data; d m,m′ is the distance between sections m and m′; Υ is the propagation attenuation index under the communication environment; constant b is the bandwidth occupied by the communication between the two sections; |C i (k)| is the number of road sections in the cluster;

[0020] S1.2 Calculate the computational cost T of the system at time kT cal ;

[0021]

[0022] Where, T evo is the evolution operation time of the optimization algorithm; t1 and t2 are constants to be fitted; M represents the total number of road sections;

[0023] S1.3 Calculate the cluster correlation R of the system at time kT d (k);

[0024]

[0025]

[0026] In the formula, ρ m,i (k) is the density of the ith interval of road segment m at time kT; is the critical density of road section m;

[0027] S1.4 According to steps S1.1 to S1.3, the system-level objective function is determined as:

[0028]

[0029] In the formula, κc , κ t Respectively represent the communication cost and computation cost weights; k t =kT / T t , T t represents the system-level control cycle; kT represents the end time of the kth road section information sampling cycle, k represents a constant, and T represents the road section information sampling cycle.

[0030] Furthermore, the optimal cluster division algorithm in step S2 is designed based on dynamic programming and includes a main algorithm and a sub-algorithm;

[0031] The main algorithm is used to preliminarily divide the sub-areas that may form the segment cluster according to the congestion level of each segment in the multi-segment area of ​​the expressway, and to call the sub-algorithm;

[0032] The sub-algorithm is used to calculate the optimal collaborative road segment clustering decision within the sub-area.

[0033] Further, the step S3 includes the following sub-steps:

[0034] S3.1 Establish five subsystem-level objective functions;

[0035] I. Establish the total driving time J of vehicles in the subsystem TTS (k c ) is the objective function of the optimization target;

[0036]

[0037] Where U m For the road segment m∈C j A control decision, including the speed limit value VSL of the main road control area m,i and the ramp release rate r of the entrance ramp control area m,i ; N p is the number of prediction time domain steps; Z = T c / T,k c = k / Z, T c is the subsystem level control cycle; w m′,i (Zk c +p) indicates that the i-th segment of segment m′ is in Zk c +p time corresponds to the queue length of the entrance ramp; p is an integer and p∈{Zk c ,Zk c +1,···,Z(k c +N p )}; represents the set of all segments of road segment m′; L m′,i represents the length of the i-th segment of the road segment m′; m′,i represents the number of lanes in the ith segment of road segment m′; ρm′,i (Zk c +p) indicates that the i-th segment of segment m′ is in Zk c + density at time p;

[0038] II. Establish a total driving distance J TTD (k c ) is the objective function of the optimization target;

[0039]

[0040] In the formula, v m′,i (Zk c +p) indicates that the i-th segment of segment m′ is in Zk c +The average speed of vehicles in the segment at time p;

[0041] III. Objective function with the entrance ramp queue length as the optimization target;

[0042]

[0043]

[0044] In the formula, J w is the number of prediction time domain steps N p Inner subsystem C j Middle entrance ramp queue length penalty; w m',i (k) is the queue length of the entrance ramp corresponding to time kT; is the maximum number of vehicles queuing at the corresponding entrance ramp;

[0045] IV. Objective function with the sudden change of speed limit value within the road section as the optimization target;

[0046]

[0047] In the formula, J ΔVSL To control the number of time domain steps N c VSL: Penalty for sudden changes in speed limit within m sections; m,i represents the speed limit value of the ith section of road section m; c is an integer and c∈{0,1,···,N c -1)};

[0048] V. Objective function with the sudden change of ramp release rate within the road section as the optimization target;

[0049]

[0050] In the formula, J Δr To control the number of time domain steps N c The penalty for the sudden change of ramp release rate in section m; r m,irepresents the ramp release rate of the ith section of road section m;

[0051] S3.2 Based on the five subsystem-level objective functions established in step S3.1, the multi-objective optimization model shown below is established with the upper and lower bounds of the speed limit and the upper and lower bounds of the ramp release rate as constraints:

[0052]

[0053] In the formula, VSL min and VSL max are the upper and lower bounds of the speed limit respectively; r min and r max are the upper and lower bounds of the ramp release rate respectively.

