A mixed traffic distribution method considering endogenous selection of vehicle types and a medium

Abstract

The application provides a mixed traffic distribution method and medium considering endogenous selection of vehicle types, and the method comprises the following steps: constructing a mixed traffic network; for each starting point, calculating a generalized cost according to a total cost per unit distance, a time value parameter and an expected average travel time of each vehicle type, and then determining a selection probability of each vehicle type; calculating travel demand of each vehicle type of each starting point-terminal point pair in combination with travel demand of each starting point-terminal point pair; taking a road section flow based on a starting point and travel demand divided according to vehicle types as joint decision variables, using a road section-node based modeling method, and uniformly representing vehicle type selection conditions and traffic distribution equilibrium conditions of each vehicle type as a variational inequality problem; iteratively solving to obtain travel demand of each vehicle type and equilibrium flow distribution in the road network. The application can endogenously determine vehicle type demand, avoid path enumeration, and is suitable for traffic distribution analysis of large-scale mixed traffic road networks.

Description

Technical Field

[0001] This invention relates to the field of mixed traffic assignment technology, specifically providing a mixed traffic assignment method and medium that considers the endogenous selection of vehicle type. Background Technology

[0002] With the development of intelligent connected and autonomous driving technologies, connected vehicles (CVs), autonomous vehicles (AVs), and connected-autonomous vehicles (CAVs) will gradually enter urban road traffic systems. Due to factors such as vehicle purchase costs, user acceptance, and infrastructure coverage, manually driven vehicles (HVs) will be difficult to completely replace for a considerable period. Therefore, manually driven vehicles, connected vehicles, autonomous vehicles, and connected-autonomous vehicles will coexist in road networks for a long time, forming a heterogeneous road network environment where different vehicle types are mixed.

[0003] Different vehicle types, due to the different technologies they employ, exhibit significant differences in information acquisition methods, control methods, and decision-making mechanisms. Manually driven vehicles primarily rely on driver experience for driving and route selection; connected vehicles, utilizing vehicle-to-everything (V2X) technology, can acquire real-time road network traffic information but still require human driver intervention; autonomous vehicles, relying on autonomous driving technology, possess autonomous perception and driving decision-making capabilities, but their collaborative interaction capabilities are limited; connected-autonomous vehicles, on the other hand, incorporate both V2X and autonomous driving technologies, possessing both real-time road network perception and autonomous driving capabilities. These differences lead to more complex road network operation characteristics under mixed traffic conditions and make the analysis of route decision-making and traffic flow spatial distribution characteristics more complex for different vehicle types.

[0004] Predicting the spatial distribution of traffic flow in road networks using traffic assignment is fundamental to road network-level traffic planning and management. Existing methods for analyzing mixed-traffic road networks still fall short in characterizing the operational characteristics of road networks under long-term coexistence of multiple vehicle types. They typically only consider mixed-traffic scenarios involving HV and CAV vehicles, and often treat the travel demand of different vehicle types as pre-given parameters, failing to reveal the endogenous mechanisms of traveler vehicle type selection in multi-vehicle-type scenarios. However, in reality, travelers use factors such as travel costs for different vehicle types as the basis for their purchasing decisions, resulting in travel demand for different vehicle types not being an externally given constant. Therefore, existing methods struggle to accurately analyze and predict the spatial distribution of traffic flow in multi-vehicle-type mixed-traffic scenarios. Summary of the Invention

[0005] This invention addresses the shortcomings of existing technologies by providing a mixed traffic assignment method that considers the endogenous selection of vehicle types. This method aims to solve or partially solve the problem that, under the long-term coexistence and mixed traffic conditions of HV, AV, CV, and CAV, existing traffic assignment methods typically treat the travel demand or penetration rate of the four types of vehicles as exogenous given parameters, failing to characterize the feedback mechanism between vehicle type selection and road network travel time. Consequently, it is difficult to simultaneously determine the travel demand of the four types of vehicles and the distribution of corresponding traffic flows in the road network. At the same time, the number of feasible paths grows rapidly under large-scale road networks, significantly increasing the model size and solution difficulty.

[0006] To achieve the above objectives, in a first aspect, the present invention provides a mixed traffic allocation method that considers the endogenous selection of vehicle types, comprising the following steps: constructing a mixed traffic network, wherein the mixed traffic network includes at least manually driven vehicles (HV), autonomous vehicles (AV), connected vehicles (CV), and connected autonomous vehicles (CAV); For each starting point, the generalized cost of each vehicle type at that starting point is calculated based on the total cost of ownership per unit distance, time value parameters, and expected average travel time. Then, the selection probability of each vehicle type is determined based on the generalized cost. Based on the travel demand of each origin-destination pair and the selection probability of each vehicle type, calculate the travel demand of each origin-destination pair for each vehicle type. Using the segment flow based on the origin and the travel demand divided by vehicle type as joint decision variables, and adopting a segment-node-based modeling approach, the vehicle type selection conditions and the traffic allocation equilibrium conditions for each vehicle type are uniformly expressed as a variational inequality problem. The variational inequality problem is solved iteratively to obtain the travel demand of each vehicle type and the equilibrium flow distribution of each vehicle type on each road segment in the mixed traffic network.

[0007] Preferably, the process of constructing a mixed traffic network includes: Representing the mixed traffic network as a directed graph Where N is the set of nodes and A is the set of road segments; Define a starting set R and an ending set S, and , and define the vehicle type set. ; For each start-end pair (r, s), define For the travel demand of this origin-destination pair, Let m be the travel demand of vehicle type m on the origin-destination pair (r,s), and satisfy: Where r and s represent the starting point and the ending point, respectively, and m represents the vehicle type; The sum of the starting point and ending point requirements from the starting point r to all ending points. That is, the travel demand of the starting point r: Vehicle type m on the road segment Traffic This is the sum of the traffic flow for this vehicle type at each starting point on the road segment: in, This indicates the vehicle type m starting from the origin r on the road segment. Traffic on the internet; Section Total flow The sum of traffic flow for each vehicle type on this road segment: .

