An integrated automated driving lane and multi-type intersection layout optimization method

Abstract

This invention discloses an integrated method for optimizing the layout of dedicated autonomous driving lanes and multiple intersection types, comprising the following steps: Step S1: Describing mixed traffic scenarios in an urban road network; Step S2: Clarifying the types of CAV-dedicated lanes and intersections, as well as CAV-dedicated entrance lanes; Step S3: Clarifying the relationship between decision variables and decision variables in the DISLP problem; Step S4: Calculating travel costs; Step S5: Constructing constraints for traffic distribution; Step S6: Building a problem model; and Step S7: Designing an improved genetic algorithm for solution. This method significantly reduces total system travel costs and improves road network operational efficiency. Based on the problem model, taking into account the accessibility of HVs within the road network, an improved genetic algorithm is employed to obtain a robust and stable solution with a finite number of iterations.

Description

Technical Field

[0001] The present invention relates to the field of networked autonomous driving vehicles, and in particular to an integrated autonomous driving lane and multi-type intersection layout optimization method. Background Art

[0002] With the increasing maturity of communications, onboard sensors, and computing technologies, connected and autonomous vehicles (CAVs) are widely recognized as the future of mobility, demonstrating significant potential for ensuring traffic safety and improving traffic efficiency. Compared to traditional human-driven vehicles (HVs), CAVs have faster reaction times and can maintain a shorter safe headway between vehicles ahead. Furthermore, thanks to real-time communication and coordinated control between vehicles, CAVs can operate in platoons, further shortening safe headway distances between CAVs. Furthermore, CAVs, relying on advanced autonomous driving technology, allow travelers to maximize their travel time and focus on activities other than driving. In summary, CAVs can significantly improve road capacity and profoundly change travelers' travel choices.

[0003] Due to limitations in transportation infrastructure, vehicle costs, and traveler acceptance, it's unrealistic for HVs to be completely replaced by CAVs in the short term. Instead, HVs and CAVs will coexist on urban road networks, sharing the same transportation infrastructure for a considerable period of time. Due to the lack of connected autonomous driving technology, the safe headway between HVs in a following mode is significantly greater than that between CAVs. Furthermore, HVs' random driving behavior significantly interferes with CAVs and hinders their platooning, ultimately impacting overall traffic efficiency. Therefore, under existing road infrastructure, the efficiency of a mixed HV and CAV landscape will be far lower than that of a fully connected environment, even lower than in traditional, all-HV travel scenarios. For urban road networks with mixed HV and CAV traffic, a major challenge for transportation planners is how to minimize interference between the two types of vehicles, improve road and intersection capacity, and leverage the technological advantages of CAVs to improve overall network efficiency.

[0004] To effectively leverage the efficiency advantages of CAVs on road sections, dedicated CAV lanes can be deployed. These lanes directly alter the right-of-way structure, separating CAV and HV traffic, reducing mutual interference and facilitating CAV platooning. However, inappropriate CAV lane deployment not only fails to maximize CAV efficiency but also leads to wasted traffic resources and increased congestion.

[0005] In addition, in terms of intersection traffic organization, based on advanced connected autonomous driving technology, the Autonomous Intersection Management (AIM) strategy has been proposed to replace traditional visual intersections and control CAVs through intersections. By planning the driving trajectory of each CAV entering and exiting the intersection and relying on communication and cooperation between CAVs, CAVs can more flexibly avoid conflicts when passing through intersections. The AIM strategy based on signal-free control is not limited by the number of signal phases, allowing CAVs to pass through intersections continuously, fully leveraging the advantages of connected autonomous driving technology to achieve the effect of alleviating intersection traffic congestion. Research has shown that in a connected autonomous driving environment, the AIM strategy based on signal-free control is significantly superior to traditional signalized intersections in terms of traffic efficiency.

[0006] Existing research mainly focuses on the layout of CAV lanes, the layout of various types of intersections, and the optimization of signal timing from the road network level in mixed traffic scenarios. For mixed traffic scenarios of HVs and CAVs, no research has been conducted on the integrated research of CAV lane layout, the layout of various types of intersections and signal timing optimization, and the number of CAV-only entrance lanes at intersections from the road network level. To this end, the present invention is aimed at urban traffic networks with mixed traffic of HVs and CAVs. From the perspective of traffic planning managers, by constructing a mathematical optimization model and designing an improved genetic algorithm for solution, an integrated optimization method for the layout of autonomous driving lanes and multiple types of intersections is proposed. The layout of CAV lanes, the layout of multiple types of intersections and signal timing, and the number of CAV-only entrance lanes are decided in the road network to optimize the flow distribution of HVs and CAVs on the road network, so as to achieve the goal of reducing the total travel cost of the urban road network.

[0007] Technical solution of prior art 1

[0008] Existing research focuses on optimizing the layout of CAV lanes in the road network to better manage HV and CAV traffic in mixed traffic scenarios. Chen et al. (2016) combined the CAV market penetration diffusion model with the CAV lane deployment optimization model, proposed a CAV lane layout optimization problem that minimizes the total travel cost, and obtained a time-varying CAV lane deployment plan, revealing the dynamic relationship between the CAV lane layout plan and the CAV market penetration rate. However, when the proportion of CAVs in mixed traffic flow is relatively low, setting up CAV lanes on the road section often results in a waste of road resources and even affects the operating efficiency of the entire road network. Inspired by high-occupancy lanes (HOV) lanes and high-occupancy / toll lanes (HOT) lanes, Liu et al. [2] (2019) first proposed the concept of CAV toll lanes (CAVT). By providing an optimized layout plan for the joint deployment of CAV lanes and CAVT lanes, Liu[2] et al. (2019) proved that the introduction of CAVT lanes can further improve the efficiency of road network operation in a mixed traffic environment with a relatively low proportion of CAVs. Wang[3] et al. (2021) proposed the optimal charging rate problem for CAVT lanes by considering the different travel behavior characteristics of HVs and CAVs under elastic demand. Zhang[4] et al. (2022) introduced the concept of electronic fares into the optimization layout problem of CAV lanes, proposed the deployment of CAV electronic ticket lanes, and proved its practicality.

[0009] Disadvantages of the prior art 1

[0010] 1. In technology 1, Chen[1] et al. (2016), Liu[2] et al. (2019), Wang[3] et al. (2021), and Zhang[4] et al. (2022) only focused on the deployment planning of CAV lanes or CAVT lanes in mixed traffic scenarios in urban road networks, and did not propose multiple types of intersections suitable for mixed traffic scenarios. At the same time, the delay time of HVs and CAVs at intersections was not considered in the calculation of travel time.

[0011] Technical solution of existing technology 2

[0012] Some research focuses on optimizing intersection layout and signal design in mixed-traffic scenarios on urban road networks. Li Tongfei et al. (2022) described three types of intersections for mixed traffic conditions involving HVs and CAVs, including traditional signalized intersections, signalized intersections with dedicated CAV phases and dedicated CAV entrances (referred to as signalized intersections with dedicated CAV phases and dedicated CAV entrances), and intelligent unsignalized intersections. They modeled the spatial layout and signal setting optimization problem for these three types of intersections and used an exact solution algorithm to obtain the intersection layout and signal design scheme that minimizes the total travel cost of the road network. The proposed spatial layout and signal setting planning scheme for the three types of intersections can not only leverage the technical advantages of CAV traffic efficiency but also ensure HV accessibility.

