Multi-vehicle queue cooperative control method and medium based on distributed model predictive control

Abstract

The present invention discloses a multi-vehicle queue cooperative control method based on distributed model predictive control, which includes the following steps: constructing a multi-vehicle queue system and a dynamic model, establishing a unidirectional communication topology structure, defining the coupling constraints of vehicles and queues based on the communication topology structure, constructing a distributed model predictive control optimization problem and a cost function, where the distributed model predictive control optimization problem includes a vehicle node local optimization problem, a queue global optimization problem, and a multi-objective optimization problem; solving the optimization control problem for each vehicle node and adjusting the control inputs and motion states of each vehicle. In the technical solution of the present invention, the unidirectional communication topology design, the introduction of the coupling constraints within and between queues, the construction of local optimization problems, and the combination of the global coordination mechanism are used to achieve dynamic synchronization and stable control between vehicles and between queues, and can effectively improve the cooperative control performance of the multi-vehicle queue system in a complex dynamic environment.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle platoon cooperative control, and particularly relates to a multi-vehicle platoon cooperative control method and medium based on distributed model predictive control. Background Art

[0002] With the rapid development of intelligent transportation systems and modern logistics, the cooperative control of large-scale vehicle platoons has become an important research direction in the field of autonomous driving. In a one-way communication environment, there are problems of delay and dynamic changes in information transmission between vehicles, and existing control methods are difficult to achieve efficient cooperation and global stability in a multi-platoon system. Existing centralized control methods are difficult to meet the real-time cooperation requirements of large-scale vehicle fleets due to high computational complexity, large communication delays, and single-point failure risks. Distributed model predictive control (DMPC), due to its characteristics of combining local optimization and global coordination, has become an effective method to solve this problem. However, current research still has deficiencies in communication topology design, computational complexity control, and dynamic environment adaptability. Therefore, for the problem of multi-platoon vehicle trajectory synchronization in a one-way communication environment, it is necessary to propose an efficient, stable, and dynamic environment-adaptive control method to achieve global consistency and robustness of the platoon system. Summary of the Invention

[0003] The main object of the present invention is to provide a multi-vehicle platoon cooperative control method based on distributed model predictive control, aiming to solve the problems that existing distributed model predictive control methods still have deficiencies in communication topology design, computational complexity control, and dynamic environment adaptability in the cooperative control application of vehicle platoons.

[0004] To achieve the above object, the multi-vehicle platoon cooperative control method based on distributed model predictive control proposed by the present invention includes the following steps:

[0005] Construct a multi-vehicle platoon system and a dynamic model, where the dynamic model includes a system state vector, a motion state equation of a single vehicle, and a platoon length state equation;

[0006] Establish a one-way communication topology structure, where the communication topology structure includes an intra-platoon communication topology and an inter-platoon communication topology. The intra-platoon communication topology includes a directed spanning tree with the leader vehicle as the root node, and the inter-platoon communication topology includes a directed spanning tree with the leader vehicle of the first platoon in the system as the root node;

[0007] Based on the communication topology structure, coupling constraints of vehicles and queues are defined, and a distributed model predictive control optimization problem and a cost function are constructed. The distributed model predictive control optimization problem includes a local optimization problem of vehicle nodes, a global optimization problem of queues, and a multi-objective optimization problem that integrates the local optimization problem of vehicle nodes and the global optimization problem of queues. The local optimization problem of vehicle nodes is used to solve the local optimal control input for each vehicle node, and the global optimization problem of queues is used to solve the coordinated control input between queues for each queue node by using the multi-objective optimization problem.

[0008] Solve the optimization control problem for each vehicle node, and adjust the control input and motion state of each vehicle according to the optimization result to achieve cooperative control within and between queues.

[0009] Preferably, the steps of constructing the multi-vehicle queue system and the dynamic model, where the dynamic model includes a system state vector, a motion state equation of a single vehicle, and a queue length state equation, include:

[0010] Define the update rules of vehicle states and vehicle state variables. The vehicle state vector includes position, velocity, acceleration, and torque, and use a discrete-time model to construct the dynamic equation of a single vehicle.

[0011] Define the queue length as the distance between the leading vehicle and the trailing vehicle, and establish a dynamic change equation of the queue length.

[0012] Preferably, the local optimization problem of vehicle nodes includes a vehicle local cost function and vehicle constraint conditions. The vehicle constraint conditions include vehicle prediction model constraints, the constraint that the initial prediction state is equal to the actual state at time t, vehicle state constraints, vehicle control input constraints, vehicle position constraints, and constraints of recursive feasibility and stability.