[0008] Preferably, for each origin, the generalized cost of each vehicle type at that origin is calculated based on the total cost of ownership per unit distance, time value parameters, and expected average travel time, including: For each starting point r and each vehicle type m, the generalized cost is calculated as follows: Total cost of ownership per unit distance based on vehicle type m Average travel distance from the starting point r Determine the average monetary cost of a single trip for vehicle type m at origin r. : Based on the time value parameter of vehicle type m Under network equilibrium, the expected average travel time of vehicle type m starting from origin r. Determine the average trip time cost for vehicle type m at origin r. ; Adding the average monetary cost per trip to the average time cost per trip yields the generalized cost of vehicle type m at the origin r. : .

[0009] Preferably, the method further includes obtaining the average travel distance of the starting point r. Specifically, it includes: For a starting point r, the average travel distance is calculated based on the weighted average of the starting-to-destination demand corresponding to r: in, This indicates the start-end pair. The average travel distance.

[0010] Preferably, the method further includes obtaining the expected average travel time. Specifically, it includes: The calculation is based on the weighted average traffic flow and travel time of this vehicle type on each road segment: in, Let r represent the travel demand of vehicle type m at the starting point r. Indicates road segment Travel time.

[0011] Preferably, determining the selection probability of each vehicle type based on the generalized cost includes: Using generalized cost as the basis for vehicle type selection, a multinomial Logit model is employed to calculate the probability that a traveler at origin r will choose vehicle type m: in, This is a scaling parameter used to represent the sensitivity of travelers to generalized cost differences between different vehicle types. Let k be the vehicle type set, and k be the vehicle type index. Let r represent the generalized cost of vehicle type k at the starting point r.

[0012] Preferably, based on the travel demand of each origin-destination pair and the selection probability of each vehicle type, the travel demand of each origin-destination pair for each vehicle type is calculated, including: For the origin-end pair Travel demand for vehicle type m Represented as: The travel demand for vehicle type m at the starting point r is obtained by summing the travel demands for vehicle type m from that starting point to all destinations, that is: .

[0013] Preferably, using the segmental traffic flow based on the origin and the travel demand categorized by vehicle type as joint decision variables, a segment-node-based modeling approach is adopted to uniformly represent the vehicle type selection conditions and the traffic allocation equilibrium conditions for each vehicle type as a variational inequality problem, including: Establish node flow conservation constraints and non-negativity constraints: in, This represents the net inflow of vehicle type m originating from r at node j. This indicates the vehicle type m starting from the origin r on the road segment. Traffic on the internet; Establish traffic allocation equilibrium conditions for each vehicle type: For manually driven vehicles (HV) and autonomous vehicles (AV), following the Stochastic User Equilibrium (SUE) principle, we define the vehicle type m relative to the origin r on the road segment. The generalized road segment cost is: in, Indicates road segment Travel time, This indicates the vehicle type m starting from the origin r on the road segment. Traffic on the internet , This represents the flow of vehicle type m originating from origin r and flowing into node j. Let j be the index of the predecessor node of node j. The perception error scale parameter for human-driven vehicles (HV) and autonomous vehicles (AV) in path selection; like Then the generalized segment cost of that segment is equal to the minimum perceived generalized cost difference between adjacent nodes; if Then the generalized segment cost of this segment is not less than the minimum perceived generalized cost difference between adjacent nodes, specifically expressed as: in, Indicates the cost of a generalized road segment The value at the equilibrium solution, and Let represent the minimum perceived generalized cost of vehicle type m from starting point r to node i and node j, respectively, under equilibrium conditions. This indicates that in equilibrium, vehicle type m starting from point r is on road segment Traffic on the internet; For connected vehicles, following the user-balanced user (UE) principle, the CV is defined relative to the starting point r on the road segment. The generalized road segment cost is: like Then the actual generalized segment cost of this segment is equal to the minimum actual generalized cost difference between adjacent nodes; if If the travel cost of this route segment is not less than the minimum actual generalized cost difference between adjacent nodes, it can be expressed as follows: in, Indicates the cost of a generalized road segment The value at the equilibrium solution, and Let represent the minimum actual generalized cost of CV reaching node i and node j from the starting point r, respectively, in the equilibrium state. This represents the CV starting from the origin r in the equilibrium state on the road segment. Traffic on the internet; For connected autonomous vehicles (CAVs), following the system optimality (SO) principle, the CAV is defined relative to the starting point r on the road segment. The generalized road segment cost is: in, Indicates CAV on the road segment Traffic on the internet Indicates road segment travel time The partial derivative of the CAV flow rate on this road segment; like Then the generalized segment cost of that segment is equal to the minimum marginal generalized cost difference between adjacent nodes; if If the cost of the generalized road segment is not less than the minimum marginal generalized cost difference between adjacent nodes, it can be expressed as follows: in, Indicates the cost of a generalized road segment The value at the equilibrium solution, and These represent the CAV starting from the point of equilibrium. The minimum marginal generalized cost to reach nodes i and j. This indicates that the CAV starting from the origin r in the equilibrium state is located on the road segment. Traffic on the internet; Finally, the vehicle type selection conditions and the traffic allocation equilibrium conditions for each vehicle type are uniformly expressed as the following variational inequality problem, where K is the joint feasible region determined by the node flow conservation constraint, the non-negativity constraint, and the vehicle type travel demand constraint: Solve , making The first three terms correspond to the stochastic user equilibrium conditions for manually driven vehicles (HV) and autonomous vehicles (AV), the user equilibrium conditions for connected vehicles (CV), and the system optimal conditions for connected autonomous vehicles (CAV), respectively. The last term corresponds to the vehicle type selection condition based on the starting point. Indicating equilibrium state The value of represents the travel demand of vehicle type m at the starting point r under equilibrium conditions.