[0013] Disadvantages of the second prior art

[0014] Technology 2 only focuses on the layout and signal design of three types of intersections in mixed traffic scenarios in urban road networks, and does not consider the layout planning of CAV-only lanes in the road network. In addition, Technology 2 believes that the number of CAV-only entrance lanes at signalized intersections with CAV-only phases and CAV-only entrance lanes is externally given, and does not consider the decision-making problem of the number of CAV-only entrance lanes. In fact, how to layout CAV-only entrance lanes is not only related to the type of intersection, but also to the number of CAV-only lanes on the upstream road section. At the entrance where CAV-only entrance lanes are required, the number of CAV-only entrance lanes needs to be greater than or equal to the number of upstream CAV-only lanes, and the remaining number of ordinary entrance lanes is greater than or equal to the number of upstream ordinary lanes.

[0015] Technical problems to be solved by the present invention

[0016] To address the above technical issues, the present invention provides an integrated automated driving lane and multi-type intersection layout optimization method, which optimizes the layout of CAV lanes, the layout and signal timing of multi-type intersections, and the number of CAV-dedicated entrance lanes in a mixed traffic network, achieving the following objectives:

[0017] 1. For scenarios where HVs and CAVs coexist, and considering the accessibility of HVs in the road network, a planning method for two types of CAV-only lanes and CAV-only entrance lanes is proposed for the layout of three types of intersections, including traditional signalized intersections, signalized intersections with CAV-only phases and CAV-only entrance lanes, and intelligent unsignalized intersections. The two types of CAV-only lanes include CAV-only lanes where HVs are prohibited and CAV-only lanes where HVs can pass after paying.

[0018] 2. From the perspective of traffic planners, we propose, model, and solve an integrated optimization layout problem for dedicated autonomous driving lanes and multiple intersection types. The resulting layout solution can significantly reduce the total system travel cost and improve road network efficiency.

[0019] 3. According to the problem model, considering the accessibility of HVs in the road network, an improved genetic algorithm is used to obtain a strong stable solution with a finite number of iterations. Summary of the Invention

[0020] To address the challenges of existing technologies, this paper provides an integrated method for optimizing the layout of dedicated autonomous driving lanes and multiple intersection types. This method significantly reduces total system travel costs and improves road network efficiency. Based on the problem model, the accessibility of HVs within the road network is considered, and a modified genetic algorithm is employed to obtain a robust, stable solution using a finite number of iterations.

[0021] The technical solutions of the present invention are as follows:

[0022] A method for optimizing the layout of integrated autonomous driving lanes and multi-type intersections includes the following steps:

[0023] Step S1: describing mixed traffic scenarios in urban road networks;

[0024] Step S2: Clarify the types of CAV lanes and intersections, and CAV entrance lanes;

[0025] Step S3: clarify the decision variables and the relationship between decision variables in the DISLP problem;

[0026] Step S4: Calculate travel cost;

[0027] Step S5: constructing flow distribution constraint conditions;

[0028] Step S6: constructing a problem model;

[0029] Step S7: Design an improved genetic algorithm.

[0030] Step S1 includes the following sub-steps:

[0031] Sub-step S11: Use represents the road network, N represents the node set, represents the set of CAV lanes on the road segment, Represents the set of common lanes on the road segment;

[0032] Sub-step S12: R and S are the sets of starting points and ending points in the road network,

[0033] Sub-step S13: r and s represent a starting point and an end point respectively, r∈R, s∈S, (i, j) represents the road segment with starting point i and end point j,

[0034] Sub-step S14: The proposed planning problem is described using a link-node modeling method, where the traffic on the road section is differentiated by different destinations and vehicle types.

[0035] Step S2 includes the following sub-steps:

[0036] Sub-step S21: Determine the type of CAV lane in the joint optimization layout solution: a CAV lane where HVs are prohibited from passing and CAVs can pass freely is called a conventional CAV lane; a CAV lane where HVs can pass by paying and CAVs can pass freely is called a CAVT lane;

[0037] Sub-step S22: Clarify the intersection type in the joint optimization layout plan: traditional signalized intersections implement visual signal control, and HVs and CAVs travel together through the intersection according to signal instructions; intelligent unsignalized intersections implement the AIM strategy based on unsignaled control, CAVs communicate and coordinate with each other to pass through the intersection, and HVs are prohibited from entering; signalized intersections with CAV-dedicated phases and CAV-dedicated entrance lanes (referred to as signalized intersections with CAV-dedicated phases and CAV-dedicated entrance lanes), the intersection is equipped with a CAV-dedicated entrance lane and a signal phase is equipped with a CAV-dedicated phase. The CAV-dedicated entrance lane only allows CAVs to enter, and HVs are prohibited from entering. The AIM strategy based on unsignaled control is implemented in the CAV-dedicated phase. At this time, only CAVs in the dedicated entrance lane are allowed to pass, and traditional visual signal control is implemented in other phases.

[0038] Step S3 includes the following sub-steps:

[0039] Sub-step S31: clarifying the decision variables of the planning problem;

[0040] Sub-step S32: Clarify the relationship between decision variables: Considering HV accessibility, clarify the relationship between multiple types of intersections in the road network; the relationship between intersection type and CAV lane type, number of CAV lanes, and number of CAV entrance lanes; the relationship between CAV lane type and HV toll collection; the relationship between signal cycle and green light time of each phase.

[0041] Step S4 includes the following sub-steps:

[0042] Sub-step S41: calculating the travel time of the road section;

[0043] Sub-step S42: Calculate the import delay time;

[0044] Sub-step S43: Calculate travel cost.

[0045] Step S5 includes the following sub-steps:

[0046] Sub-step S51: constructing flow non-negativity and demand constraints;

[0047] Sub-step S52: Constructing user balancing constraints under the mixed traffic conditions of HVs and CAVs.

[0048] Step S7 includes the following sub-steps:

[0049] Sub-step S71: Considering the accessibility of HVs in the road network, generate a set of initial feasible solutions;

[0050] Sub-step S72: Bring the feasible solution into the GAMS solver to solve a network traffic distribution sub-problem, calculate the objective function value of the original problem corresponding to the feasible solution, and sort according to the obtained original problem objective value;

[0051] Sub-step S73: selecting the father and mother generations according to the principle that the smaller the target value, the greater the probability of becoming the father and mother generations, and performing genetic mutation to generate offspring;

[0052] Sub-step S74: determining whether the generated offspring meet the accessibility requirements of the HV in the road network. If the offspring do not meet the requirements, they will be replaced by newly generated offspring that meet the accessibility requirements of the HV until all the generated offspring meet the accessibility requirements of the HV in the road network.

[0053] Sub-step S75: Return to step S72 until the maximum evolutionary generation is reached, and the program terminates.

[0054] The beneficial effects of the integrated automatic driving lane and multi-type intersection layout optimization method of the present invention are as follows:

[0055] 1. The present invention is aimed at scenarios where HVs and CAVs coexist, and takes into account the accessibility of HVs in the road network. It proposes a planning method for the layout of three types of intersections, including traditional signalized intersections, signalized intersections with CAV-dedicated phases and CAV-dedicated entrance lanes, and intelligent unsignalized intersections. The method also includes two types of CAV-dedicated lanes: CAV-dedicated lanes where HVs are prohibited, CAV lanes where HVs can pass by paying, and CAV-dedicated entrance lanes.

