Intersection signal lamp-free optimization control method and system based on intelligent network connection vehicle team

Abstract

The invention discloses an intersection no-signal-lamp optimization control method and system based on an intelligent network connection fleet, and the method comprises the steps: (1) building an intersection no-signal-lamp optimization control model, according to the model, the minimum sum of the total driving delay of a CAV queue passing through an intersection and the total error between performance indexes and final state indexes of the CAV queue in all entrance lane directions is used as a target function, CAV queue control is used as a constraint condition, and a formation mode of the CAV queue is obtained; (2) a CAV queue control model is established, the model takes the minimum sum of CAV operation cost and terminal cost as an objective function, and the acceleration and state of each CAV at each moment in the CAV queue are obtained; (3) establishing an optimization control double-layer model, wherein the optimization control double-layer model comprises an intersection signal-lamp-free optimization control model located at the upper layer and a CAV queue control model located at the lower layer; and (4) solving the optimization control double-layer model. According to the invention, intersection signal lamp-free optimization control based on the intelligent network connection fleet can be realized, and the method is simple and easy to implement.

Description

Technical Field

[0001] The present invention relates to intelligent transportation technology, and in particular to an intersection without signal lights optimization control method and system based on an intelligent network-connected vehicle fleet. Background Art

[0002] As an emerging product of the cross-integration of information, automobile, transportation and other industries, the Connected Autonomous Vehicle (CAV) has emerged in recent years. With the help of network communication and autonomous driving technology, multiple CAVs can form a team to shorten the distance between vehicles, thereby improving traffic efficiency and reducing energy consumption and emissions.

[0003] Most of the existing studies in the literature focus on determining the passing sequence, allocation time, and trajectory planning of CAV platoons to intersection conflict zones and freeway merge zones based on platoon-based unsignalized and merge-coordinated control. To date, no one has paid attention to the optimization of tactical decisions such as platoon size, which may greatly affect the overall traffic efficiency of the intersection. Intuitively, a large-scale CAV platoon can improve the travel efficiency of the current entrance direction of the intersection. However, adopting a large platoon size will have a negative impact on the travel efficiency of other conflicting directions. In addition, based on a two-stage / two-layer model, the passing sequence and allocation time are first determined, and then the trajectories of CAVs when the vehicles are platooned are determined, which is also the focus of many research works. Due to the simplification of modeling, the impact of the trajectory of CAV platoons in formation on the passing sequence and allocation time is rarely considered. Although a few studies have jointly optimized the passing sequence and allocation time as well as the trajectory of CAVs forming a platoon for intersection management problems, the trajectories obtained by simple motion planning methods may not be applicable to the actual formation of CAV platoons. Therefore, unsignalized intersection control considering intelligent connected vehicle platooning has become one of the main problems that need to be solved in intersection management. Summary of the invention

[0004] In view of the problems existing in the prior art, the purpose of the present invention is to provide an intersection without signal lights optimization control method and system based on an intelligent connected vehicle fleet.

[0005] In order to achieve the above-mentioned purpose of the invention, the present invention provides an intersection without signal light optimization control method based on an intelligent networked fleet, comprising the following steps:

[0006] (1) Establishing an optimization control model for intersections without signal lights, wherein the optimization control model for intersections without signal lights takes the total travel delay of a CAV queue passing through an intersection and the minimum sum of the total errors between the performance index and the final state index of the CAV queue in all entrance road directions as the objective function, takes the CAV queue control as the constraint condition, and obtains the formation mode of the CAV queue;

[0007] (2) establishing a CAV queue control model, wherein the CAV queue control model, when the queue information of the CAV queue is known, takes the minimum sum of the CAV operation cost and the terminal cost as the objective function, and obtains the acceleration and state of each CAV in the CAV queue at each moment;

[0008] (3) establishing an optimization control double-layer model, wherein the optimization control double-layer model includes an intersection without signal lights optimization control model located at an upper layer and a CAV queue control model located at a lower layer;

[0009] (4) Solving the optimization control double-layer model, obtaining the optimal formation mode of the CAV queue when passing through an intersection without a traffic light, and the optimal acceleration and optimal state of each CAV in the CAV queue at each moment.

[0010] The present invention provides an intersection without signal lights optimization control system based on an intelligent networked fleet, comprising:

[0011] The first model building module is used to establish an intersection without signal lights optimization control model, wherein the intersection without signal lights optimization control model takes the total travel delay of the CAV queue through the intersection and the minimum sum of the total errors between the performance index and the final state index of the CAV queue in all entrance road directions as the objective function, takes the CAV queue control as the constraint condition, and obtains the formation mode of the CAV queue;

[0012] The second model building module is used to build a CAV queue control model, wherein the CAV queue control model takes the minimum sum of CAV operation cost and terminal cost as the objective function when the queue information of the CAV queue is known, and obtains the acceleration and state of each CAV in the CAV queue at each moment;

[0013] A two-layer model building module, used to build an optimization control two-layer model, wherein the optimization control two-layer model includes an intersection without signal lights optimization control model located at an upper layer and a CAV queue control model located at a lower layer;

[0014] The model solving module is used to solve the optimization control double-layer model to obtain the optimal formation mode of the CAV queue when passing through the intersection without a traffic light and the optimal acceleration and optimal state of each CAV in the CAV queue at each moment.