[0014] Preferably, the travel time for the aforementioned route is calculated using a method based on passenger car equivalents, specifically including: Introducing passenger car equivalent conversion factor This is used to convert the flow rate of vehicle type m into the standard passenger car equivalent flow rate: in, For road section free flow velocity, Let m be the reaction time for vehicle type m. HV is the reaction time, and L is the vehicle length; Road sections calculated using the U.S. Highway Bureau's Road Resistance Function (BPR) Travel time on road sections with mixed traffic conditions: in, For road section Free-flow travel time, and For the parameters of the BPR function, This indicates the road section under conditions of purely manual driving. The baseline traffic capacity.

[0015] The beneficial effects of this invention are as follows: By constructing a mixed traffic assignment model that considers the endogenous selection of vehicle types, this invention can determine the travel demand of four types of vehicles and the distribution of traffic flow, total traffic flow, and travel time of the four types of vehicles in each road segment. This model can be used to analyze the spatial distribution characteristics of mixed traffic flow of the four types of vehicles in the road network. Furthermore, this invention adopts a segment-node-based modeling approach, which does not rely on explicit enumeration of feasible paths, thereby reducing the modeling and solution complexity of traffic assignment problems in large-scale mixed road networks. Attached Figure Description

[0016] The disclosure of this invention will become more readily understood with reference to the accompanying drawings. It will be readily understood by those skilled in the art that these drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. Furthermore, similar numbers in the drawings are used to denote similar components, wherein: Figure 1 This is a schematic diagram of the iterative solution process of a mixed traffic assignment method considering the endogenous selection of vehicle type according to an embodiment of the present invention.

[0017] Figure 2 This is an embodiment of the Nguyen-Dupuis test road network of the present invention.

[0018] Figure 3 This is a test road network segment total traffic spatial distribution according to an embodiment of the present invention. Detailed Implementation

[0019] Some embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0020] Example 1 like Figure 1-3 As shown, the present invention provides a mixed traffic assignment method that considers the endogenous selection of vehicle type, including the following steps: Step S1: Model the traffic network for scenarios where human-driven vehicles (HV), autonomous vehicles (AV), connected vehicles (CV), and connected autonomous vehicles (CAV) coexist and mix in the long term.

[0021] Mixed traffic networks are represented as directed graphs. Where N is the set of nodes, A is the set of road segments, R and S are the set of starting points and the set of ending points, respectively. , The vehicle type set is as follows: .

[0022] For the origin-end pair , This indicates the travel demand for that origin-destination pair. This represents the travel demand for vehicle type m in this origin-destination pair. The travel demand for this origin-destination pair consists of the travel demand for four vehicle types, therefore: (1) Where r and s represent the starting point and the ending point, respectively; m represents the vehicle type.

[0023] Furthermore, Let the travel demand at origin r be equal to the sum of the origin-to-destination demands for all journeys from that origin to all destinations, i.e.: (2) Section The flow rate of vehicle type m is the sum of the flow rates of that vehicle type on each road segment starting from each origin. Indicates vehicle type m on road segment Traffic on the internet This indicates that vehicle type m, originating from r, is on the road segment. The traffic on the internet includes: (3) Accordingly, Indicates road segment The total traffic flow on this road segment consists of the traffic flow of four types of vehicles: (4) Step S2: To reflect the impact of mixed traffic of four types of vehicles on road segment capacity and travel time, this invention adopts a road segment travel time calculation method based on passenger car equivalent PCU. Since HV, AV, CV, and CAV vehicles differ in operational parameters such as reaction time, and different vehicle types occupy road segment capacity to varying degrees, the traffic flow of the four types of vehicles is converted to standard passenger car equivalent traffic flow using a PCU conversion factor. Based on this, the road segment travel time is calculated using the U.S. Federal Highway Administration's road resistance function.

[0024] To reflect the impact of different vehicle types' reaction time differences on road segment capacity, a passenger car equivalent conversion factor is introduced. This is used to convert the flow rate of vehicle type m into the standard passenger car equivalent flow rate, that is: (5) in, For road section free flow velocity, Let m be the reaction time for vehicle type m. HV represents the reaction time, and L represents the vehicle length.

[0025] Under pure HV conditions, road section Baseline traffic capacity It can be represented as: (6) in, This refers to the number of lanes in the road segment.

[0026] After converting the traffic flow of the four vehicle classes to the standard passenger car equivalent traffic flow using the passenger car equivalent conversion factor, the road block function of the U.S. Federal Highway Administration is used to calculate the traffic flow of the road segment. Travel time under mixed traffic conditions is specifically expressed as follows: (7) in, Free-flow travel time for the road segment and These are the road resistance function parameters for the U.S. Federal Highway Administration.

[0027] Step S3: In order to determine the probability of travelers choosing different vehicle types from the same starting point and to further obtain the travel demand for different vehicle types for each starting point-end point pair, this invention establishes a vehicle type selection model for each starting point r, and uses HV, AV, CV and CAV as four types of vehicles that travelers can choose.

[0028] The generalized cost of vehicle type m at origin r consists of the average monetary cost per trip and the time cost, where the time cost is calculated by discounting the expected average trip time. Therefore, the generalized cost can be expressed as: (8) in, This represents the generalized cost of vehicle type m at the starting point r. This represents the average monetary cost of a single trip for vehicle type m at the origin r. This represents the expected average travel time of vehicle type m starting from origin r under network equilibrium conditions. This represents the time value parameter for the corresponding vehicle type m, used to convert the expected average travel time into time cost.

[0029] Specifically, the total cost of ownership per unit distance for this vehicle type Average travel distance from the starting point r Jointly determine the average monetary cost of a single trip for vehicle type m at origin r. ,Right now: (9) Among them, the total cost of ownership per unit distance This refers to the cost of converting vehicle purchase, depreciation, energy consumption, insurance, and other usage costs into the unit of driving distance.