[0056] 2. From the perspective of traffic planners, this paper proposes, models, and solves an integrated optimization layout problem for dedicated autonomous driving lanes and multiple intersection types. The resulting layout solution can significantly reduce the system's total travel cost and improve road network efficiency.

[0057] 3. The present invention targets the problem model, considers the accessibility of HVs in the road network, and adopts an improved genetic algorithm to obtain a strong stable solution with a finite number of iterations. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 This is the road network diagram of the present invention.

[0059] Figure 2 This is the ND network diagram of the present invention.

[0060] Figure 3 It is a diagram of the iterative process of the solution algorithm of the present invention. DETAILED DESCRIPTION

[0061] The specific embodiments of the present invention are described below to facilitate understanding of the present invention by those skilled in the art. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the appended claims, these changes are obvious, and all inventions and creations utilizing the concepts of the present invention are protected.

[0062] Step 1. Description of mixed traffic scenarios in urban road networks

[0063] First, the road network in the mixed traffic scenario of HV and CAV is described. Represents a road network, where N represents a node set, represents the set of CAV lanes on the road segment, Represents the set of common lanes on the road segment. R and S are the sets of starting points and end points in the road network, respectively. r and s represent a starting point and an end point respectively, r∈R, s∈S. (i,j) represents the road segment with starting point i and end point j. The link-node modeling method is used to describe the proposed planning problem. The traffic on the road segment will be differentiated by the different end points and vehicle types (HV or CAV). For example, Represents the HV flow on the link (i, j) to the end point s.

[0064] Step 2. Clarify the types of CAV lanes and intersections, and the concept of CAV entrance lanes

[0065] The types of CAV lanes in the joint optimization layout plan are clarified: CAV lanes where HVs are prohibited and CAVs can freely pass are called regular CAV lanes; CAV lanes where HVs can pass by paying and CAVs can freely pass are called CAVT lanes. Both HVs and CAVs can freely pass through regular lanes.

[0066] There are three types of intersections:

[0067] Traditional signalized intersection: The intersection is controlled by visual signals, and HVs and CAVs travel together through the intersection according to signal instructions.

[0068] Intelligent Unsignalized Intersection: Implements an AIM strategy based on unsignalized control. CAVs communicate with each other and coordinate through the intersection. The all-red traffic light on the upstream exit of the intersection restricts HVs from entering.

[0069] Signalized intersection with a dedicated CAV phase and a dedicated CAV entrance lane, referred to as a signalized intersection with a dedicated CAV phase and a dedicated CAV entrance lane: the intersection is equipped with a dedicated CAV entrance lane and the signalized phase is equipped with a dedicated CAV phase. The dedicated CAV entrance lane only allows CAVs to enter. The AIM strategy based on non-signalized control is implemented in the dedicated CAV phase. At this time, only CAVs in the dedicated entrance lane are allowed to pass. Traditional visual signal control is implemented in other phases.

[0070] It should be noted that at signalized intersections with CAV dedicated phases and CAV dedicated entrance lanes, CAVs are defaulted to enter the CAV dedicated entrance lanes. In summary, after integrating the layout of CAV dedicated lanes, the layout of multiple types of intersections and signal timing, and the deployment of CAV dedicated entrance lanes at intersections, the original traffic network will be transformed into a widened network, such as Figure 1 .

[0071] Step 3. Clarify the decision variables and the relationship between decision variables in the DISLP problem

[0072] 1. Decision variables

[0073] Existing research has not considered the integrated optimization layout of autonomous driving lanes and multi-type intersections, nor the relationship between different types of intersections and CAV lanes and CAV entrance lanes in the planning scheme. To this end, the present invention first defines the decision variables of the planning problem.

[0074]

[0075]

[0076]

[0077]

[0078]

[0079]

[0080]

[0081]

[0082]

[0083] Where, Indicates the number of CAV lanes on the road section (i, j). As an integer variable, it can take the value from 0 to the maximum number of lanes on the road section (i, j). Indicates the type of the CAV lane on the road section (i, j). As a 0-1 variable, 0 indicates that the CAV lane on the road section (i, j) is a regular CAV lane, and 1 indicates that the CAV lane on the road section (i, j) is a CAVT lane; τ ij It represents the fee charged by the CAVT lane on the road section (i, j) for entering HVs. It is a continuous variable with a given value range. Indicates the number of CAV dedicated entrance lanes connecting road section (i, j) and intersection j. As an integer variable, it can be 0 to the maximum number of entrance lanes on road section (i, j). z j Indicates whether intersection j has a normal signal phase, as a 0-1 variable, 0 means no normal signal phase, 1 means a normal signal phase; z′ j Indicates whether intersection j has a CAV dedicated signal phase, as a 0-1 variable, 0 indicates no CAV dedicated signal phase, and 1 indicates a CAV dedicated signal phase; c j Represents the signal period of intersection j, as a continuous variable, with a given value interval It represents the hth common signal phase of intersection j, as a continuous variable, with a given value interval The CAV signal phase at intersection j is a continuous variable with a given value interval. By z j 、z′ j The value combination of can represent the intersection type, z j =1, z′ j =0 means that intersection j is a traditional signalized intersection; j =1, z′ j =1 means that intersection j is a signalized intersection with a CAV-dedicated phase and a CAV-dedicated entrance lane; j =0, z′ j =0 indicates that intersection j is an intelligent unsignalized intersection.

[0084] 2. Relationships between decision variables

[0085] Considering the accessibility of HVs in the road network, the relationship between multiple types of intersections in the road network is clarified.

[0086]

[0087]

[0088]

[0089]

[0090]

[0091]

[0092]

[0093] Where Θ represents the set of HV reachability between nodes.

[0094] Among them, θ ij is a 0-1 variable, where 0 indicates that the HV between node i and node j is unreachable, and 1 indicates that the HV between node i and node j is reachable; Γ (0) represents the adjacency matrix of the original road network, where is a 0-1 variable, 1 indicates that node i and node j are adjacent in the original road network, otherwise it is 0; Γ represents the road network adjacency matrix for HV travel, where γ ij is a 0-1 variable, where 1 indicates that both node i and node j are not intelligent unsignalized intersections and are adjacent to each other in the original road network, otherwise 0; i is a 0-1 variable indicating whether node i is an intelligent unsignalized intersection. If it is an intelligent unsignalized intersection, it takes 1, otherwise it takes 0. As shown in formula (14), since intelligent unsignalized intersections prohibit HVs from passing through, HVs can only pass when both adjacent nodes are not intelligent unsignalized intersections. In formula (15), (Γ) n The set Θ of HV accessibility between nodes is obtained by Boolean addition of 1 to n powers of Γ. Formula (16) shows that if the HV travel demand from the starting point r to the end point s is When it is greater than 0, HV can reach from the starting point r to the end point s, that is, θ rs is 1. Combining the above constraints, we can limit the variable combination z representing the intersection type. j 、z′ j The value of is chosen so that it satisfies the accessibility condition of HV.