[0015] Compared with the prior art, the present invention has the following beneficial effects: the present invention proposes a two-layer control model for signal-free intersections considering intelligent connected vehicle formation, combines optimization management and optimal control schemes, considers the formation process of CAV fleets in the model, wherein the upper model determines the formation mode of the CAV fleet, and the lower model determines the optimal trajectory of the CAV fleet formation; develops a hybrid artificial bee colony algorithm that integrates the artificial bee colony method and the dynamic programming algorithm for optimal control, and gives full play to the advantages of the two algorithms to solve the proposed model; studies the maximum fleet size change of the fleet-based control strategy and its impact on the traffic operation performance under balanced and unbalanced motion states between the entrance lane directions. In addition, compared with the baseline strategy, the fleet-based control strategy has better traffic throughput performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flow chart of a method for optimizing control of no-signal lights at intersections of an intelligent network-connected vehicle fleet provided by an embodiment of the present invention;

[0017] Figure 2 It is a schematic diagram of the CAV queue status at the intersection;

[0018] Figure 3 It is a schematic diagram of the solution method of the optimization control two-layer model. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0020] Embodiment 1

[0021] Figure 1 The invention provides an optimization control method for intersections without signal lights based on an intelligent networked fleet. Before describing the invention, the problem to be solved by the invention is first described. Figure 2 An intersection in the control area shown in (a) (see the circle area within the orange dashed line). Let Z represent the set of all entrance road directions at the intersection, and I z is the set of CAVs in the z direction of the inlet channel, z∈Z.

[0022] The leading vehicle in the region (such as Figure 1 The time instant when CAV1 in (a) reaches the decision position of its direction is t 0 The present invention will study the time instant t 0 is in a pre-specified decision position (see Figure 1All scattered CAVs between the yellow line in (a) and the boundary of the control area near the intersection; it is assumed that all CAVs in the control area can exchange real-time motion information, i.e., position, speed, and acceleration, with each other, and share this information with the unsignalized intersection controller through V2V and V2I communications. CAVs in each direction can form CAV convoys of different convoy sizes to cross the intersection. The present invention needs to determine the formation mode of the CAV queue, the maximum convoy size, the passing order and the allocation time, as well as the optimal acceleration and optimal state (position and speed) of each CAV at each moment.

[0023] In order to solve the above problems, the present invention specifically comprises the following steps:

[0024] (1) Establish an optimal control model for intersections without traffic lights.

[0025] The intersection non-signal light optimization control model takes the total travel delay of the CAV queue through the intersection and the minimum sum of the total errors between the performance index and the final state index of the CAV queue in all entrance road directions as the objective function, and takes the CAV queue control as the constraint condition to obtain the formation mode of the CAV queue. Specifically, the intersection non-signal light optimization control model is:

[0026]

[0027]

[0028] s={s kk′ ∈{0,1}|z∈Z,z′∈Z\{z},k=1,2,...,|I z |,k′=1,2,...,|I z′ |}

[0029]

[0030] Where T represents the total travel delay of the CAV queue through the intersection, E represents the total error between the performance index and the final state index of the CAV queue in all entrance directions, M represents a positive constant greater than the preset threshold; Z represents the set of all entrance directions of the intersection entrance, z represents an entrance direction in Z, and I z represents the set of CAVs in the z direction, |I z | indicates I z Number of CAVs in the medium; They represent the allocation time of the k+1th and kth queues arriving at the conflict point in the z direction, represents the allocation time of the k′th queue in the z′ direction to reach the conflict point; is the CAV non-empty indicator variable. If the kth queue in the z direction is not empty, then otherwise are all vehicle position indicator variables. If CAV i is in the kth queue in the z direction, then otherwise If CAV i - In the z direction, the k′th queue otherwise i - Indicates the number of the CAV before CAV i. If CAV i is in the k′th queue in the z′ direction, then otherwise Indicates the CAV i of the kth queue in the z direction from The additional time required to add the distance from the location at the time to the conflict point, It represents the optimal allocation time for the k+1th and kth queues to reach the conflict point in the z direction. represents the shortest driving time required for CAV i to reach the conflict point along the z direction. It should be noted that according to the basic kinematic principle, when each CAV drives to the conflict point alone, it can first accelerate to a pre-specified terminal speed and then cruise to calculate the shortest driving time. The terminal state index is specifically Indicates the location of the conflict point, Indicates the final headway distance, Indicates the final state speed. The performance indicators of the CAV queue in the z direction include represents the optimal position of the k-th queue CAV in the z direction, represents the optimal heading of the k-th CAV in the z direction, represents the optimal speed of the kth CAV in the z direction; c z represents the maximum number of vehicles allowed in the CAV queue in the z direction, L is the safety buffer distance, s kk′ Indicates that through the order indicator variable, if the k-th queue CAV has priority over the k′-th queue CAV, then s kk′ =1, otherwise s kk′ = 0; the formation mode of the CAV queue includes {w, y, c, s, t}, where w, y, c, s, t represent the vehicle position information set, empty position information set, maximum vehicle number set, passing order information set, and allocation time set of the CAV queue respectively. Figure 1 The red dots in (a) are the conflict points that are pre-specified terminal locations.

[0031] Among them, since the present invention aims to manage and control CAVs so that they form a convoy while crossing the conflict point and reaching a predetermined forward speed and speed, the large -M component in the objective function will penalize infeasible CAV trajectories. The first constraint ensures that each CAV joins a CAV convoy. The second and third constraints ensure the continuous grouping of the CAV convoy. The fourth constraint ensures that there is no overtaking of CAVs during platoon formation. The fifth constraint limits the total number of CAVs in a convoy to an upper limit on the size of the convoy. The sixth and seventh constraints ensure that the distribution time difference between two CAV queues from the same direction or from different directions to reach the conflict point is not less than the sum of the time distances of each CAV in the previous CAV queue (i.e., the sum of the predetermined end point and the safety distance divided by the predetermined terminal speed). The eighth to twelfth constraints respectively define the variables s kk′ ,c z and domain.