[0030] For the origin-end pair , This indicates the start-end pair. If the average travel distance of origin r is given, then the average travel distance of origin r can be calculated by weighting the demand of each origin-destination pair corresponding to that origin as follows: (10) The expected average travel time of vehicle type m at origin r .

[0031] (11) in, Let r represent the travel demand of vehicle type m at the starting point r. Indicates travel time for a particular route.

[0032] According to equation (8), the generalized cost of each vehicle type at each starting point can be calculated. Using generalized cost as the basis for vehicle type selection, a multinomial Logit model is employed to calculate the probability that a traveler at origin r will choose vehicle type m. Specifically, it is expressed as: (12) in, This is a scaling parameter used to represent the sensitivity of travelers to generalized cost differences between different vehicle types. Let k be the vehicle type set, and k be the vehicle type index. Let r represent the generalized cost of vehicle type k at the starting point r.

[0033] For the origin-end pair The travel demand for vehicle type m is determined by the origin-destination travel demand. The probability of choosing vehicle type m at the starting point r This was jointly determined, specifically as follows: (13) The travel demand for vehicle type m at the starting point r can be obtained by summing the travel demands for vehicle type m from that starting point to all destinations, that is: (14) Therefore, the travel demand for the four types of vehicles is no longer a pre-given parameter, but is determined by the vehicle purchase and use costs, time value parameters, and network equilibrium travel time, and is calculated through the probability of vehicle type selection.

[0034] Step S4: Considering the selection of four vehicle types (HV, AV, CV, and CAV), this invention establishes a mixed traffic assignment model that considers the endogenous selection of vehicle types to determine the distribution of mixed traffic flow in the road network. This model adopts a segment-node-based modeling approach, using segment flow based on the origin and travel demand categorized by vehicle type as joint decision variables. It also unifies the vehicle type selection conditions with the traffic assignment equilibrium conditions for the four vehicle types in a unified model, thereby determining the travel demand of each vehicle type and its flow distribution in the road network, and avoiding explicit enumeration of feasible paths.

[0035] Regarding path selection behavior, this invention assumes that HV and AV follow the Stochastic User Equilibrium (SUE) principle, CV follows the User Equilibrium (UE) principle, and CAV follows the System Optimum (SO) principle. Based on the above behavioral settings, flow conservation constraints and equilibrium conditions corresponding to the four types of vehicles are established respectively, and the vehicle type selection conditions and traffic assignment equilibrium conditions are further unified into variational inequality problems.

[0036] (1) Node flow conservation constraints and nonnegativity constraints In the segment-node-based modeling approach, segment traffic flow is based on the starting point. The node flow conservation constraint and non-negativity constraint must be satisfied, specifically as follows: (15) (16) in, This represents the net inflow of vehicle type m originating from r at node j. This indicates the vehicle type m starting from the origin r on the road segment. Traffic on the internet.

[0037] Net inflow at node j It can be represented as: (17) in, This represents the travel demand for vehicle type m from the starting point r to the destination j.

[0038] (2) Stochastic user equilibrium conditions for HV and AV For manually driven vehicles (HV) and autonomous vehicles (AV), considering the perception error in their path selection process, their path selection follows the Stochastic User Equilibrium (SUE) principle. Define the vehicle type m relative to the starting point r on the road segment... The generalized road segment cost is: (18) in, Indicates road segment Travel time on This indicates the vehicle type m starting from the origin r on the road segment. Traffic on the internet , This represents the flow of vehicle type m originating from origin r and flowing into node j. Let j be the index of the predecessor node of node j. The perception error scale parameter for human-driven vehicles (HV) and autonomous vehicles (AV) in path selection; In equilibrium, if Then the generalized segment cost of that segment is equal to the minimum perceived generalized cost difference between adjacent nodes; if Then the generalized segment cost of this segment is not less than the minimum perceived generalized cost difference between adjacent nodes, specifically expressed as: (19) in, Indicates the cost of a generalized road segment The value at the equilibrium solution, and Let represent the minimum perceived generalized cost of vehicle type m from starting point r to node i and node j, respectively, under equilibrium conditions. This indicates that in equilibrium, vehicle type m starting from point r is on road segment Traffic on the internet.

[0039] In this specific embodiment, to demonstrate the consistency between the multiple Logit random user equilibrium form satisfied by the HV / AV path selection probability and the segment-node form expression constituted by equations (18) and (19) in this invention, it can be achieved and verified through the following steps.

[0040] First, verify the sufficiency condition: Based on the HV / AV generalized segment cost definition given in equation (18), if the nodal potential equilibrium condition shown in equation (19) holds. Next, we prove that equations (18) and (19) can derive a multinomial Logit stochastic user equilibrium form at the path level. For any condition satisfying... section of road From equations (18) and (19), we can obtain: For ease of explanation, a new variable is introduced. It is defined as HV / AV relative to the starting point. Select road segment The probability is calculated as follows: Substituting equation (21) into equation (20), it can be transformed into the following equation: make The conditional probability on the left side of equation (22) is expressed as follows: Indicates the successor node after a given choice. Under these conditions, HV / AV relative to the starting point Select node The probability of, where . Indicates HV / AV from the starting point To the finish line Select path The probability of, where Since random assignment based on the Logit model has the Markov property, therefore... in, It is a binary variable used to represent a road segment. Is it located in HV / AV from the start point? To the finish line path Above, among which Specifically, when the road section Located from the starting point To the finish line path When, ;otherwise, . This represents the set of feasible paths between the start-end pair (r, s).