[0095] 3. Relationship between intersection types, CAV lanes, and CAV entrance lanes

[0096]

[0097]

[0098]

[0099]

[0100]

[0101]

[0102]

[0103]

[0104]

[0105]

[0106] As mentioned above, the variable combination z representing the intersection type j 、z′ j The values ​​correspond to three types of intersections. Formula (17) can avoid z j =0, z′ j =1 value combination appears; Formula (18) indicates that the number of entrance lanes at the intersection entrance is equal to the sum of the number of lanes in the section and the number of widened lanes, where Y represents the number of widened lanes at the intersection entrance. Formula (19) indicates that if the intersection is an intelligent unsignalized intersection, all lanes in the upstream section are CAV lanes, otherwise at least one ordinary lane is reserved. Formula (20) indicates that the total number of lanes in the section is the sum of the number of CAV lanes and the number of ordinary lanes. Formula (21) indicates that if the intersection is a traditional signalized intersection, no CAV entrance lane is set, otherwise the number of CAV entrance lanes is greater than or equal to the number of upstream CAV lanes and does not exceed the sum of the number of upstream CAV lanes and the number of widened lanes. Formula (22) indicates that if the intersection is an intelligent unsignalized intersection, all entrance lanes are CAV entrance lanes. Formula (23) indicates that if the intersection is a CAV-dedicated phase signal intersection or an intelligent unsignalized intersection, there is at least one CAV entrance lane at the entrance. Equation (24) indicates that if the intersection is an intelligent, unsignalized intersection, no ordinary entrance lane is set. Otherwise, all entrance lanes other than the CAV-dedicated entrance lane are ordinary entrance lanes. Since the CAV-dedicated lane and ordinary lane on the same road section (i, j) connect to the same entrance, as shown in Equations (25-26), the layout of the CAV-dedicated entrance lane and ordinary lane downstream of the ordinary lane is the same.

[0107] 4. Relationship between CAV lane types and HV service fees

[0108]

[0109]

[0110]

[0111] In formula (27), ω ijis a 0-1 variable indicating whether a CAV lane is set on the road section (i, j). If a CAV lane is set, it takes 1, otherwise it takes 0. Formula (28) stipulates that if a CAV lane is not set on the road section (i, j), the CAV lane type variable = 0. Formula (29) stipulates that if the CAV lane is a regular CAV lane, the charging variable τ ij is 0.

[0112] 5. Relationship between signal cycle and green light time of each phase

[0113]

[0114]

[0115]

[0116]

[0117]

[0118] Formula (30) indicates that if intersection j is not a signalized intersection with a CAV-dedicated phase and a CAV-dedicated entrance, the duration of the CAV-dedicated phase is 0. Formulas (31-33) indicate the relationship between the upper and lower bounds of the cycle duration and the upper and lower bounds of the green light duration of each phase for signalized intersections. Where w is the green light interval, including the yellow light time and the full red time. Formula (34) calculates the green light duration of the normal signal phase of intersection j at the entrance (i, j). Where, It is a 0-1 variable indicating whether the entrance (i, j) corresponds to the hth common phase of intersection j.

[0119] Step 4. Calculate travel costs

[0120] 1. Calculation of travel time for road sections

[0121]

[0122]

[0123]

[0124]

[0125]

[0126] In formula (35-36), u ij The traffic capacity of pure HVs on CAV lanes and ordinary lanes is equal to the number of CAV lanes on road section (i, j). Number of ordinary lanes Multiply the average safe headway time of a HV following a HV by In formula (37), the BPR function is used to calculate the time t required for a vehicle to pass through the road section (i, j). ij , It represents the time required for a vehicle to freely pass through the road section (i, j), α and β are two positive parameters in the BPR function. are the HV and CAV traffic on road section (i, j), pcu ij is the conversion coefficient of CAV flow to HV flow. In formula (38), according to the derivation of formula, the conversion coefficient of CAV flow to HV flow pcu is obtained ij , is a ratio of CAV traffic to mixed traffic flow p ij Related functions. Among them, is the average safe headway time between CAV and HV, is the average safe headway between HV and CAV, is the average safe headway time of CAV following CAV. Formula (39) shows that the proportion of CAV flow in mixed traffic flow on road section (i, j) is p. ij It is equal to the CAV flow rate on road segment (i, j) divided by the sum of CAV flow rate and HV flow rate.

[0127] 2. Calculation of import delay time

[0128] 1) Inlet flow constraint

[0129]

[0130]

[0131]

[0132] Equation (40) indicates that the CAV flow at the intersection entrance downstream of the CAV lane on road section (i, j) is equal to the sum of the CAV flow on the CAV lane and the ordinary lane. The same is true for HV. Equation (41) indicates that the ordinary lane and the CAV lane downstream on road section (i, j) have the same flow at the intersection entrance. Equation (42) calculates the conversion coefficient from CAV flow to HV flow on the entrance road (i, j).

[0133] 2) Calculation of delay time for ordinary entrance

[0134]

[0135]

[0136]

[0137]

[0138] In formula (43), λ ij is a 0-1 variable, indicating whether the common inlet channel (i, j) is oversaturated. If it is oversaturated, it takes 1, otherwise it takes 0. Formula (44) calculates the delay time d of the common inlet channel (i, j) ij Among them, o ij is a non-negative continuous variable, representing the saturation of the common inlet (i, j); T is a fixed parameter, representing the duration of the derived flow; Equation (45) establishes λ by introducing a sufficiently large number M ij Saturation o of the common inlet (i, j) ij If the saturation of the common inlet channel (i, j) is o ij >1,λ ij Taking 0 will not satisfy the constraint, so it can only be 1; if the saturation of the common inlet channel (i, j) is o ij ≤1,λ ij Taking 1 will not satisfy the constraint, so it can only be taken as 0. Equation (46) calculates the saturation o of the common inlet channel (i, j) ij , where the numerator is the equivalent HV flow at the common inlet, and the denominator is the vehicle dissipation capacity of the common inlet. X is a fixed parameter representing the saturation flow rate of HV flow through a common inlet.

[0139] 3) Calculation of CAV dedicated entrance delay time under CAV dedicated phase

[0140]

[0141]

[0142]

[0143]

[0144] In formula (47), is a 0-1 variable indicating whether the CAV dedicated entrance lane (i, j) is oversaturated in the CAV dedicated phase. If it is oversaturated, it takes 1, otherwise it takes 0. Formula (48) calculates the delay time of the CAV dedicated entrance lane (i, j) in the CAV dedicated phase. in, is a non-negative continuous variable, representing the saturation of the CAV dedicated inlet channel (i, j) under the CAV dedicated phase; Equation (49) is established by introducing a sufficiently large number M. Saturation of CAV dedicated inlet channel (i, j) under CAV dedicated phase If the saturation of the CAV dedicated inlet channel (i, j) under the CAV dedicated phase is Taking 0 will not satisfy the constraint, so it can only take 1; if the saturation of the CAV dedicated inlet channel (i, j) in the CAV dedicated phase Taking 1 will not satisfy the constraint, so it can only be 0. Equation (50) calculates the saturation of the CAV dedicated inlet channel (i, j) under the CAV dedicated phase The numerator is the CAV flow rate at the CAV dedicated entrance, and the denominator is the vehicle dissipation capacity of the CAV dedicated entrance. is a fixed parameter, representing the saturated flow rate of CAV traffic passing through a CAV-dedicated entrance lane, and ρ is the improvement coefficient of the intersection traffic efficiency of the unsignaled AIM control strategy compared with the traditional signalized intersection.