[0032] (2) Establish a CAV queue control model.

[0033] Among them, the CAV queue control model, when the queue information of the CAV queue is known, takes the minimum sum of the CAV operating cost and the terminal cost as the objective function, and obtains the acceleration and state of each CAV in the CAV queue at each moment. Since the operating cost is related to the cost of the transition process before the CAV reaches a stable following state, it is necessary to consider the system state and acceleration of each time step within the specified time span. In order to achieve the formation and keep the following CAV at the predetermined terminal vehicle distance and speed, the operating cost should include three cost items. The first item is used to adjust the actual forward direction so that it reaches the predetermined terminal forward direction; the second item is used to maintain the speed uniformity between the two CAVs; and the last item is used to ensure comfortable and energy-saving driving within the time span. Therefore, the established CAV queue control model is specifically as follows:

[0034]

[0035] In the formula, J represents the total cost, represents the operating cost function, ζ() represents the terminal cost function, t 0 Indicates the initial time, represents the optimal allocation time for the kth queue CAV to reach the conflict point in the z direction, They represent the state vector and acceleration vector of the kth queue CAV along the z direction at time t, They represent the position column vector and velocity column vector of the k-th queue CAV along the z direction at time t, respectively. Indicates that the k-th queue CAV is The state column vector along the z direction at each moment; p i (t) respectively represent CAVi - 、CAVi + , the position of CAV i, i - 、i + Respectively The vehicle numbers before and after vehicle i, Indicates the final headway distance, v i (t) respectively represent CAVi + , the speed of CAV i, u1(t), u i (t) respectively represent The acceleration of the first CAV, CAV i, represents the optimal set of CAVs of the kth queue in the z direction, and n represents The number of CAVs in the middle, β1, β2, β3 are weight parameters; represents the final position, represents the final velocity, Respectively expressed as The position and speed of the first CAV in the k-th queue at time z, α1 and α2 are weighting coefficients; express Derivative, and A is a 2n*2n coefficient matrix, B is a 2n*1 coefficient matrix, is an n×n matrix of all 1s, is an n×1 matrix of all 1s; v min 、v max Respectively represent the minimum and maximum speed, u min 、u max Respectively represent the minimum and maximum acceleration values, represents the optimal position of CAV j in the k-1th row in the z direction, represents the optimal set of the k-1th row of CAVs in the z direction, p i (t0) represents the initial position of CAV i of the kth queue in the z direction, D c Indicates the length of the control area. Represents a state set, Represents a collection of accelerations.

[0036] The first constraint represents the system dynamics of the CAV fleet. The second and third constraints ensure that the acceleration [u min ,u max ] and speed [v min ,v max] are within the reasonable range of and at any time within the time range. The fourth constraint restricts the actual distance between vehicle j in the (k-1)th row and vehicle i in the kth queue in the same direction z to maintain a safe buffer distance. The fifth constraint ensures that the actual distance between two consecutive CAVs in the same row is not less than the safe buffer distance. The sixth constraint ensures that information sharing can be achieved between CAVs in the control area.

[0037] (3) Establishing an optimization control double-layer model, wherein the optimization control double-layer model includes an intersection without signal lights optimization control model at the upper layer and a CAV queue control model at the lower layer.

[0038] (4) Solving the optimization control double-layer model, obtaining the optimal formation mode of the CAV queue when passing through an intersection without a traffic light, and the optimal acceleration and optimal state of each CAV in the CAV queue at each moment.

[0039] In order to solve the optimization control double-layer model, the present invention integrates the artificial bee colony (ABC) algorithm and the dynamic programming algorithm to propose a hybrid ABC algorithm, such as Figure 3 As shown, this step specifically includes the following steps:

[0040] (4.1) Randomly generate an initial solution; For the ABC algorithm in the proposed method with a maximum number of iterations of Nm, in the startup phase, the employed bees will generate an initial solution group. Each bee is responsible for a randomly generated solution, and the procedures of the employed bee stage and the observation bee stage are similar.

[0041] (4.2) Through the artificial bee colony search method, according to the initial solution or the performance indicators of the CAV queue in the previous iteration, the optimal solution of the intersection non-signal light optimization control model is searched, and the optimal formation mode of the CAV queue in this iteration is obtained; wherein, each employed bee selects its relevant solution, and each observer bee selects a solution according to the fitness value, and the fitness value of each solution can be calculated in each iteration. Then, after evaluating the fitness value of the new solution, a greedy selection strategy is performed on the two stages. If the new solution is better than the old solution, the new solution can replace the old solution, and its limit value is set to 0. After all employed bees and observer bees complete the neighborhood research respectively, the employed bee and observer bee stages end. In the scout bee stage, all solutions are scanned, and solutions that cannot be improved under the limit of continuous iterations will be abandoned and replaced by randomly generated new solutions;

[0042] (4.3) Based on the optimal formation mode of the CAV queue solved by the upper model, the dynamic programming algorithm is used to solve the CAV queue control model, so as to obtain the optimal acceleration and optimal state of each CAV in the CAV queue at each moment in this iteration;

[0043] (4.4) The performance index of the CAV queue is calculated based on the optimal acceleration and optimal state at this iteration;

[0044] (4.5) Return to step (4.2) to perform the next iteration until the iteration reaches a preset number of times or meets a preset condition, and then the iteration ends. The optimal formation mode of the CAV queue at the end of the iteration, as well as the optimal acceleration and optimal state of the CAV queue are output as the optimal solution.