[0013] Preferably, the vehicle local cost function includes:

[0014] ;

[0015] In the formula: and respectively represent the predicted state and the predicted control input sequence of vehicle node i at time t, represents the assumed state of vehicle node i at time t; represents the assumed state of the vehicle in front of vehicle node i at time t, represents the distance between vehicle node i and the vehicle in front; the weight matrix is set as a symmetric non-negative definite matrix, and respectively represents the control state error, control input, assumed state error, and vehicle distance error in front.

[0016] Preferably, the queuing global optimization problem includes a queuing global cost function and queuing constraint conditions. The queuing constraint conditions include communication topology constraints, queuing position constraints, and queuing length change constraints. The communication topology constraints are used to constrain the state update of the leading vehicle to depend on the state information of other leading vehicles defined by the communication topology, ensuring that the behavior of the leading vehicle is restricted by the communication graph structure and can only obtain information from neighbors in the communication topology. The queuing position constraints are set for the position trajectories of the leading vehicle and the trailing vehicle of the queue to ensure that the spatial constraints between queues are satisfied.

[0017] Preferably, the queuing global cost function includes:

[0018] ;

[0019] In the formula: represents the predicted state sequence of the leading vehicle; represents the predicted control input sequence of the leading vehicle, represents the length of the (k - 1)th queue; represents the distance error between adjacent leading vehicles of queues, represents the distance between the leading vehicle of queue k and the trailing vehicle of the previous queue; is set as a symmetric non - negative definite matrix, which respectively controls the state error, control input, assumed state error, and the distance error of the leading vehicle, to ensure the stability and coordination of the vehicle in the queue.

[0020] Preferably, the step of adjusting the motion states of each vehicle according to the optimization result, where the motion states include position, speed, and acceleration, to achieve cooperative control within and between queues includes:

[0021] Initializing the vehicle state, input state, and queue length;

[0022] For each queue, generating the desired trajectory of the leading vehicle within the prediction time domain;

[0023] At each moment, solving the optimization control problem for each vehicle node, and according to the optimization result, obtaining the optimal control input sequence to achieve cooperative control within and between queues.

[0024] Preferably, the step of solving the optimization control problem for each vehicle node at each moment to obtain the optimal control input sequence to achieve cooperative control within and between queues includes:

[0025] In the continuously looping queuing state update step, node state update step, and information exchange and synchronization step at each moment, gradually updating the control inputs and states of all queues and vehicle nodes to achieve cooperative control within and between queues;

[0026] The queue state update steps include: updating the state sequence of the leading vehicle to ensure following the desired trajectory; within the prediction horizon, updating the state of each vehicle according to the optimal control input sequence, calculating the next assumed input sequence of the leading vehicle and updating the assumed state sequence;

[0027] The node state update steps include: updating the current node state using the optimal control input, within the prediction horizon, calculating the optimal state sequence and updating the node state, updating the assumed input sequence of the vehicle, and predicting the assumed state sequence based on the assumed input sequence;

[0028] The information exchange and synchronization steps include: sending the assumed state sequence and assumed input sequence of the queue leading vehicle node to adjacent leading vehicle nodes, and sending the node assumed state sequence and assumed input sequence to adjacent following vehicle nodes to achieve information sharing.

[0029] Preferably, after the step of establishing a unidirectional communication topology, the communication topology includes an intra - queue communication topology and an inter - queue communication topology. The intra - queue communication topology includes a directed spanning tree with the leading vehicle as the root node, and the inter - queue communication topology includes a directed spanning tree with the leading vehicle of the first queue in the system as the root node, the following steps are included:

[0030] Establishing multiple communication strategies, the communication strategies include a leading - vehicle - to - leading - vehicle strategy, a leading - vehicle - to - rear - vehicle strategy, and a leading - vehicle - to - rear - vehicle strategy; the leading - vehicle - to - leading - vehicle strategy includes a communication method where information is passed step - by - step through the queue leading vehicle, the leading - vehicle - to - rear - vehicle strategy includes a communication method where the leading vehicle directly receives the information of the rear vehicle of the previous queue, and the leading - vehicle - to - rear - vehicle strategy includes a communication method that allows the leading vehicle to receive the information of both the leading vehicle and the rear vehicle of the previous queue simultaneously;

[0031] Selecting different communication strategies according to the queue length, response speed, and dynamic environment.

[0032] A multi - vehicle queue cooperative control medium based on distributed model predictive control stores executable instructions, and the executable instructions contain program codes for executing the multi - vehicle queue cooperative control method based on distributed model predictive control as described in any one of the above when being executed.