[0054] Further, the step S5 includes the following sub-steps:

[0055] S5.1 Calculate the congestion level of the road section;

[0056]

[0057] In the formula, MC m represents the congestion index of road section m; is the critical density of the ith interval of section m;

[0058] S5.2 Definitions They are the control triggering time sequence and the control decision updating time sequence, and the calculation expressions are as follows:

[0059]

[0060]

[0061] In the formula, θ c , σ u and θ u are all hyperparameters.

[0062] Further, in step S6, the expected speed The calculation expression is:

[0063]

[0064]

[0065] In the formula, is the distance between the vehicle and the preceding vehicle in lane l of the current road section; is the recommended speed; represents the vehicle spacing when ICVs are evenly distributed on lane l in the current road section; L vehis the average length of ICV.

[0066] Beneficial effects:

[0067] 1. The strategy proposed in the present invention can dynamically adjust the network connection relationship between roadside devices according to the traffic status of each road section, realize targeted information sharing and collaborative control, avoid unnecessary communication resource overhead and computing resource overhead that is ineffective in improving traffic efficiency, and realize distributed collaborative control of highways under the constraints of limited cyber resources.

[0068] 2. Based on the vehicle-level speed guidance strategy, ICV can implement single-vehicle speed control based on the recommended speed and surrounding traffic conditions, ensuring the uniform distribution of vehicles on the road section, so as to avoid the negative impact of random and non-uniform distribution of ICV in traffic flow on highway traffic efficiency.

[0069] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 Shows the three-layer cyber-physical system architecture based on ICV;

[0071] Figure 2 A schematic diagram of a highway scene is shown;

[0072] Figure 3 The CPS dynamic centralized-distributed control system architecture for highway traffic efficiency is shown;

[0073] Figure 4 The pseudo code of the main algorithm in the optimal clustering algorithm is shown;

[0074] Figure 5 The pseudo code of the sub-algorithm of the optimal clustering algorithm is shown;

[0075] Figure 6 The flowchart of the subsystem-level multi-objective optimization algorithm is shown. DETAILED DESCRIPTION

[0076] In order to make the technical solutions, advantages and purposes of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the protection scope of this application.

[0077] Figure 1 The invention shows a three-layer information-physical system architecture based on ICV, including three levels: system level, subsystem level and vehicle level. The system level is for a highway consisting of multiple sections, and can dynamically adjust the network connection to group the sections according to the traffic density of each section of the highway to form a section cluster, i.e., a subsystem; the subsystem level includes multiple section clusters, and each section can trigger the optimized speed recommendation for its main road vehicles and the control signal for the ramp based on the traffic flow situation according to the traffic information of the section cluster to which it belongs; the vehicle level can guide the speed of the ICV on the main road based on the recommended speed, and control the ICV on the ramp to drive according to the ramp flow control signal.

[0078] Figure 2 A highway consisting of five sections R1, R2, R3, R4, and R5 is shown, each of which includes an on-ramp and an off-ramp and is equipped with the same roadside infrastructure. This scenario will be used as an embodiment of the present invention. It should be noted that all vehicles in this scenario are ICVs.

[0079] Based on the three-layer cyber-physical system architecture, the present invention discloses a CPS control strategy for optimizing the traffic efficiency of highways with large traffic volume under cyber resource constraints, including a system-level dynamic grouping strategy, a subsystem-level collaborative control strategy, and a vehicle-level speed guidance strategy. Figure 3 The paper demonstrates the CPS dynamic centralized-distributed control system architecture for highway traffic efficiency based on this strategy. Figure 1 The further refinement of the system architecture shown in the figure shows in detail the architecture of the system level, subsystem level, and vehicle level, as well as the specific relationship between the three levels. Figure 2 The scenario shown is an embodiment, based on Figure 1 The basic system architecture shown and Figure 3 The detailed system architecture shown in the figure introduces the specific implementation methods of the three strategies in detail.