[0045] Wherein, step (4.3) specifically includes:

[0046] (4.3.1) The time interval The objective function of the CAV queue control model is discretized into R control steps, and the discretized objective function is obtained according to the control step.

[0047]

[0048] in, represents the control step length, r is the discrete index, They represent the state vector and acceleration vector of the k-th queue CAV along the z direction at control step r, and satisfy A r and B r is the coefficient matrix for the control step r, and A r =I n×n +A×Δt,B r =B×Δt,I n×n is the n×n identity matrix;

[0049] (4.3.2) Calculate the weight matrix Q of r∈{0,1,…,R} according to the following formula r , D r and O r :

[0050]

[0051] D r =0

[0052]

[0053] in,

[0054]

[0055]

[0056] are all intermediate variables. is an n×n matrix, is an n×1 matrix;

[0057] (4.3.3) Settings in, They represent the control step r=R when Q r , D r The optimal solution of

[0058] (4.3.4) According to the following formula, we can reversely deduce the intermediate variable G of r∈{R-1,…,1,0}. r , H r , S r 、T r :

[0059]

[0060] S r =A r +B r G r

[0061] T=BH

[0062] T r =B r H r

[0063]

[0064] Among them, P r , G r , H r , S r 、T r They are all intermediate variables;

[0065] (4.3.5) According to G r , H r , S r 、T r The state vector and acceleration vector at each control step r are calculated according to the following formula;

[0066] x(r)=S r x(r-1)+T r , r∈{1,…,R}

[0067] u(r)=min[max(G r x(r)+H r ,u min ),u max ],r∈{1,…,R}

[0068] Where x(r) and x(r-1) represent the state vectors at control steps r and r-1, respectively, and x(0) = [p(t0)T ,v(t0) T ] T , p(t0), v(t0) represent the initial time position vector and velocity vector, respectively, u(r) represents the acceleration vector at control step r, u min 、u max are the minimum and maximum acceleration values ​​respectively;

[0069] (4.3.6) Substitute the state column vector and acceleration column vector at each control step r into the discretized objective function to obtain the optimal state column vector and acceleration column vector, that is, the optimal acceleration and optimal state of each CAV at each moment.

[0070] Embodiment 2

[0071] The embodiment of the present invention provides an intersection without signal lights optimization control system based on an intelligent networked fleet, including:

[0072] The first model building module is used to establish an intersection without signal lights optimization control model, wherein the intersection without signal lights optimization control model takes the total travel delay of the CAV queue through the intersection and the minimum sum of the total errors between the performance index and the final state index of the CAV queue in all entrance road directions as the objective function, takes the CAV queue control as the constraint condition, and obtains the formation mode of the CAV queue;

[0073] The second model building module is used to build a CAV queue control model, wherein the CAV queue control model takes the minimum sum of CAV operation cost and terminal cost as the objective function when the queue information of the CAV queue is known, and obtains the acceleration and state of each CAV in the CAV queue at each moment;

[0074] A two-layer model building module, used to build an optimization control two-layer model, wherein the optimization control two-layer model includes an intersection without signal lights optimization control model located at an upper layer and a CAV queue control model located at a lower layer;

[0075] The model solving module is used to solve the optimization control double-layer model to obtain the optimal formation mode of the CAV queue when passing through the intersection without a traffic light and the optimal acceleration and optimal state of each CAV in the CAV queue at each moment.

[0076] The optimization control model for intersections without signal lights is specifically as follows:

[0077]

[0078] s={s kk′ ∈{0,1}|z∈Z,z′∈Z\{z},k=1,2,...,|I z|,k′=1,2,...,|I z′ |}

[0079]

[0080] Where T represents the total travel delay of the CAV queue through the intersection, E represents the total error between the performance index and the final state index of the CAV queue in all entrance directions, M represents a positive constant greater than the preset threshold; Z represents the set of all entrance directions of the intersection entrance, z represents an entrance direction in Z, and I z represents the set of CAVs in the z direction, |I z | indicates I z Number of CAVs in the medium; They represent the allocation time of the k+1th and kth queues arriving at the conflict point in the z direction, represents the allocation time of the k′th queue in the z′ direction to reach the conflict point; is the CAV non-empty indicator variable. If the kth queue in the z direction is not empty, then otherwise are all vehicle position indicator variables. If CAV i is in the kth queue in the z direction, then otherwise If CAV i - In the z direction, the k′th queue otherwise i - Indicates the number of the CAV before CAV i. If CAV i is in the k′th queue in the z′ direction, then otherwise Indicates the CAV i of the kth queue in the z direction from The additional time required to add the distance from the location at the time to the conflict point, It represents the optimal allocation time for the k+1th and kth queues to reach the conflict point in the z direction. represents the shortest driving time required for CAV i to reach the conflict point along the z direction; the final state index is specifically: Indicates the location of the conflict point, Indicates the final headway distance, Indicates the final state speed. The performance indicators of the CAV queue in the z direction include represents the optimal position of the k-th queue CAV in the z direction, represents the optimal heading of the k-th CAV in the z direction, represents the optimal speed of the kth CAV in the z direction; c zrepresents the maximum number of vehicles allowed in the CAV queue in the z direction, L is the safety buffer distance, s kk′ Indicates that through the order indicator variable, if the k-th queue CAV has priority over the k′-th queue CAV, then s kk′ =1, otherwise s kk′ =0; the formation mode of the CAV queue includes {w, y, c, s, t}, where w, y, c, s, t respectively represent the vehicle position information set, empty position information set, maximum vehicle number set, passing order information set, and allocation time set of the CAV queue.