[0145] 4) Calculation of Delay Time for CAV Entrance Lanes at Intelligent Unsignalized Intersections

[0146]

[0147]

[0148]

[0149] In formula (51), ξ ij is a 0-1 variable, indicating whether the CAV dedicated entrance lane (i, j) of the intelligent unsignalized intersection is oversaturated. If it is oversaturated, it takes 1, otherwise it takes 0. Formula (52) calculates the delay time of the CAV dedicated entrance lane (i, j) of the intelligent unsignalized intersection Formula (53) introduces a sufficiently large number M to establish ξ ij and relationship, if ξ ij Taking 0 will not satisfy the constraint, so it can only take 1; if ξ ij Taking 1 will not satisfy the constraint, so it can only take 0.

[0150] 3. Travel cost calculation

[0151] 1) Calculation of travel cost by road section

[0152]

[0153]

[0154]

[0155] Equation (54) calculates the travel cost of an HV on a road section (i, j) through a CAV lane: The travel cost of HVs using CAV lanes is calculated based on the following conditions: the intersection is an intelligent non-signalized intersection, there is no CAV lane on the road section, and the CAV lane is a conventional CAV lane. is a sufficiently large number M. Otherwise, the travel cost of HVs through CAV lanes is is the travel time t through the CAV lane ij and the fee charged τ ij The sum of the values, multiplied by the time value coefficient κ of HV HV Equation (55) calculates the travel cost of a CAV on a road section (i, j) through a CAV dedicated lane: When there is no CAV lane on road section (i, j), the travel cost of a CAV through the CAV lane is is a sufficiently large number M. Otherwise, the travel cost of CAV through the CAV lane is is the travel time t through the CAV lane ij Multiply by the time value coefficient κ of CAV CAV Equation (56) calculates the travel cost of HV and CAV on road section (i, j) through the ordinary lane When the intersection is an intelligent non-signalized intersection, the travel cost of HV and CAV through the ordinary lane is a sufficiently large number M. Otherwise, the travel cost of HV and CAV through ordinary lanes is are the travel time t through the ordinary lane ij Multiply by the time value coefficient κ of HV and CAV m .

[0156] 2) Calculation of travel costs at intersection entrances

[0157]

[0158]

[0159]

[0160] Equation (57) calculates the travel cost of an HV on the CAV lane (i, j) passing through intersection j: If intersection j is an intelligent non-signalized intersection, HVs are prohibited from passing through it. Therefore, the travel cost of an HV on the CAV lane (i, j) passing through intersection j is a sufficiently large number M. Otherwise, it is the delay time d for an HV on the CAV lane (i, j) passing through the ordinary entrance lane (i, j). ij Multiply by the time value coefficient κ of HV HVEquation (58) calculates the travel cost of a CAV on the CAV lane (i, j) passing through intersection j: If intersection j is a traditional signalized intersection, the travel cost of a CAV on the CAV lane (i, j) passing through intersection j is is the delay time d of the CAV on the CAV dedicated lane (i, j) passing through the common entrance lane (i, j) ij Multiply by the time value coefficient κ of CAV CAV If intersection j is a signalized intersection with a CAV-only phase and a CAV-only entrance lane, the travel cost of a CAV on the CAV-only lane (i, j) through intersection j is is the delay time of the CAV on the CAV lane (i, j) passing through the CAV entrance lane (i, j) during the CAV phase Multiply by the time value coefficient κ of CAV CAV ; If intersection j is an intelligent non-signalized intersection, the travel cost of a CAV on the CAV lane (i, j) passing through intersection j is is the delay time of the CAV on the CAV lane (i, j) passing through the CAV entrance lane (i, j) Multiply by the time value coefficient κ of CAV CAV Formula (59) indicates that HVs and CAVs on the ordinary lane and CAV lane on the road section (i, j) have the same travel cost when passing through intersection j. Step 5. Traffic distribution constraints

[0161] 1. Non-negative flow and demand constraints

[0162]

[0163]

[0164]

[0165]

[0166]

[0167] Equation (60) indicates that the HV and CAV flows on link (i, j) to endpoint s are both nonnegative. Equations (51-64) indicate node flow conservation. In a road network, for endpoint s, the flow into endpoint s should equal the total attraction of node s; for starting point r, the flow out of starting point r and ending at node s is the travel demand between the OD pair; for an intermediate node k, the flow out of node k and ending at s should equal the flow into node k and ending at s.

[0168] 2. User balance constraints under mixed HV and CAV traffic conditions

[0169] The UE equilibrium constraint is constructed based on the link-node modeling method: when equilibrium is reached, if the road segment (i, j) is on the path from node i to end point s, and there is HV or CAV traffic on the road segment, then the travel cost of HV or CAV on the road segment (i, j) is equal to the minimum travel cost from node i to end point s minus the minimum travel cost from node j to end point s.

[0170]

[0171]

[0172]

[0173]

[0174] Equations (65-66) represent the UE balancing constraints for HVs on the road network; Equations (67-68) represent the UE balancing constraints for CAVs on the road network. HV and CAV traffic on roads with different end points The relationship between and They represent the minimum HV travel cost and the minimum CAV travel cost from node j to end point s, respectively.

[0175] Step 6. Problem Model

[0176] For the mixed traffic scenario of connected autonomous driving, an integrated autonomous driving lane and multi-type intersection layout optimization method is proposed, with the goal of minimizing the total travel time in order to improve the system operation efficiency and alleviate traffic congestion. In summary, the problem model is:

[0177]

[0178] constraint:

[0179] a) Decision variables and relationships between them in DISLP problems (1-34)

[0180] b) Calculation of travel costs for HV and CAV (35-59)

[0181] c) HV and CAV traffic distribution constraints (60-68)

[0182] Step 7. Improve the genetic algorithm

[0183] The improved genetic algorithm is used to solve the DISLP problem:

[0184] Step 1: Consider the accessibility of HVs in the road network and generate a set of initial feasible solutions.

[0185] Step 2: Bring the feasible solutions into the GAMS solver to solve a network traffic allocation sub-problem and sort them according to the obtained original problem objective value.

[0186] Step 3: Select the father and mother generations according to the principle that the smaller the target value, the greater the probability of becoming the father and mother, and perform genetic mutation to generate offspring.

[0187] Step 4: Determine whether the generated offspring meet the accessibility requirements of HV in the road network. If the offspring do not meet the requirements, they will be replaced by newly generated offspring that meet the accessibility requirements of HV until all the generated offspring meet the accessibility requirements of HV in the road network.

[0188] Step 5: Repeat steps 2, 3, and 4 above until the maximum evolutionary generation is reached and the program terminates.

[0189] The present invention proposes an integrated layout optimization method for autonomous driving lanes and multiple types of intersections and a solution algorithm for mixed traffic scenarios of connected autonomous driving. This method can significantly improve road network performance, increase the overall travel efficiency of mixed traffic networks of HVs and CAVs under different market penetration rates, and give full play to the advantages of connected autonomous driving technology.

[0190] The road network is described by example analysis. The Nguyen-Dupuis (ND) network consists of 13 nodes (starting nodes 1 and 4, ending nodes 2 and 3), 19 road arcs, and 4 OD pairs (1-2, 1-3, 4-2, and 4-3). Figure 2 In the extended network, each road segment arc is divided into a common lane arc and a CAV dedicated lane arc, for a total of 38 arcs. The road parameters of the ND network are shown in Table 1.