[0080] The system-level dynamic grouping strategy includes steps S1 and S2 in the implementation steps of the present invention:

[0081] S1. Taking the communication overhead, computational overhead and cluster correlation required by the road segment cluster structure as the optimization target, determine the system-level objective function;

[0082] That is, the communication overhead is quantified by network bandwidth occupancy and communication energy consumption, and the calculation overhead of the road section is evaluated by the calculation time. Then, the evaluation index of the correlation degree of the road section within the cluster is designed. Then, the communication overhead, calculation overhead and cluster correlation degree (the sum of the evaluation indexes of the correlation degree of the road section within the cluster) required for communication and calculation between roadside devices of each section are used as optimization targets, and the system-level objective function is determined by weighted method to establish a single-objective optimization problem.

[0083] S1.1 Calculate the communication overhead E of the system at time kT c (k);

[0084]

[0085] In the formula, C i (k) is the i-th subsystem; is a subsystem set; m represents C i (k) the mth road segment; c m The amount of data sent for the road segment m within the cluster; E elec The energy consumed for each bit of data sent or received; ε amp The energy consumption per square meter for sending each bit of data; d m,m′ is the distance between sections m and m′; Υ is the propagation attenuation index under the communication environment; constant b is the bandwidth occupied by the communication between the two sections; |C i (k)| is the number of road sections in the cluster;

[0086] S1.2 Calculate the computational cost T of the system at time kT cal ;

[0087]

[0088] Where, T evo is the evolution operation time of the optimization algorithm; t1 and t2 are constants to be fitted; M represents the total number of road sections, and in this embodiment, M=5;

[0089] S1.3 Calculate the cluster correlation R of the system at time kT d (k);

[0090]

[0091]

[0092] In the formula, ρ m,i (k) is the density of the ith interval of road segment m at time kT; is the critical density of road section m;

[0093] S1.4 According to steps S1.1 to S1.3, the system-level objective function is determined as:

[0094]

[0095] In the formula, κ c , κ t They represent the weights of communication overhead and computational overhead respectively, and are taken as κ according to the results of multiple experiments. c =0.1,κ t =0.01; k t =kT / T t , T t represents the system-level control cycle; kT represents the end time of the kth road section information sampling cycle, k represents a constant, and T represents the road section information sampling cycle;

[0096] S2. Using the designed optimal clustering algorithm to solve the single-objective optimization problem established in step S1, the optimal section cluster structure of the highway under the current traffic conditions is obtained, and the communication connection relationship (communication / non-communication) between the roadside equipment of each section and the roadside equipment of other sections is determined;

[0097] The optimal clustering algorithm in step S2 is designed based on dynamic programming (DP), which includes a main algorithm and a sub-algorithm. The main algorithm implements the preliminary segmentation of sub-areas that may form segment clusters according to the congestion level of each segment in the multi-segment area of ​​the highway. Further, for the segmented sub-areas, the main algorithm will calculate the optimal coordinated segment clustering decision in the sub-area by calling the sub-algorithm. The pseudo code of the main algorithm is as follows: Figure 4 As shown, the pseudo code of the sub-algorithm is as follows Figure 5 shown.

[0098] The subsystem-level collaborative control strategy includes S3, S4 and S5 in the implementation steps of the present invention:

[0099] S3. Based on the optimal road segment cluster structure obtained in step S2, the total time spent (TTS), total travel distance (TTD), entrance ramp queue length, speed limit mutation within the road segment, and ramp release rate mutation within the subsystem are taken as optimization targets, and five subsystem-level objective functions are established. The upper and lower bounds of the road segment speed limit and the upper and lower bounds of the ramp release rate are taken as constraints. Combined with the five subsystem-level objective functions, a multi-objective optimization problem is established.

[0100] S3.1 Establish five subsystem-level objective functions;

[0101] I. Establish the total driving time J of vehicles in the subsystemTTS (k c ) is the objective function of the optimization target;

[0102]

[0103] Where U m For the road segment m∈C j A control decision, including the speed limit value VSL of the main road control area m,i and the ramp release rate r of the entrance ramp control area m,i ; N p is the number of prediction time domain steps; Z = T c / T,k c = k / Z, T c is the subsystem level control cycle; w m′,i (Zk c +p) indicates that the i-th segment of segment m′ is in Zk c +p time corresponds to the queue length of the entrance ramp; p is an integer and p∈{Zk c ,Zk c +1,···,Z(k c +N p )}; represents the set of all segments of road segment m′; L m′,i represents the length of the i-th segment of the road segment m′; m′,i represents the number of lanes in the ith segment of road segment m′; ρ m′,i (Zk c +p) indicates that the i-th segment of segment m′ is in Zk c + density at time p;