[0081] The CAV queue control model is specifically:

[0082]

[0083]

[0084] In the formula, J represents the total cost, represents the operating cost function, ζ() represents the terminal cost function, t 0 Indicates the initial time, represents the optimal allocation time for the kth queue CAV to reach the conflict point in the z direction, They represent the state vector and acceleration vector of the kth queue CAV along the z direction at time t, They represent the position column vector and velocity column vector of the k-th queue CAV along the z direction at time t, respectively. Indicates that the k-th queue CAV is The state column vector along the z direction at each moment; p i (t) respectively represent CAVi - 、CAVi + , the position of CAV i, i - 、i + Respectively The vehicle numbers before and after vehicle i, Indicates the final headway distance, v i (t) respectively represent CAVi + , the speed of CAV i, u1(t), u i (t) respectively represent The acceleration of the first CAV, CAV i, represents the optimal set of CAVs of the kth queue in the z direction, and n represents The number of CAVs in the middle, β1, β2, β3 are weight parameters; represents the final position, represents the final velocity, Respectively expressed as The position and speed of the first CAV in the k-th queue at time z, α1 and α2 are weighting coefficients; express Derivative, and A is a 2n*2n coefficient matrix, B is a 2n*1 coefficient matrix, is an n×n matrix of all 1s, is an n×1 matrix of all 1s; v min 、v max Respectively represent the minimum and maximum speed, u min 、u max Respectively represent the minimum and maximum acceleration values, represents the optimal position of CAV j in the k-1th row in the z direction, represents the optimal set of CAVs in the k-1th row in the z direction, p i (t0) represents the initial position of CAV i of the kth queue in the z direction, D c Indicates the length of the control area. Represents a state set, Represents a collection of accelerations.

[0085] Wherein, the model solving module specifically includes:

[0086] Initialization unit, used to randomly generate an initial solution;

[0087] The first model solving unit is used to search for the optimal solution of the intersection non-signal light optimization control model by an artificial bee colony search method according to the performance index of the CAV queue in the initial solution or the last iteration, and obtain the optimal formation mode of the CAV queue in this iteration;

[0088] The second model solving unit is used to solve the CAV queue control model using a dynamic programming algorithm according to the optimal formation mode of the CAV queue solved by the upper model, so as to obtain the optimal acceleration and optimal state of each CAV in the CAV queue at each moment in this iteration;

[0089] A performance indicator calculation unit, used to calculate the performance indicator of the CAV queue according to the optimal acceleration and optimal state in this iteration;

[0090] The iteration judgment unit is used to return to execute the first model solving unit for the next iteration until the iteration reaches a preset number of times or meets a preset condition, and then terminate the iteration, and output the optimal formation mode of the CAV queue at the end of the iteration as well as the optimal acceleration and optimal state of the CAV queue as the optimal solution.

[0091] Wherein, the second model solving unit specifically includes:

[0092] Discretization subunit, used to divide the time interval The objective function of the CAV queue control model is discretized into R control steps, and the discretized objective function is obtained according to the control step.

[0093]

[0094] in, represents the control step length, r is the discrete index, They represent the state vector and acceleration vector of the k-th queue CAV along the z direction at control step r, and satisfy A r and B r is the coefficient matrix for the control step r, and A r =I n×n +A×Δt,B r =B×Δt,I n×n is the n×n identity matrix;

[0095] The weight matrix calculation subunit is used to calculate the weight matrix Q of r∈{0,1,...,R} according to the following formula r , D r and O r :

[0096]

[0097] D r =0

[0098]

[0099] in,

[0100]

[0101]

[0102] are all intermediate variables. is an n×n matrix, is an n×1 matrix;

[0103] The optimal solution determination subunit is used to set in, They represent the control step r=R when Q r , D r The optimal solution of

[0104] The reverse derivation subunit is used to reversely deduce the intermediate variable G of r∈{R-1,…,1,0} in turn according to the following formula r , H r , S r 、T r :

[0105]

[0106] S r =A r +B r G r

[0107] T r =B r H r

[0108]

[0109] Among them, P r , G r , H r , S r 、T r They are all intermediate variables;

[0110] The state and acceleration column vector calculation subunit is used to calculate the acceleration column vector according to G r , H r , S r 、T r The optimal state column vector and the optimal acceleration column vector at each control step r are calculated according to the following formula;

[0111] x(r)=S r x(r-1)+T r ,r∈{1,...,R}

[0112] u(r)=min[max(G r x(r)+H r ,u min ),u max ],r∈{1,...,R}

[0113] Where x(r) and x(r-1) represent the state column vectors at control steps r and r-1, respectively, and x(0) = [p(t0) T ,v(t0) T ] T , p(t0), v(t0) represent the initial time position vector and velocity vector, respectively, u(r) represents the acceleration column vector at control step r, u min 、u max are the minimum and maximum acceleration values ​​respectively;

[0114] The optimal acceleration and optimal state determination subunit is used to obtain the optimal acceleration and optimal state of each CAV at each moment according to the optimal state column vector and the optimal acceleration column vector at each control step r.

[0115] The system provided in the embodiment of the present invention can be used to execute the method provided in the first embodiment of the present invention, and has the corresponding functions and beneficial effects of the execution method.

[0116] It is worth noting that in the embodiment of the above system, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0117] The embodiments described above are only illustrative, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, i.e., they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, or of course, it can be implemented only by hardware, as long as the function or effect can be achieved.