[0191] Related parameter settings in the model: HV and CAV travel demand between OD See Table 2. The parameters in the BPR function are set to α = 0.15 and β = 4. The saturation flow rate X of HVs passing through the intersection is 1700 veh / h. The coefficient ρ, which indicates the efficiency improvement of the unsignalized intelligent intersection over the traditional signalized intersection, is 2.1566. All signalized intersections use two-phase round-robin control in the ordinary phase, with a cycle length of c. j The upper limit is 200 seconds, the lower limit is 37 seconds, and the green light time of each phase The lower bound of is 15 seconds, the green light interval w is 3 seconds; the CAVT lane charges HV a fee τ ij The upper limit is 3 minutes and the lower limit is 0; the number of CAV dedicated lanes The upper bound of the number of lanes in the road section is shown in Table 1, and the lower bound is 0; the number of widened entrance lanes Y is 2; the four types of headway time They are 1.8 seconds, 1.8 seconds, 1.2 seconds, and 0.9 seconds respectively; the time value is given as HV is 1 and CAV is 1; the maximum evolutionary generation is 200 generations.

[0192] Table 1: Segment parameters of the Nguyen-Dupuis network

[0193]

[0194]

[0195] Table 2 Travel demand table

[0196] OD Travel demand (vehicles / hour) OD Travel demand (vehicles / hour) q_1^{HV,2} 2500 q_1^{CAV,2} 2500 q_1^{HV,3} 2800 q_1^{CAV,3} 2800 q_4^{HV,2} 2250 q_4^{CAV,2} 2250 q_4^{HV,3} 2600 q_4^{CAV,3} 2600

[0197] By giving the given parameters, the proposed model and algorithm are used to optimize the layout of integrated autonomous driving lanes and multi-type intersections for all road sections and intersections of the ND network. The optimized layout schemes are shown in Tables 3-4, and the traffic distribution of HVs and CAVs on the optimized road network is shown in Table 5. The iterative process of solving the layout optimization problem of the integrated autonomous driving lanes and multi-type intersections by the improved genetic algorithm is shown in Figure 3 ,It can be seen that the obtained optimization scheme is a strong stable solution of the solving algorithm. The total travel time of vehicles on the road network under the initial scheme is 15379.97931, and the total travel time of vehicles on the road network under the optimized scheme is 12438.01248, which is a year-on-year decrease of 19.13%.

[0198] Table 3 Optimization plan for the layout of CAV dedicated lanes and CAV dedicated entrance lanes on road sections

[0199]

[0200]

[0201] Table 4 Intersection types and signal design optimization schemes

[0202]

[0203] Table 5 Traffic distribution of HV and CAV on the road network

[0204]

[0205]

[0206] The present invention proposes an integrated optimization method for the layout of autonomous driving lanes and multiple types of intersections in the mixed traffic scenario of connected autonomous driving, clarifies the concepts of two types of CAV lanes, three types of intersections, and CAV entrance lanes, and analyzes the constraints between the two types of CAV lanes, three types of intersections, signal design, and CAV entrance lanes. The proposed DISLP problem is modeled. In the problem, the type of each intersection in the decision-making road network (traditional signalized intersection, signalized intersection with CAV dedicated phase and CAV dedicated entrance lane, and intelligent non-signaled intersection) and intersection signal design (cycle duration, ordinary phase duration, CAV dedicated phase duration) are determined, and the deployment of CAV lanes (conventional CAV lanes, CAVT lanes) on each road section in the decision-making road network and the deployment of CAV dedicated entrance lanes at each entrance are determined, so that the total travel cost of the overall road network in the mixed traffic scenario is minimized. Considering the accessibility of HVs in the road network, an improved genetic algorithm is used to obtain a strong and stable solution to the original problem with a finite number of iterations. The modeling method and solution algorithm for the DISLP problem proposed in this invention can significantly improve road network performance, increase the travel efficiency of HVs and CAVs under different CAV market shares, ensure the accessibility of HVs in the road network, and give full play to the advantages of connected autonomous driving technology.