[0041] Substituting equation (22) into equation (24), we obtain the path selection probability satisfying the following equation: in, Indicates vehicle type Choose a path between the start-end pair (r, s). The cost of travel along a route is determined by the route itself. The total travel time for each section of the route is calculated by adding up the travel times. From the starting point To the finish line Summing all paths on both sides of equation (25), we have Therefore, we can obtain Substituting equation (28) into equation (25), we get in, Indicates vehicle type Choose a path between the start-end pair (r, s). The probability of.

[0042] From equation (29), it can be seen that under the HV / AV generalized segment cost defined by equation (18), if the segment-node equilibrium relationship shown in equation (19) can lead to a multinomial Logit stochastic user equilibrium at the path level, then equations (18) and (19) together constitute a sufficient condition for HV / AV stochastic user equilibrium in the segment-node form. Thus, the sufficiency proof is complete.

[0043] Next, we will prove the necessary condition for the above proposition. Assume... The equilibrium segment flow is obtained by solving the multinomial Logit random user equilibrium at the path level. According to equation (18) and the conditional probability defined by equation (23) on the left side of equation (22), we have... From the starting point To the finish line path Generalized travel costs for all routes Summing, we can get As mentioned above, random assignment based on the Logit model has the Markov property. Substituting equation (24) into the right side of equation (31), we can obtain... in, For path set The path index in the file.

[0044] Since equation (32) holds true for any path between all OD pairs, we select a path starting from the origin. To the node path ,in . It is a road starting from the beginning To the node The path, which passes through the path in sequence and road sections ,in Therefore, we can obtain the following equation: in, and They represent the starting point. To the node and nodes The set of feasible paths and Representing vehicle type Along the path From the starting point To the node and nodes The cost of traveling along the route.

[0045] Then, subtracting equation (34) from equation (33) gives the road segment. The generalized cost of travel, namely As shown below: For any starting point and vehicle type The first term on the right side of equation (35) is only related to the node. The second item is only related to the node. This is relevant. This is consistent with the meaning expressed by equation (19). Therefore, it can be seen that under the HV / AV generalized segment cost defined by equation (18), the multinomial Logit stochastic user equilibrium at the path level can derive the segment-node equilibrium relationship shown in equation (19). Therefore, equations (18) and (19) together constitute the necessary condition for HV / AV stochastic user equilibrium in the segment-node form. Thus, the proof of the necessity condition is complete.

[0046] (3) User equilibrium condition of CV For the CV (Cyclic Component Controller), the path selection result is determined based on the acquired road network traffic information and the User Equalization (UE) principle. The CV is defined relative to the starting point r on the road segment... The generalized road segment cost is: (36) In equilibrium, if Then the generalized segment cost of that segment is equal to the minimum actual generalized cost difference between adjacent nodes; if If the generalized segment cost of the road segment is not less than the minimum actual generalized cost difference between adjacent nodes, it can be expressed as follows: (37) in, Indicates the cost of a generalized road segment The value at the equilibrium solution, and Let represent the minimum actual generalized cost of CV reaching node i and node j from the starting point r, respectively, in the equilibrium state. This represents the CV starting from the origin r in the equilibrium state on the road segment. Traffic on the internet.

[0047] (4) System optimal conditions for CAV For CAVs, based on their cooperative control capabilities, their path selection results are determined according to the system optimal SO principle. The CAV is defined relative to the starting point r on the road segment... The generalized road segment cost is: (38) in, Indicates CAV on the road segment Traffic on the internet Indicates travel time for the route. The partial derivative of the CAV flow rate on this road segment; In equilibrium, if Then the cost of the generalized road segment is equal to the minimum marginal generalized cost difference between adjacent nodes; if If the cost of the generalized road segment is not less than the minimum marginal generalized cost difference between adjacent nodes, it can be expressed as follows: (39) in, Indicates the cost of a generalized road segment The value at the equilibrium solution, and These represent the CAV starting from the point of equilibrium. The minimum marginal generalized cost to reach nodes i and j. This indicates that the CAV starting from the origin r in the equilibrium state is located on the road segment. Traffic on the internet.

[0048] (5) Mixed traffic assignment model in variational inequality form Based on the above-mentioned flow conservation constraints, the definition of net inflow to nodes, the path selection conditions for the four types of vehicles, and the vehicle type selection model, this invention expresses vehicle type selection and mixed traffic assignment in a unified manner as the following joint variational inequality problem.

[0049] Joint decision variables are represented as Where x represents the road segment traffic flow variable divided by origin and vehicle type, and q represents the travel demand variable divided by origin and vehicle type. The joint feasible region K is defined as: (40) The variational inequality problem can then be expressed as: Solving , making (41) In Equation (41), the first three terms correspond to the stochastic user equilibrium conditions for HV and AV, the user equilibrium conditions for CV, and the system optimal conditions for CAV, respectively, while the last term corresponds to the vehicle type selection condition based on the starting point. By solving the variational inequality problem expressed by Equation (41), the travel demand of the four types of vehicles and the equilibrium traffic distribution in the mixed traffic network can be obtained.

[0050] Step S5: To solve the vehicle type selection and mixed traffic assignment model in the form of the variational inequality problem, this invention employs a nested iterative method: the inner layer (mixed traffic assignment model) updates the segment flow based on the origin using the successive mean squared average (MSA) method; the outer layer (vehicle type selection model) updates the vehicle type selection probability using a fixed-point iterative method with a relaxation factor. The specific process is as follows: Figure 1 As shown.

[0051] Step S51: Initialization.

[0052] Initialize the outer iteration count n=0, and determine the inner convergence threshold. Outer layer convergence threshold And the maximum number of iterations; given the initial vehicle type selection probability at each starting point. ,Depend on Determine the initial vehicle type travel demand for each origin-destination pair and initialize the segment travel time according to free-flow conditions. .

[0053] Step S52: Load inner network traffic.