[0118] It should be understood that the above embodiments and descriptions only describe the principles, main features and advantages of the present invention. Without departing from the spirit and scope of the present invention, the present invention may be subject to various changes and improvements, and these changes and improvements all fall within the scope of protection of the present invention.

Claims

1. An optimization control method for intersections without signal lights based on an intelligent networked fleet, characterized in that: The steps include: (1) Establishing an optimization control model for intersections without signal lights, wherein the optimization control model for intersections without signal lights takes the total travel delay of a CAV queue passing through an intersection and the minimum sum of the total errors between the performance index and the final state index of the CAV queue in all entrance road directions as the objective function, takes the CAV queue control as the constraint condition, and obtains the formation mode of the CAV queue; (2) establishing a CAV queue control model, wherein the CAV queue control model, when the queue information of the CAV queue is known, takes the minimum sum of the CAV operation cost and the terminal cost as the objective function, and obtains the acceleration and state of each CAV in the CAV queue at each moment; (3) establishing an optimization control double-layer model, wherein the optimization control double-layer model includes an intersection without signal lights optimization control model located at an upper layer and a CAV queue control model located at a lower layer; (4) Solving the optimization control double-layer model, obtaining the optimal formation mode of the CAV queue when passing through an intersection without a traffic light, and the optimal acceleration and optimal state of each CAV in the CAV queue at each moment.

2. The method for optimizing control of intersections without signal lights based on an intelligent networked fleet according to claim 1 is characterized in that: The intersection without signal lights optimization control model is specifically: Where T represents the total travel delay of the CAV queue through the intersection, E represents the total error between the performance index and the final state index of the CAV queue in all entrance directions, M represents a positive constant greater than the preset threshold; Z represents the set of all entrance directions of the intersection entrance, z represents an entrance direction in Z, and I z represents the set of CAVs in the z direction, |I z | indicates I z Number of CAVs in the medium; They represent the allocation time of the k+1th and kth queues arriving at the conflict point in the z direction, represents the allocation time of the k′th queue in the z′ direction to reach the conflict point; is a CAV non-empty variable. If the kth queue in the z direction is not empty, then otherwise are all vehicle teaming variables. If CAV i In the z direction, the kth queue otherwise If CAV i - In the z direction, the k′th queue otherwise i - Indicates the number of the CAV before CAV i. If CAV i is in the k′th queue in the z′ direction, then otherwise Indicates the CAV i of the kth queue in the z direction from The additional time required to add the distance from the location at the time to the conflict point, represents the optimal allocation time for the kth queue to reach the conflict point in the z direction, represents the shortest driving time required for CAV i to reach the conflict point along the z direction; the final state indicator is specifically: Indicates the location of the conflict point, Indicates the final headway distance, Indicates the final state speed. The performance indicators of the CAV queue in the z direction include represents the optimal position of the k-th queue CAV in the z direction, represents the optimal headway distance of the k-th CAV queue in the z direction, represents the optimal speed of the k-th queue CAV in the z direction; c z represents the maximum number of vehicles allowed in the CAV queue in the z direction, L is the safety buffer distance, s kk′ Indicates that through the order variable, if the k-th queue CAV has priority over the k′-th queue CAV, then s kk′ =1, otherwise s kk′ =0; the formation mode of the CAV queue includes {w, y, c, s, t}, where w, y, c, s, t respectively represent the vehicle information set in the CAV queue, the non-empty information set of the queue, the maximum number of vehicles set, the passing order information set, and the allocation time set.

3. The method for optimizing control of intersections without signal lights based on an intelligent networked fleet according to claim 1, characterized in that: The CAV queue control model is specifically: In the formula, J represents the total cost, represents the operating cost function, ζ() represents the terminal cost function, t 0 Indicates the initial time, represents the optimal allocation time for the kth queue CAV to reach the conflict point in the z direction, They represent the state vector and acceleration vector of the k-th queue CAV along the z direction at time t, They represent the position vector and velocity vector of the k-th queue CAV along the z direction at time t, respectively. Indicates that the k-th queue CAV is The state vector along the z direction at the moment; p i (t) respectively represent CAVi - 、CAVi + , the position of CAV i, i - 、i + Respectively The vehicle numbers before and after vehicle i, Indicates the final headway distance, v i (t) respectively represent Medium CAV i + , the speed of CAV i, u1(t), u i (t) respectively represent The acceleration of the first CAV, CAV i, represents the optimal set of CAVs of the kth queue in the z direction, and n represents The number of CAVs in the middle, β1, β2, β3 are weight parameters; represents the final position, represents the final velocity, Respectively expressed as The position and speed of the first CAV in the k-th queue at time z, α1 and α2 are weighting coefficients; express Derivative, and A is a 2n*2n coefficient matrix, B is a 2n*1 coefficient matrix, is an n×n matrix of all 1s, is an n×1 matrix of all 1s; v min 、v max Respectively represent the minimum and maximum speed, u min 、u max Respectively represent the minimum and maximum acceleration values, represents the optimal position of CAV j in the k-1th row in the z direction, represents the optimal set of the k-1th row of CAVs in the z direction, p i (t0) represents the initial position of CAV i of the kth queue in the z direction, D c Indicates the length of the control area. Represents a state set, Represents a collection of accelerations.