[0104] II. Establish a total driving distance J TTD (k c ) is the objective function of the optimization target;

[0105]

[0106] In the formula, v m′,i (Zk c +p) indicates that the i-th segment of segment m′ is in Zk c +The average speed of vehicles in the segment at time p;

[0107] III. Objective function with the entrance ramp queue length as the optimization target;

[0108]

[0109]

[0110] In the formula, J w is the number of prediction time domain steps Np Inner subsystem C j Middle entrance ramp queue length penalty; w m',i (k) is the queue length of the entrance ramp corresponding to time kT; is the maximum number of vehicles queuing at the corresponding entrance ramp;

[0111] IV. Objective function with the sudden change of speed limit value within the road section as the optimization target;

[0112]

[0113] In the formula, J ΔVSL To control the number of time domain steps N c VSL: Penalty for sudden changes in speed limit within m sections; m,i represents the speed limit value of the ith section of road section m; c is an integer and c∈{0,1,···,N c -1)};

[0114] V. Objective function with the sudden change of ramp release rate within the road section as the optimization target;

[0115]

[0116] In the formula, J Δr To control the number of time domain steps N c The penalty for the sudden change of ramp release rate in section m; r m,i represents the ramp release rate of the ith section of road section m;

[0117] S3.2 Based on the five subsystem-level objective functions established in step S3.1, the multi-objective optimization model shown below is established with the upper and lower bounds of the speed limit and the upper and lower bounds of the ramp release rate as constraints:

[0118]

[0119] In the formula, VSL min and VSL max are the upper and lower bounds of the speed limit respectively; r min and r max are the upper and lower bounds of the ramp release rate respectively.

[0120] S4. Using a subsystem-level multi-objective optimization algorithm, solve the multi-objective optimization problem established in step S3 to obtain the optimal control decision, that is, the speed recommendation for the main road of each section and the ramp control signal;

[0121] The flow chart of the subsystem-level multi-objective optimization algorithm is as follows: Figure 6As shown in the figure, the algorithm first uses the non-dominated sorting evolutionary algorithm (NSGA-II) to solve the multi-objective optimization problem established in step 3, and obtains the Pareto optimal solution set under the multi-objective function, which contains multiple optimal solutions. Then, based on the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS), an optimal control decision is selected from the Pareto optimal solution set. Specifically:

[0122] The Pareto solution in the Pareto optimal solution set is represented as a control decision object with multiple objective function values. When the number of Pareto solutions in the Pareto optimal solution set is p, assuming that the optimal solution U + and the worst solution U - All Pareto solutions in the Pareto optimal solution set are z , the optimal value of z∈{TTS,TTD,ΔVSL,Δr,o} and the worst value At this time, the Euclidean distance between any Pareto solution p in the Pareto optimal solution set and the optimal solution and the worst solution is expressed as:

[0123]

[0124] Among them, D + represents the Euclidean distance between the Pareto solution p and the optimal solution, D - represents the Euclidean distance between the Pareto solution p and the worst solution. At this time, the relative distance between the Pareto solution p and the optimal solution is expressed as:

[0125]

[0126] Among them, D*∈[0,1] is used to evaluate the closeness of the Pareto solution p to the optimal solution. When D* is close to 1, the Pareto solution p is closer to the optimal solution. Each Pareto solution is scored by calculating the relative distance to the optimal solution. Finally, the scores are sorted to select the optimal control decision for the current section. It should be noted that since the dimensions of different objective function values ​​are not uniform, each objective function value needs to be standardized before calculation.

[0127] The subsystem-level multi-objective optimization algorithm can be used to solve the speed recommendations for the main roads of each road section and the ramp control signals.

[0128] S5. Design the congestion index of the road section, and use the congestion index to calculate the congestion level of each road section in the subsystem. After that, each road section will determine whether to start active control measures and update the control decision based on its own traffic status and the congestion level of the adjacent road section in the subsystem. Usually, each road section needs to update the current control decision only when the traffic status of the road section under the jurisdiction of the subsystem or the congestion level of other coordinated roads is significantly different from the information obtained when the control decision was last updated.