Claims

1. An integrated automated driving lane and multi-type intersection layout optimization method, characterized in that: The following steps are involved: Step S1: describing mixed traffic scenarios in urban road networks; Sub-step S11: Use represents the road network, N represents the node set, represents the set of CAV lanes on the road segment, Represents the set of common lanes on the road segment; Sub-step S12: R and S are the sets of starting points and ending points in the road network, Sub-step S13: r and s represent a starting point and an end point respectively, r∈R, s∈S, (i, j) represents the road segment with starting point i and end point j, Sub-step S14: Use the link-node modeling method to describe the proposed DISLP planning problem, where the traffic on the road segment will be differentiated by different destinations and vehicle types; Step S2: Clarify the types of CAV lanes and intersections, and CAV entrance lanes; Sub-step S21: Determine the type of CAV lane in the joint optimization layout solution: a CAV lane where HVs are prohibited from passing and CAVs can pass freely is called a conventional CAV lane; a CAV lane where HVs can pass by paying and CAVs can pass freely is called a CAVT lane; Sub-step S22: Determine the intersection type in the joint optimization layout plan: Traditional signalized intersections implement visual signal control, and HVs and CAVs travel together through the intersection according to signal instructions; Intelligent unsignalized intersections implement an AIM strategy based on unsignalized control, and CAVs can pass through the intersection through real-time communication and coordinated control with each other, while HVs are prohibited from entering; Signalized intersections with CAV-dedicated phases and CAV-dedicated entrance lanes are set at the signalized intersection, and the signal phases are set as CAV-dedicated phases. The CAV-dedicated entrance lanes allow CAVs to enter, but HVs are prohibited from entering. The AIM strategy based on unsignalized control is implemented in the CAV-dedicated phases, allowing CAVs on the dedicated entrance lanes to pass, and traditional visual signal control is implemented in other phases; Step S3: Clarify the relationship between decision variables and decision variables in the DISLP problem. The decision variables include Where, Indicates the number of CAV lanes on the road section (i, j). As an integer variable, it can take the value from 0 to the maximum number of lanes on the road section (i, j). Indicates the type of the CAV lane on the road section (i, j). As a 0-1 variable, 0 indicates that the CAV lane on the road section (i, j) is a regular CAV lane, and 1 indicates that the CAV lane on the road section (i, j) is a CAVT lane; τ ij It represents the fee charged by the CAVT lane on the road section (i, j) for entering HVs. It is a continuous variable with a given value range. Indicates the number of CAV dedicated entrance lanes connecting road section (i, j) and intersection j. As an integer variable, it can be 0 to the maximum number of entrance lanes on road section (i, j). z j Indicates whether intersection j has a normal signal phase, as a 0-1 variable, 0 means no normal signal phase, 1 means a normal signal phase; z′ j Indicates whether intersection j has a CAV dedicated signal phase, as a 0-1 variable, 0 indicates no CAV dedicated signal phase, and 1 indicates a CAV dedicated signal phase; c j Represents the signal period of intersection j, as a continuous variable, with a given value interval It represents the hth common signal phase of intersection j, as a continuous variable, with a given value interval The CAV signal phase at intersection j is a continuous variable with a given value interval. By z j 、z′ j The value combination of can represent the intersection type, z j =1, z′ j =0 means that intersection j is a traditional signalized intersection; j =1, z′ j =1 means that intersection j is a signalized intersection with a CAV-dedicated phase and a CAV-dedicated entrance lane; j =0, z′ j =0 means that intersection j is an intelligent unsignalized intersection; The relationship between decision variables Consider the accessibility of HVs in the road network and clarify the relationship between multiple types of intersections in the road network Where Θ represents the set of HV reachability between nodes; Among them, θ ij is a 0-1 variable, where 0 indicates that the HV between node i and node j is unreachable, and 1 indicates that the HV between node i and node j is reachable; Γ (0) represents the adjacency matrix of the original road network, where is a 0-1 variable, 1 indicates that node i and node j are adjacent in the original road network, otherwise it is 0; Γ represents the road network adjacency matrix for HV travel, where γ ij is a 0-1 variable, where 1 indicates that both node i and node j are not intelligent unsignalized intersections and are adjacent to each other in the original road network, otherwise 0; i is a 0-1 variable, indicating whether node i is an intelligent unsignalized intersection. If it is an intelligent unsignalized intersection, it takes 1, otherwise it takes 0. As shown in formula (14), since intelligent unsignalized intersections prohibit HVs from passing through, HVs can only pass when both adjacent nodes are not intelligent unsignalized intersections. In formula (15), (Γ) n The set Θ of HV accessibility between nodes is obtained by Boolean addition of Γ from 1 to n powers. Equation (16) shows that if the HV travel demand from the starting point r to the end point s is When it is greater than 0, HV can reach from the starting point r to the end point s, that is, θ rs =1, combining the above constraints to limit the variable combination z representing the intersection type j 、z′ j The value of is chosen so that it satisfies the accessibility condition of HV; Relationship between intersection types and CAV lanes and CAV entrance lanes Variable combination z representing the intersection type j 、z′ j The values ​​correspond to three types of intersections. Formula (17) can avoid z j =0, z′ j =1 value combination appears; Formula (18) indicates that the number of entrance lanes at the intersection entrance is equal to the sum of the number of lanes in the section and the number of widened lanes, where Y represents the number of widened lanes at the intersection entrance. Formula (19) indicates that if the intersection is an intelligent non-signalized intersection, all lanes in the upstream section are CAV lanes, otherwise at least one ordinary lane is reserved. Formula (20) indicates that the total number of lanes in the section is the sum of the number of CAV lanes and the number of ordinary lanes. Formula (21) indicates that if the intersection is a traditional signalized intersection, no CAV entrance lane is set, otherwise the number of CAV entrance lanes is greater than or equal to the number of upstream CAV lanes and does not exceed the sum of the number of upstream CAV lanes and the number of widened lanes. , formula (22) indicates that if the intersection is an intelligent non-signalized intersection, all entrance lanes are CAV-only entrance lanes; formula (23) indicates that if the intersection is a CAV-only phase signalized intersection or an intelligent non-signalized intersection, there is at least one CAV-only entrance lane at the entrance; formula (24) indicates that if the intersection is an intelligent non-signalized intersection, no ordinary entrance lane is set; otherwise, all entrance lanes other than the CAV-only entrance lane are ordinary entrance lanes; since the CAV-only lanes and ordinary lanes of the same road section (i, j) are connected to the same entrance, as shown in formulas (25-26), the layout of the CAV-only entrance lanes and ordinary entrance lanes is the same downstream of the ordinary lanes; The relationship between CAV lane types and HV service fees In formula (27), ω ij is a 0-1 variable indicating whether a CAV lane is set up on the road section (i, j). It takes 1 if a CAV lane is set up, otherwise it takes 0. Formula (28) stipulates that if a CAV lane is not set up on the road section (i, j), the CAV lane type variable = 0; Formula (29) stipulates that if the CAV lane is a regular CAV lane, the charging variable τ ij is 0; Relationship between signal cycle and green light time of each phase Formula (30) indicates that if intersection j is not a signalized intersection with a CAV-dedicated phase and a CAV-dedicated entrance, the duration of the CAV-dedicated phase is 0. Formulas (31-33) indicate the relationship between the upper and lower bounds of the cycle duration and the upper and lower bounds of the green light duration of each phase for signalized intersections, where w is the green light interval, including the yellow light time and the full red time. Formula (34) calculates the green light duration of the normal signal phase of intersection j corresponding to the entrance (i, j); where is a 0-1 variable indicating whether the entrance (i, j) corresponds to the hth common phase of intersection j; Step S4: Calculate travel cost; Road section travel time calculation In formula (35-36), u ij The traffic capacity of pure HVs on CAV lanes and ordinary lanes is equal to the number of CAV lanes on road section (i, j). Number of ordinary lanes Multiply the average safe headway time of a HV following a HV by In formula (37), the BPR function is used to calculate the time t required for a vehicle to pass through the road section (i, j) ij , It represents the time required for a vehicle to freely pass through the road section (i, j), α and β are two positive parameters in the BPR function. are the HV and CAV traffic on road section (i, j), pcu ij is the conversion coefficient of CAV flow to HV flow; in formula (38), according to the formula derivation, the conversion coefficient of CAV flow to HV flow pcu is obtained ij , is a ratio of CAV traffic to mixed traffic flow p ij Related functions; among them, is the average safe headway time between CAV and HV, is the average safe headway between HV and CAV, is the average safe headway time between CAV and CAV; Formula (39) shows that the proportion of CAV flow in mixed traffic flow on road section (i, j) is p. ij It is equal to the CAV flow rate on road segment (i, j) divided by the sum of CAV flow