[0054] Given the current travel demand for the vehicle type Travel time by road segment Under these conditions, inner-layer calculations are performed based on a mixed-traffic assignment model, and the successive mean average (MSA) method is used to update the segment traffic based on the origin. During the inner-layer network traffic loading process, path selection-sensitive parameters are introduced. This is used to adjust the sensitivity of vehicle type m to road segment cost differences during road segment loading; The larger the value, the more traffic tends to be allocated to road segments with lower costs. This is determined by the vehicle type m starting from the origin r and then moving along the road segment. Traffic And calculate the travel time for the corresponding road segment. Compared with the expected average travel time It is used for updating the outer vehicle type selection model.

[0055] During the inner layer iterative calculation, when the change in inner layer flow is less than the inner layer convergence threshold... Alternatively, stop the inner layer iteration when the preset maximum number of inner layer iterations is reached.

[0056] Step S53: Update outer layer requirements.

[0057] Calculate the generalized cost of the four types of vehicles based on the current expected average travel time. The selection probabilities of the four types of vehicles at each starting point are calculated using the aforementioned vehicle type selection model based on the starting point. Furthermore, a relaxation coefficient is employed. Probability of vehicle type selection Update the data and determine the travel demand by vehicle type for each origin-destination pair. This is to improve the stability of the iterative process.

[0058] Step S54: Convergence check.

[0059] Calculate the maximum change in travel demand for vehicle type between two consecutive outer iterations. .like If the result is positive, stop the iteration and output the current solution; otherwise, let n = n + 1 and return to step S52 to continue the next round of iteration.

[0060] To further verify the effectiveness of the technical solution of this invention, the Nguyen-Dupuis test road network was selected for illustration, and its topology is as follows: Figure 2 As shown. This test road network is a commonly used benchmark road network in the field of traffic assignment, containing 13 nodes and 19 road segments. The set of origins R={1,4} and the set of destinations S={2,3} form 4 origin-destination pairs, namely (1,2), (1,3), (4,2) and (4,3), respectively. The travel demand of each origin-destination pair is shown. The corresponding speeds are 5,000 vehicles / hour, 4,500 vehicles / hour, 5,600 vehicles / hour, and 5,200 vehicles / hour.

[0061] Table 1 shows the information for each road segment in the test road network, including the road segment number and corresponding road segment. Free-flow travel time, segment length, and number of lanes. The parameters listed in Table 1 are used for calculating segment travel time and mixed traffic assignment. The speed is set to 60.0 km / h. The travel time for the road segment is calculated using the BPR function, with parameters α=0.15 and β=4.

[0062] Table 1. Road segment information for the Nguyen-Dupuis test road network.

[0063] In this test road network, four vehicle types—HV, AV, CV, and CAV—are considered, and a segmental travel time calculation method based on passenger car equivalent PCU is adopted. The vehicle length L is taken as 4.0m, and the reaction times of HV, AV, CV, and CAV are... The time intervals are 1.5s, 0.8s, 1.0s, and 0.6s, respectively. The vehicle type selection model uses a multinomial Logit model with the following scaling parameters: Take 0.15. This embodiment will assign each start-endpoint pair... average travel distance Given as input parameters, where the start-end pair (1,2), (1,3), (4,2), (4,3) The distances are 35.1250, 37.6667, 35.6000, and 36.8333 km, respectively. The total cost of ownership per unit distance for HV, AV, CV, and CAV is given in RMB (yuan). The time value parameters are set at 0.9356, 1.1831, 0.9823, and 1.2161 yuan / km respectively. The rates were set to 0.298, 0.176, 0.298 and 0.176 yuan / min, respectively.

[0064] During the solution process, the initial vehicle type selection probabilities of HV, AV, CV, and CAV at each starting point are calculated. All are set to 0.25; Path selection sensitivity parameters for HV, AV, CV, and CAV during inner network loading. The values ​​are 0.4, 0.4, 50.0, and 50.0 respectively; the convergence threshold in the outer iteration is... Set to 0.001, maximum number of iterations to 30, relaxation coefficient Set to 0.3; The convergence threshold for flow rate changes in the inner iteration is... Set it to 1.0, and the maximum number of iterations is 10000.

[0065] Based on the above parameters, the iterative solution method for vehicle type selection and mixed traffic assignment described in step S5 is used to calculate the travel demand for each origin-destination pair according to vehicle type. The results are shown in Table 2. Table 2 lists the travel demand for each origin-destination pair and the travel demand corresponding to the four vehicle types: HV, AV, CV, and CAV.

[0066] Table 2. Travel demand by vehicle type for each origin-destination pair

[0067] Table 2 shows that for different origin-destination pairs, the proposed method can determine the travel demand for four vehicle types: HV, AV, CV, and CAV. Furthermore, the sum of the travel demands for these four vehicle types is consistent with the corresponding origin-destination travel demand. These results demonstrate that the proposed method can intrinsically determine the allocation of origin-destination travel demand among different vehicle types while maintaining the conservation of total origin-destination demand. Compared to existing methods that use vehicle penetration rate as an externally given parameter, this method comprehensively considers the impact of vehicle purchase and usage costs, time value, and network equilibrium travel time on vehicle type selection, thereby obtaining the travel demand for different vehicle types.

[0068] Through iterative solutions of vehicle type selection and mixed traffic assignment, the travel demand for each origin-destination pair by vehicle type, as well as the traffic flow of different vehicle types, total traffic flow, and travel time for each road segment, are obtained. The results are shown in Table 3. Table 3 lists the HV, AV, CV, and CAV traffic flows, total traffic flow, and corresponding travel times for each road segment.

[0069] To visually demonstrate the spatial distribution characteristics of traffic flow in the test road network, Figure 3 The spatial distribution of total traffic flow in each road segment of the test road network is given. The number above the road segment indicates the road segment number, and the number below indicates the total traffic flow (vehicles / hour) of the corresponding road segment. The shade of gray indicates the size of the total traffic flow, and the darker the gray, the greater the total traffic flow.