4. The method for optimizing control of intersections without signal lights based on an intelligent networked fleet according to claim 3 is characterized in that: Step (4) specifically includes the following steps: (4.1) Randomly generate an initial solution; (4.2) Using the artificial bee colony search method, based on the acceleration and motion state trajectory of the CAV queue at the initial solution or the previous iteration and the final state performance index, the optimal solution of the intersection non-signal light optimization control model is searched to obtain the optimal formation mode of the CAV queue at this iteration; (4.3) Based on the optimal formation mode of the CAV queue solved by the upper model, the dynamic programming algorithm is used to solve the CAV queue control model, so as to obtain the optimal acceleration and optimal state of each CAV in the CAV queue at each moment in this iteration; (4.4) The performance index of the CAV queue is calculated based on the optimal acceleration and optimal state at this iteration; (4.5) Return to step (4.2) to perform the next iteration until the iteration reaches a preset number of times or meets a preset condition, and then the iteration ends. The optimal formation mode of the CAV queue at the end of the iteration, as well as the optimal acceleration and optimal state of the CAV queue are output as the optimal solution.

5. The method for optimizing control of intersections without signal lights based on an intelligent networked fleet according to claim 4 is characterized in that: Step (4.3) specifically includes: (4.3.1) The time interval The objective function of the CAV queue control model is discretized into R control steps, and the discretized objective function is obtained according to the control step. in, represents the control step length, r is the discrete index, They represent the state vector and acceleration vector of the k-th queue CAV along the z direction at control step r, and satisfy A r and B r is the coefficient matrix for the control step r, and A r =I n×n +A×Δt,B r =B×Δt,I n×n is the n×n identity matrix; (4.3.2) Calculate the weight matrix Q of r∈{0,1,…,R} according to the following formula r , D r and O r : D r ＝0 in, are all intermediate variables. is an n×n matrix, is an n×1 matrix; (4.3.3) Settings in, They represent the control step r=R when Q r , D r The optimal solution of (4.3.4) According to the following formula, we can reversely deduce the intermediate variable G of r∈{R-1,…,1,0}. r , H r , S r , T r : S r =A r +B r G r T=BH rrr Among them, P r , G r , H r , S r , T r They are all intermediate variables; (4.3.5) According to G r , H r , S r , T r The state vector and acceleration vector at each control step r are calculated according to the following formula; x(r)=S r x(r-1)+T r ,r∈{1,...,R} u(r)=min[max(G r x(r)+H r ,u min ),u max ],r∈{1,...,R} Where x(r) and x(r-1) represent the state column vectors at control steps r and r-1, respectively, and x(0) = [p(t0) T ,v(t0) T ] T , p(t0), v(t0) represent the initial time position vector and velocity vector respectively, u(r) represents the acceleration column vector at control step r, u min 、u max are the minimum and maximum acceleration values ​​respectively; (4.3.6) Substitute the state column vector and acceleration column vector at each control step r into the discretized objective function to obtain the optimal state column vector and acceleration column vector, that is, the optimal acceleration and optimal state of each CAV at each moment.

6. An optimized control system for intersections without signal lights based on an intelligent networked fleet, characterized in that: include: The first model building module is used to establish an intersection without signal lights optimization control model, wherein the intersection without signal lights optimization control model takes the total travel delay of the CAV queue through the intersection and the minimum sum of the total errors between the performance index and the final state index of the CAV queue in all entrance road directions as the objective function, takes the CAV queue control as the constraint condition, and obtains the formation mode of the CAV queue; The second model building module is used to build a CAV queue control model, wherein the CAV queue control model takes the minimum sum of CAV operation cost and terminal cost as the objective function when the queue information of the CAV queue is known, and obtains the acceleration and state of each CAV in the CAV queue at each moment; A two-layer model building module, used to build an optimization control two-layer model, wherein the optimization control two-layer model includes an intersection without signal lights optimization control model located at an upper layer and a CAV queue control model located at a lower layer; The model solving module is used to solve the optimization control double-layer model to obtain the optimal formation mode of the CAV queue when passing through the intersection without a traffic light and the optimal acceleration and optimal state of each CAV in the CAV queue at each moment.

7. The intersection non-signal light optimization control system based on the intelligent networked fleet according to claim 6 is characterized in that: The intersection without signal lights optimization control model is specifically: s={s kk′ ∈{0,1}|z∈Z,z′∈Z\{z},k=1,2,...,|I z |,k′=1,2,...,|I z′ |} Where T represents the total travel delay of the CAV queue through the intersection, E represents the total error between the performance index and the final state index of the CAV queue in all entrance directions, M represents a positive constant greater than the preset threshold; Z represents the set of all entrance directions of the intersection entrance, z represents an entrance direction in Z, and I z represents the set of CAVs in the z direction, |I z | indicates I z Number of CAVs in the medium; They represent the allocation time of the k+1th and kth queues arriving at the conflict point in the z direction, represents the allocation time of the k′th queue in the z′ direction to reach the conflict point; is the CAV non-empty indicator variable. If the kth queue in the z direction is not empty, then otherwise are all vehicle position indicator variables. If CAV i is in the kth queue in the z direction, then otherwise If CAV i - In the z direction, the k′th queue otherwise i - Indicates the number of the CAV before CAV i. If CAV i is in the k′th queue in the z′ direction, then otherwise Indicates the CAV i of the kth queue in the z direction from The additional time required to add the distance from the location at the time to the conflict point, It represents the optimal allocation time for the k+1th and kth queues to reach the conflict point in the z direction. represents the shortest driving time required for CAV i to reach the conflict point along the z direction; the final state indicator is specifically: Indicates the location of the conflict point, Indicates the final headway distance, Indicates the final state speed. The performance indicators of the CAV queue in the z direction include represents the optimal position of the k-th queue CAV in the z direction, represents the optimal heading of the k-th CAV in the z direction, represents the optimal speed of the kth CAV in the z direction; c z represents the maximum number of vehicles allowed in the CAV queue in the z direction, L is the safety buffer distance, s kk′ Indicates that through the order indicator variable, if the k-th queue CAV has priority over the k′-th queue CAV, then s kk′ =1, otherwise s kk′ =0; the formation mode of the CAV queue includes {w, y, c, s, t}, where w, y, c, s, t respectively represent the vehicle position information set, empty position information set, maximum vehicle number set, passing order information set, and allocation time set of the CAV queue.