[0129] S5.1 Calculate the congestion level of the road section;

[0130]

[0131] In the formula, MC m represents the congestion index of road section m; is the critical density of the ith interval of section m;

[0132] S5.2 Definitions They are the control triggering time sequence and the control decision updating time sequence, and the calculation expressions are as follows:

[0133]

[0134]

[0135] In the formula, θ c , σ u and θ u are all hyperparameters.

[0136] The vehicle-level speed guidance strategy includes steps 6 and 7 in the implementation steps of the present invention:

[0137] S6. Calculate the expected speed of the vehicle based on the recommended speed of the road section and the distance between the vehicle and the vehicle in front;

[0138] In step S6, the expected speed The calculation expression is:

[0139]

[0140]

[0141] In the formula, is the distance between the vehicle and the preceding vehicle in lane l of the current road section; is the recommended speed; represents the vehicle spacing when ICVs are evenly distributed on lane l in the current road section; L veh is the average length of ICV.

[0142] S7. The ICV on the main road of the road section travels at the desired speed, while the ICV on the ramp travels according to the ramp control signal.

[0143] At this point, according to the above detailed steps, you can get Figure 2 The control decisions of each section of the highway in the illustrated embodiment optimize the traffic efficiency of highways with large traffic volumes under the constraints of cyber resources.

[0144] It is hereby declared that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solutions, which should be included in the scope of the claims of the present invention.

Claims

1. A CPS control strategy for optimizing high-volume traffic efficiency on highways, characterized in that: The following steps are involved: S1. Taking the communication overhead, computational overhead and cluster correlation required by the road segment cluster structure as the optimization target, determine the system-level objective function; S2. Use the optimal clustering algorithm to solve the optimal section cluster structure of the highway under the current traffic conditions; S3. Based on the optimal road segment cluster structure obtained in step S2, the total driving time, total driving distance, entrance ramp queue length, speed limit mutation within the subsystem, and ramp release rate mutation are used as optimization targets, and five subsystem-level objective functions are established. The upper and lower bounds of the road segment speed limit and the upper and lower bounds of the ramp release rate are further used as constraints to establish a multi-objective optimization problem; S4. Using a subsystem-level multi-objective optimization algorithm, solve the multi-objective optimization problem established in step S3 to obtain the optimal control decision, that is, the speed recommendation for the main road of each section and the ramp control signal; The subsystem-level multi-objective optimization algorithm uses a non-dominated sorting evolutionary algorithm to solve the multi-objective optimization problem established in step S3, obtains a Pareto optimal solution set under the multi-objective function, and then selects an optimal control decision in the Pareto optimal solution set based on the approximate ideal solution sorting method; S5. Design a congestion index for a road section, and use the congestion index to calculate the congestion level of each road section in the subsystem. After that, each road section determines whether to initiate active control measures and whether to update the control decision based on its own traffic status and the congestion level of the adjacent road sections in the subsystem to which it belongs; S6. Calculate the expected speed of the vehicle based on the recommended speed of the road section and the distance between the vehicle and the vehicle in front; S7. The ICV on the main road of the road section travels at the desired speed, while the ICV on the ramp travels according to the ramp control signal.

2. According to claim 1, a CPS control strategy for optimizing high-volume traffic efficiency on highways is characterized by: The step S1 comprises the following sub-steps: S1.1 Calculate the communication overhead E of the system at time kT c (k); In the formula, C i (k) is the i-th subsystem; C(k) is the subsystem set; m represents C i (k) the mth road segment; c m The amount of data sent for the road segment m within the cluster; E elec The energy consumed for each bit of data sent or received; ε amp The energy consumption per square meter for sending each bit of data; d m,m′ is the distance between sections m and m′; Υ is the propagation attenuation index under the communication environment; constant b is the bandwidth occupied by the communication between the two sections; |C i (k)| is the number of road sections in the cluster; S1.2 Calculate the computational cost T of the system at time kT cal ; Where, T evo is the evolution operation time of the optimization algorithm; t1 and t2 are constants to be fitted; M represents the total number of road sections; S1.3 Calculate the cluster correlation R of the system at time kT d (k); In the formula, ρ m,i (k) is the density of the ith interval of road segment m at time kT; is the critical density of road section m; S1.4 According to steps S1.1 to S1.3, the system-level objective function is determined as: In the formula, κ c , κ t Respectively represent the communication cost and computation cost weights; k t =kT / T t , T t represents the system-level control cycle; kT represents the end time of the kth road section information sampling cycle, k represents a constant, and T represents the road section information sampling cycle.