rate and HV flow rate; Calculation of entry delay time 1) Inlet flow constraint Formula (40) indicates that the CAV flow at the intersection entrance downstream of the CAV lane on road section (i, j) is equal to the sum of the CAV flow on the CAV lane and the ordinary lane, and the same is true for HV. Formula (41) indicates that the ordinary lane and the CAV lane downstream on road section (i, j) have the same flow at the intersection entrance. Formula (42) calculates the conversion coefficient of CAV flow to HV flow on the entrance road (i, j) 2) Calculation of delay time for ordinary entrance In formula (43), λ ij is a 0-1 variable indicating whether the common inlet channel (i, j) is oversaturated. If it is oversaturated, it takes 1, otherwise it takes 0. Formula (44) calculates the delay time d of the common inlet channel (i, j) ij ; Among them, o ij is a non-negative continuous variable, representing the saturation of the common inlet (i, j); T is a fixed parameter, representing the duration of the derived flow; Equation (45) establishes λ by introducing a sufficiently large number M ij Saturation o of the common inlet (i, j) ij If the saturation of the common inlet channel (i, j) is o ij >1,λ ij Taking 0 will not satisfy the constraint, so it can only be 1; if the saturation of the common inlet channel (i, j) is o ij ≤1,λ ij Taking 1 will not satisfy the constraint, so it can only be taken as 0; Equation (46) calculates the saturation o of the common inlet channel (i, j) ij , where the numerator is the equivalent HV flow at the common inlet, and the denominator is the vehicle dissipation capacity of the common inlet; X is a fixed parameter, representing the saturation flow rate of the HV flow through a common inlet; Calculation of CAV dedicated entrance lane delay time under CAV dedicated phase: In formula (47), is a 0-1 variable, indicating whether the CAV dedicated entrance lane (i, j) in the CAV dedicated phase is oversaturated. If it is oversaturated, it takes 1, otherwise it takes 0. Equation (48) calculates the delay time of the CAV dedicated entrance lane (i, j) in the CAV dedicated phase. in, is a non-negative continuous variable, representing the saturation of the CAV dedicated inlet channel (i, j) under the CAV dedicated phase; Equation (49) is established by introducing a sufficiently large number M. Saturation of CAV dedicated inlet channel (i, j) under CAV dedicated phase If the saturation of the CAV dedicated inlet channel (i, j) under the CAV dedicated phase is Taking 0 will not satisfy the constraint, so it can only take 1; if the saturation of the CAV dedicated inlet channel (i, j) in the CAV dedicated phase Taking 1 will not satisfy the constraint, so it can only be taken as 0. Equation (50) calculates the saturation of the CAV dedicated inlet channel (i, j) under the CAV dedicated phase The numerator is the CAV flow rate at the CAV dedicated entrance, and the denominator is the vehicle dissipation capacity of the CAV dedicated entrance; is a fixed parameter, representing the saturation flow rate of CAV traffic passing through a CAV-dedicated entrance lane, and ρ is the improvement coefficient of the intersection traffic efficiency of the unsignaled AIM control strategy compared with the traditional signalized intersection; Calculation of Delay Time for CAV-Dedicated Entrance Lanes at Intelligent Unsignalized Intersections In formula (51), ξ ij is a 0-1 variable, indicating whether the CAV dedicated entrance lane (i, j) of the intelligent unsignalized intersection is oversaturated. If it is oversaturated, it takes 1, otherwise it takes 0. Formula (52) calculates the delay time of the CAV dedicated entrance lane (i, j) of the intelligent unsignalized intersection Formula (53) introduces a sufficiently large number M to establish ξ ij and relationship, if ξ ij Taking 0 will not satisfy the constraint, so it can only take 1; if ξ ij Taking 1 will not satisfy the constraint, so it can only take 0; Travel cost calculation 1) Calculation of travel cost by road section Equation (54) calculates the travel cost of an HV on a road section (i, j) through a CAV lane: The travel cost of HVs using CAV lanes is calculated based on the following conditions: the intersection is an intelligent non-signalized intersection, there is no CAV lane on the road section, and the CAV lane is a conventional CAV lane. is a sufficiently large number M, otherwise, the travel cost of HV through the CAV lane is is the travel time t through the CAV lane ij and the fee charged τ ij The sum of the values, multiplied by the time value coefficient κ of HV HV , Equation (55) calculates the travel cost of a CAV on a road section (i, j) through a CAV lane: When there is no CAV lane on road section (i, j), the travel cost of a CAV through the CAV lane is is a sufficiently large number M, otherwise, the travel cost of CAV through the CAV lane is the travel time t through the CAV lane ij Multiply by the time value coefficient κ of CAV CAV , Equation (56) calculates the travel cost of HV and CAV through the ordinary lane on road section (i, j) When the intersection is an intelligent non-signalized intersection, the travel cost of HV and CAV through the ordinary lane is a sufficiently large number M, otherwise, the travel cost of HV and CAV through the ordinary lane are the travel time t through the ordinary lane ij Multiply by the time value coefficient κ of HV and CAV m ; Calculation of travel costs at intersection entrances Equation (57) calculates the travel cost of an HV on the CAV lane (i, j) passing through intersection j: If intersection j is an intelligent non-signalized intersection, HVs are prohibited from passing through the intelligent non-signalized intersection. Therefore, the travel cost of an HV on the CAV lane (i, j) passing through intersection j is a sufficiently large number M. Otherwise, it is the delay time d of an HV on the CAV lane (i, j) passing through the ordinary entrance lane (i, j). ij Multiply by the time value coefficient κ of HV HV , Equation (58) calculates the travel cost of a CAV on the CAV lane (i, j) passing through intersection j If intersection j is a traditional signalized intersection, the travel cost of a CAV on the CAV lane (i, j) passing through intersection j is is the delay time d of the CAV on the CAV dedicated lane (i, j) passing through the common entrance lane (i, j) ij Multiply by the time value coefficient κ of CAV CAV If intersection j is a signalized intersection with a CAV-only phase and a CAV-only entrance lane, the travel cost of a CAV on the CAV-only lane (i, j) through intersection j is is the delay time of the CAV on the CAV lane (i, j) passing through the CAV entrance lane (i, j) during the CAV phase Multiply by the time value coefficient κ of CAV CAV ; If intersection j is an intelligent non-signalized intersection, the travel cost of a CAV on the CAV lane (i, j) passing through intersection j is is the delay time of the CAV on the CAV lane (i, j) passing through the CAV entrance lane (i, j) Multiply by the time value coefficient κ of CAV CAV , Formula (59) indicates that HV and CAV on the ordinary lane and CAV lane on the road section (i, j) have the same travel cost when passing through intersection j Step S5: constructing flow distribution constraint conditions; Sub-step S51: Constructing non-negative flow and demand constraints Formula (60) indicates that the HV and CAV flows on the road segment (i, j) to the end point s are non-negative. Formulas (61-64) indicate that the node flow conservation is true. In the road network, for the end point s, the flow into the end point s should be equal to the total attraction of the node s; for the starting point r, the flow out of the starting point r and ending at the node s is the travel demand between the OD pair; for the intermediate node k, the flow out of the node k and ending at s should be equal to the flow into the node k and ending at s; Sub-step S52: Constructing user balance constraints under mixed HV and CAV traffic conditions The UE equilibrium constraint is constructed based on the link-node modeling method: when equilibrium is reached, if a road segment (i, j) is on the path from node i to end point s and there is HV or CAV traffic on the road segment, then the travel cost of the HV or CAV on the road segment (i, j) is equal to the minimum travel cost from node i to end point s minus the minimum travel cost from node j to end point s; Formula (65-66) represents the UE balance constraint of HV on the road network; Formula (67-68) represents the UE balance constraint of CAV on the road network; HV and CAV traffic on the road section HV and CAV traffic on roads with different end points The relationship between and They represent the minimum travel cost of HV and CAV from node j to destination s respectively; Step S6: Construct the DISLP problem model with the minimum total travel time as the goal; The problem model is: The constraints are a), b), c) a) Decision variables and the relationships between them in the DISLP problem; b) Calculation of travel costs for HV and CAV; c) HV and CAV traffic distribution constraints; Step S7: Design an improved genetic algorithm to solve the DISLP problem; Sub-step S71: Considering the accessibility of HVs in the road network, generate a set of initial feasible solutions; Sub-step S72: Bring the feasible solution into the GAMS solver to solve a network traffic distribution sub-problem, calculate the objective function value of the original problem corresponding to the feasible solution, and sort according to the obtained original problem objective value; Sub-step S73: selecting the father and mother generations according to the principle that the smaller the target value, the greater the probability of becoming the father and mother generations, and performing genetic mutation to generate offspring; Sub-step S74: determining whether the generated offspring meet the accessibility requirements of the HV in the road network. If the offspring do not meet the requirements, they will be replaced by newly generated offspring that meet the accessibility requirements of the HV until all the generated offspring meet the accessibility requirements of the HV in the road network. Sub-step S75: Return to step S72 until the maximum evolutionary generation is reached, and the program terminates.

Images