[0070] Table 3. Traffic flow, total traffic flow, and travel time for four types of vehicles on each road segment.

[0071] From Table 3 and Figure 3 As can be seen, through iterative solutions of vehicle type selection and mixed traffic assignment, the proposed method can obtain the traffic flow of HV, AV, CV, and CAV, the total traffic flow of the road segment, and the travel time of the road segment under mixed traffic conditions for four types of vehicles. Furthermore, the spatial distribution characteristics of heterogeneous traffic flow in the test road network are visually reflected through a spatial distribution map of road segment traffic flow. These results demonstrate that the proposed method can simultaneously determine the travel demand of different vehicle types and the road network traffic flow distribution results within the same framework, providing a quantitative basis for traffic operation status analysis and traffic management strategy evaluation under mixed traffic conditions for four types of vehicles.

[0072] To verify whether the results obtained by the method satisfy the equilibrium conditions for different vehicle types, taking starting point 1 as an example, Table 4 lists the generalized segment costs and minimum generalized cost differences for HV and AV on each road segment. ) and traffic flow of relevant road sections from starting point 1 Table 5 lists the generalized segment cost, minimum generalized cost difference, and traffic flow of the relevant segment at starting point 1 for CV and CAV on each segment.

[0073] Table 4. Generalized segment costs, minimum generalized cost difference, and traffic flow of relevant segments at starting point 1 for HV and AV on various road segments.

[0074] Table 5 Generalized segment cost, minimum generalized cost difference, and traffic flow of CV and CAV on various road segments, and related road segments from starting point 1.

[0075] Note: In Tables 4 and 5, "-" indicates that the road segment is not a reachable road segment for the origin-end point trip corresponding to origin 1, so the corresponding generalized road segment cost and minimum generalized cost difference are not listed.

[0076] As shown in Tables 4 and 5, taking starting point 1 as an example, the traffic flow of the relevant road segments at starting point 1 is... For road segments that are accessible but have a starting point of 1, the generalized road segment cost is consistent with the minimum generalized cost difference; for road segments with a starting point of 1, the relevant road segment flow... For a given road segment, the generalized segment cost is not less than the minimum generalized cost difference. The above results demonstrate that, taking starting point 1 as an example, the obtained traffic assignment results satisfy the stochastic user equilibrium conditions for HV and AV, the user equilibrium conditions for CV, and the system optimal conditions for CAV.

[0077] Example 2 The present invention also provides a computer-readable storage medium. In one embodiment of the present invention, the computer-readable storage medium can be configured to store a program that performs the mixed traffic allocation method considering endogenous selection of vehicle type in the above-described method embodiments. This program can be loaded and run by a processor to implement the above-described mixed traffic allocation method considering endogenous selection of vehicle type. For ease of explanation, only the parts related to the embodiments of the present invention are shown; for specific technical details not disclosed, please refer to the method section of the embodiments of the present invention. The computer-readable storage medium can be a storage device comprising various electronic devices. Optionally, in the embodiments of the present invention, the computer-readable storage medium is a non-transitory computer-readable storage medium.

[0078] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the original technical features, and the technical solutions resulting from these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A mixed-traffic assignment method considering endogenous vehicle type selection, characterized in that, Includes the following steps: Construct a mixed traffic network, which includes at least manually driven vehicles (HV), autonomous vehicles (AV), connected vehicles (CV), and connected autonomous vehicles (CAV); For each starting point, the generalized cost of each vehicle type at that starting point is calculated based on the total cost of ownership per unit distance, time value parameters, and expected average travel time. Then, the selection probability of each vehicle type is determined based on the generalized cost. Based on the travel demand of each origin-destination pair and the selection probability of each vehicle type, calculate the travel demand of each origin-destination pair for each vehicle type. Using the segment flow based on the origin and the travel demand divided by vehicle type as joint decision variables, and adopting a segment-node-based modeling approach, the vehicle type selection conditions and the traffic allocation equilibrium conditions for each vehicle type are uniformly expressed as a variational inequality problem. The variational inequality problem is solved iteratively to obtain the travel demand of each vehicle type and the equilibrium flow distribution of each vehicle type on each road segment in the mixed traffic network.

2. The method according to claim 1, characterized in that, The process of building a mixed-traffic network includes: Representing the mixed traffic network as a directed graph Where N is the set of nodes and A is the set of road segments; Define a starting set R and an ending set S, and , and define the vehicle type set. ; For each start-end pair (r, s), define For the travel demand of this origin-destination pair, Let m be the travel demand of vehicle type m on the origin-destination pair (r,s), and satisfy: Where r and s represent the starting point and the ending point, respectively, and m represents the vehicle type; The sum of the starting point and ending point requirements from the starting point r to all ending points s. That is, the travel demand of the starting point r: Vehicle type m on the road segment Traffic This is the sum of the traffic flow for this vehicle type at each starting point on the road segment: in, This indicates the vehicle type m starting from the origin r on the road segment. Traffic on the internet; Section Total flow The sum of traffic flow for each vehicle type on this road segment: 。 3. The method according to claim 2, characterized in that, For each origin, the generalized cost of each vehicle type at that origin is calculated based on the total cost of ownership per unit distance, time value parameters, and expected average travel time, including: For each starting point r and each vehicle type m, the generalized cost is calculated as follows: Total cost of ownership per unit distance based on vehicle type m Average travel distance from the starting point r Determine the average monetary cost of a single trip for vehicle type m at origin r. : Based on the time value parameter of vehicle type m Under network equilibrium, the expected average travel time of vehicle type m starting from origin r. Determine the average trip time cost for vehicle type m at origin r. ; Adding the average monetary cost per trip to the average time cost per trip yields the generalized cost of vehicle type m at the origin r. : 。 4. The method according to claim 3, characterized in that, The method also includes obtaining the average travel distance from the starting point r. Specifically, it includes: For a starting point r, the average travel distance is calculated based on the weighted average of the starting-to-destination demand corresponding to r: in, This indicates the start-end pair. The average travel distance.