8. The intersection non-signal light optimization control system based on the intelligent networked fleet according to claim 6 is characterized in that: The CAV queue control model is specifically: In the formula, J represents the total cost, represents the operating cost function, ζ() represents the terminal cost function, t 0 Indicates the initial time, represents the optimal allocation time for the kth queue CAV to reach the conflict point in the z direction, They represent the state vector and acceleration vector of the k-th queue CAV along the z direction at time t, They represent the position column vector and velocity column vector of the k-th queue CAV along the z direction at time t, respectively. Indicates that the k-th queue CAV is The state column vector along the z direction at each moment; p i (t) respectively represent CAVi - 、CAVi + , the position of CAV i, i - 、i + Respectively The vehicle numbers before and after vehicle i, Indicates the final headway distance, v i (t) respectively represent CAVi + , the speed of CAV i, u1(t), u i (t) respectively represent The acceleration of the first CAV, CAV i, represents the optimal set of CAVs of the kth queue in the z direction, and n represents The number of CAVs in the middle, β1, β2, β3 are weight parameters; represents the final position, represents the final velocity, Respectively expressed as The position and speed of the first CAV in the k-th queue at time z, α1 and α2 are weighting coefficients; express Derivative, and A is a 2n*2n coefficient matrix, B is a 2n*1 coefficient matrix, is an n×n matrix of all 1s, is an n×1 matrix of all 1s; v min 、v max Respectively represent the minimum and maximum speed, u min 、u max Respectively represent the minimum and maximum acceleration values, represents the optimal position of CAV j in the k-1th row in the z direction, represents the optimal set of the k-1th row of CAVs in the z direction, p i (t0) represents the initial position of CAV i of the kth queue in the z direction, D c Indicates the length of the control area. Represents a state set, Represents a collection of accelerations.

9. The intersection non-signal light optimization control system based on the intelligent networked fleet according to claim 8 is characterized in that: The model solving module specifically includes: Initialization unit, used to randomly generate an initial solution; The first model solving unit is used to search for the optimal solution of the intersection non-signal light optimization control model by an artificial bee colony search method according to the performance index of the CAV queue in the initial solution or the last iteration, and obtain the optimal formation mode of the CAV queue in this iteration; The second model solving unit is used to solve the CAV queue control model using a dynamic programming algorithm according to the optimal formation mode of the CAV queue solved by the upper model, so as to obtain the optimal acceleration and optimal state of each CAV in the CAV queue at each moment in this iteration; A performance indicator calculation unit, used to calculate the performance indicator of the CAV queue according to the optimal acceleration and optimal state in this iteration; The iteration judgment unit is used to return to execute the first model solving unit for the next iteration until the iteration reaches a preset number of times or meets a preset condition, and then terminate the iteration, and output the optimal formation mode of the CAV queue at the end of the iteration as well as the optimal acceleration and optimal state of the CAV queue as the optimal solution.

10. The method for optimizing control of intersections without signal lights based on an intelligent networked fleet according to claim 9, characterized in that: The second model solving unit specifically includes: Discretization subunit, used to divide the time interval The objective function of the CAV queue control model is discretized into R control steps, and the discretized objective function is obtained according to the control step. in, represents the control step length, r is the discrete index, They represent the state vector and acceleration vector of the k-th queue CAV along the z direction at control step r, and satisfy A r and B r is the coefficient matrix for the control step r, and A r =I n×n +A×Δt,B r =B×Δt,I n×n is the n×n identity matrix; The weight matrix calculation subunit is used to calculate the weight matrix Q of r∈{0,1,…,R} according to the following formula r , D r and O r : D r ＝0 in, are all intermediate variables. is an n×n matrix, is an n×1 matrix; The optimal solution determination subunit is used to set in, They represent the control step r=R when Q r , D r The optimal solution of The reverse derivation subunit is used to reversely deduce the intermediate variable G of r∈{R-1,…,1,0} in turn according to the following formula r , H r , S r , T r : S r =A r +B r G r T=BH rrr Among them, P r , G r , H r , S r , T r They are all intermediate variables; The state and acceleration column vector calculation subunit is used to calculate the acceleration column vector according to G r , H r , S r , T r The optimal state column vector and the optimal acceleration column vector at each control step r are calculated according to the following formula; x(r)=S r x(r-1)+T r ,r∈{1,…,R} u(r)=min[max(G r x(r)+H r ,u min ),u max ],r∈{1,…,R} In the formula, x(r) and x(r-1) represent the state column vectors at control steps r and r-1, respectively, and x(0)=[p(t0) T ,v(t0) T ] T , p(t0), v(t0) represent the initial time position vector and velocity vector, respectively, u(r) represents the acceleration column vector at control step r, u min 、u max are the minimum and maximum acceleration values ​​respectively; The optimal acceleration and optimal state determination subunit is used to obtain the optimal acceleration and optimal state of each CAV at each moment according to the optimal state column vector and the optimal acceleration column vector at each control step r.

Images