3. The CPS control strategy for optimizing high-volume traffic efficiency on highways according to claim 2 is characterized by: The optimal cluster division algorithm in step S2 is designed based on dynamic programming and includes a main algorithm and a sub-algorithm; The main algorithm is used to preliminarily divide the sub-areas that may form the segment cluster according to the congestion level of each segment in the multi-segment area of ​​the expressway, and to call the sub-algorithm; The sub-algorithm is used to calculate the optimal collaborative road segment clustering decision within the sub-area.

4. The CPS control strategy for optimizing high-volume traffic efficiency on highways according to claim 3 is characterized by: The step S3 comprises the following sub-steps: S3.1 Establish five subsystem-level objective functions; I. Establish the total driving time J of vehicles in the subsystem TTS (k c ) is the objective function of the optimization target; Where U m For the road segment m∈C j A control decision, including the speed limit value VSL of the main road control area m,i and the ramp release rate r of the entrance ramp control area m,i ; N p is the number of prediction time domain steps; Z = T c / T,k c = k / Z, T c is the subsystem level control cycle; w m′,i (Zk c +p) indicates that the i-th segment of segment m′ is in Zk c +p time corresponds to the queue length of the entrance ramp; p is an integer and p∈{Zk c ,Zk c +1,···,Z(k c +N p )}; represents the set of all segments of road segment m′; L m′,i represents the length of the i-th segment of the road segment m′; m′,i represents the number of lanes in the ith segment of road segment m′; ρ m′,i (Zk c +p) indicates that the i-th segment of segment m′ is in Zk c + density at time p; II. Establish a total driving distance J TTD (k c ) is the objective function of the optimization target; In the formula, v m′,i (Zk c +p) indicates that the i-th segment of segment m′ is in Zk c +The average speed of vehicles in the segment at time p; III. Objective function with the entrance ramp queue length as the optimization target; In the formula, J w is the number of prediction time domain steps N p Inner subsystem C j Mid-entry ramp queue length penalty; w m',i (k) is the queue length of the entrance ramp corresponding to time kT; is the maximum number of vehicles queuing at the corresponding entrance ramp; IV. Objective function with the sudden change of speed limit value within the road section as the optimization target; In the formula, J ΔVSL To control the number of time domain steps N c VSL: Penalty for sudden changes in speed limit within m sections; m,i represents the speed limit value of the ith section of road section m; c is an integer and c∈{0,1,···,N c -1)}; V. Objective function with the sudden change of ramp release rate within the road section as the optimization target; In the formula, J Δr To control the number of time domain steps N c The penalty for the sudden change of ramp release rate in section m; r m,i represents the ramp release rate of the ith section of road section m; S3.2 Based on the five subsystem-level objective functions established in step S3.1, the multi-objective optimization model shown below is established with the upper and lower bounds of the speed limit and the upper and lower bounds of the ramp release rate as constraints: In the formula, VSL min and VSL max are the upper and lower bounds of the speed limit respectively; r min and r max are the upper and lower bounds of the ramp release rate respectively.

5. The CPS control strategy for optimizing high-volume traffic efficiency on highways according to claim 4 is characterized by: The step S5 comprises the following sub-steps: S5.1 Calculate the congestion level of the road section; In the formula, MC m represents the congestion index of road section m; is the critical density of the ith interval of section m; S5.2 Definitions They are the control triggering time sequence and the control decision updating time sequence, and the calculation expressions are as follows: In the formula, θ c , σ u and θ u are all hyperparameters.

6. The CPS control strategy for optimizing high-volume traffic efficiency on highways according to claim 5 is characterized by: In step S6, the expected speed The calculation expression is: In the formula, is the distance between the vehicle and the preceding vehicle in lane l of the current road section; is the recommended speed; represents the vehicle spacing when ICVs are evenly distributed on lane l in the current road section; L veh is the average length of ICV.

Images