5. The method according to claim 3, characterized in that, The method also includes obtaining the expected average travel time. Specifically, it includes: The calculation is based on the weighted average traffic flow and travel time of this vehicle type on each road segment: in, Let r represent the travel demand of vehicle type m at the starting point r. Indicates road segment Travel time.

6. The method according to claim 3, characterized in that, Determining the selection probability of each vehicle type based on the generalized cost includes: Using generalized cost as the basis for vehicle type selection, a multinomial Logit model is employed to calculate the probability that a traveler at origin r will choose vehicle type m: in, This is a scaling parameter used to represent the sensitivity of travelers to generalized cost differences between different vehicle types. Let k be the vehicle type set, and k be the vehicle type index. Let r represent the generalized cost of vehicle type k at the starting point r.

7. The method according to claim 6, characterized in that, Based on the travel demand for each origin-destination pair and the selection probability of each vehicle type, the travel demand for each origin-destination pair and each vehicle type is calculated, including: For the origin-end pair Travel demand for vehicle type m Represented as: The travel demand for vehicle type m at the starting point r is obtained by summing the travel demands for vehicle type m from that starting point to all destinations, that is: 。 8. The method according to claim 7, characterized in that, Using segment traffic flow based on origin and travel demand categorized by vehicle type as joint decision variables, and employing a segment-node-based modeling approach, the vehicle type selection conditions and traffic allocation equilibrium conditions for each vehicle type are uniformly expressed as variational inequality problems, including: Establish node flow conservation constraints and non-negativity constraints: in, This represents the net inflow of vehicle type m originating from r at node j. This indicates the vehicle type m starting from the origin r on the road segment. Traffic flow; establish traffic distribution equilibrium conditions for each vehicle type: For manually driven vehicles (HV) and autonomous vehicles (AV), following the Stochastic User Equilibrium (SUE) principle, we define the vehicle type m relative to the origin r on the road segment. The generalized road segment cost is: in, Indicates road segment Travel time, This indicates the vehicle type m starting from the origin r on the road segment. Traffic on the internet , This represents the flow of vehicle type m originating from origin r and flowing into node j. Let j be the index of the predecessor node of node j. The perception error scale parameter for human-driven vehicles (HV) and autonomous vehicles (AV) in path selection; like Then the generalized segment cost of that segment is equal to the minimum perceived generalized cost difference between adjacent nodes; if Then the generalized segment cost of this segment is not less than the minimum perceived generalized cost difference between adjacent nodes, specifically expressed as: in, Indicates the cost of a generalized road segment The value at the equilibrium solution, and Let represent the minimum perceived generalized cost of vehicle type m from starting point r to node i and node j, respectively, under equilibrium conditions. This indicates that in equilibrium, vehicle type m starting from point r is on road segment Traffic on the internet; For connected vehicles, following the user-balanced user (UE) principle, the CV is defined relative to the starting point r on the road segment. The generalized road segment cost is: like Then the actual generalized segment cost of this segment is equal to the minimum actual generalized cost difference between adjacent nodes; if If the travel cost of this route segment is not less than the minimum actual generalized cost difference between adjacent nodes, it can be expressed as follows: in, Indicates the cost of a generalized road segment The value at the equilibrium solution, and Let represent the minimum actual generalized cost of CV reaching node i and node j from the starting point r, respectively, in the equilibrium state. This represents the CV starting from the origin r in the equilibrium state on the road segment. Traffic on the internet; For connected autonomous vehicles (CAVs), following the system optimality (SO) principle, the CAV is defined relative to the starting point r on the road segment. The generalized road segment cost is: in, Indicates CAV on the road segment Traffic on the internet Indicates road segment travel time The partial derivative of the CAV flow rate on this road segment; like Then the generalized segment cost of that segment is equal to the minimum marginal generalized cost difference between adjacent nodes; if If the cost of the generalized road segment is not less than the minimum marginal generalized cost difference between adjacent nodes, it can be expressed as follows: in, Indicates the cost of a generalized road segment The value at the equilibrium solution, and These represent the CAV starting from the point of equilibrium. The minimum marginal generalized cost to reach nodes i and j. This indicates that the CAV starting from the origin r in the equilibrium state is located on the road segment. Traffic on the internet; Finally, the vehicle type selection conditions and the traffic allocation equilibrium conditions for each vehicle type are uniformly expressed as the following variational inequality problem, where K is the joint feasible region determined by the node flow conservation constraint, the non-negativity constraint, and the vehicle type travel demand constraint: Solve , making The first three terms correspond to the stochastic user equilibrium conditions for manually driven vehicles (HV) and autonomous vehicles (AV), the user equilibrium conditions for connected vehicles (CV), and the system optimal conditions for connected autonomous vehicles (CAV), respectively. The last term corresponds to the vehicle type selection condition based on the starting point. Indicating equilibrium state The value of represents the travel demand of vehicle type m at the starting point r under equilibrium conditions.

9. The method according to claim 5, characterized in that, The travel time for the aforementioned road segment is calculated using a method based on passenger car equivalent PCU, specifically including: Introducing passenger car equivalent conversion factor This is used to convert the flow rate of vehicle type m into the standard passenger car equivalent flow rate: in, For road section free flow velocity, Let m be the reaction time for vehicle type m. HV is the reaction time, and L is the vehicle length; Road sections calculated using the U.S. Highway Bureau's Road Resistance Function (BPR) Travel time on road sections with mixed traffic conditions: in, For road section Free-flow travel time, and These are parameters for the road resistance function of the U.S. Federal Highway Administration. This indicates the road section under conditions of purely manual driving. The baseline traffic capacity.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 9.

Images