[0033] In the technical solution of the present invention, the unidirectional communication topology design: establish the communication structure inside and between queues. Inside the queue, a directed tree structure with the leading vehicle as the root node is adopted, and between queues, a directed graph structure with the leading vehicle of the system's first queue as the root node is adopted to achieve low-latency and efficient information transmission. The distributed model predictive control method: by introducing the coupling constraints inside and between queues, construct a local optimization problem, and combine with the global coordination mechanism to achieve dynamic synchronization and stable control between vehicles and between queues, which can effectively improve the cooperative control performance of the multi-vehicle queue system in a complex dynamic environment, and realize efficient, stable and safe vehicle formation operation, with great improvements in communication topology design, computational complexity control and dynamic environment adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on the structures shown in these drawings.

[0035] Figure 1 It is the communication topology structure diagram of the leading vehicle - leading vehicle in the multi-vehicle queue cooperative control method based on distributed model predictive control of the present invention;

[0036] Figure 2 It is the speed change diagram under the leading vehicle - leading vehicle communication structure in the multi-vehicle queue cooperative control method based on distributed model predictive control of the present invention (the red, green, and blue vehicle fleets, and the dotted line is the leading vehicle);

[0037] Figure 3 It is the displacement change diagram under the leading vehicle - leading vehicle communication structure in the multi-vehicle queue cooperative control method based on distributed model predictive control of the present invention (the red, green, and blue vehicle fleets, and the dotted line is the leading vehicle);

[0038] Figure 4 It is the torque change diagram under the leading vehicle - leading vehicle communication structure in the multi-vehicle queue cooperative control method based on distributed model predictive control of the present invention (the red, green, and blue vehicle fleets, and the dotted line is the leading vehicle);

[0039] Figure 5 It is the acceleration change diagram under the leading vehicle - leading vehicle communication structure in the multi-vehicle queue cooperative control method based on distributed model predictive control of the present invention (the red, green, and blue vehicle fleets, and the dotted line is the leading vehicle);

[0040] Figure 6 It is the communication topology structure diagram of the leading vehicle - trailing vehicle in the multi-vehicle queue cooperative control method based on distributed model predictive control of the present invention;

[0041] Figure 7 The speed change diagram under the leader-follower communication structure in the multi-vehicle queue cooperative control method based on distributed model predictive control of the present invention (red, green, and blue vehicle fleets, the dotted line represents the leader vehicle);

[0042] Figure 8 The displacement change diagram under the leader-follower communication structure in the multi-vehicle queue cooperative control method based on distributed model predictive control of the present invention (red, green, and blue vehicle fleets, the dotted line represents the leader vehicle);

[0043] Figure 9 The torque change diagram under the leader-follower communication structure in the multi-vehicle queue cooperative control method based on distributed model predictive control of the present invention (red, green, and blue vehicle fleets, the dotted line represents the leader vehicle);

[0044] Figure 10 The acceleration change diagram under the leader-follower communication structure in the multi-vehicle queue cooperative control method based on distributed model predictive control of the present invention (red, green, and blue vehicle fleets, the dotted line represents the leader vehicle);

[0045] Figure 11 The leader-leader-follower communication topology diagram in the multi-vehicle queue cooperative control method based on distributed model predictive control of the present invention;

[0046] Figure 12 The speed change diagram under the leader-leader-follower communication structure in the multi-vehicle queue cooperative control method based on distributed model predictive control of the present invention (red, green, and blue vehicle fleets, the dotted line represents the leader vehicle);

[0047] Figure 13 The displacement change diagram under the leader-leader-follower communication structure in the multi-vehicle queue cooperative control method based on distributed model predictive control of the present invention (red, green, and blue vehicle fleets, the dotted line represents the leader vehicle);

[0048] Figure 14 The torque change diagram under the leader-leader-follower communication structure in the multi-vehicle queue cooperative control method based on distributed model predictive control of the present invention (red, green, and blue vehicle fleets, the dotted line represents the leader vehicle);

[0049] Figure 15 The acceleration change diagram under the leader-leader-follower communication structure in the multi-vehicle queue cooperative control method based on distributed model predictive control of the present invention (red, green, and blue vehicle fleets, the dotted line represents the leader vehicle);

[0050] Figure 16 The weight matrix values of the DMPC controller in one embodiment of the multi-vehicle queue cooperative control method based on distributed model predictive control of the present invention.

[0051] The implementation, functional features, and advantages of the present invention will be further described in conjunction with embodiments with reference to the accompanying drawings. Specific embodiments

[0052] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts belong to the scope of protection of the present invention.

[0053] In addition, in the present invention, descriptions such as "first" and "second" are only for descriptive purposes, and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include at least one of such features. In the description of the present invention, the meaning of "a plurality" is at least two, such as two, three, etc., unless otherwise specifically and clearly defined.

[0054] In the present invention, unless otherwise clearly specified and defined, terms such as "connection" and "fixation" should be understood in a broad sense. For example, "fixation" may be a fixed connection, a detachable connection, or integrated; it may be a mechanical connection or an electrical connection; it may be directly connected or indirectly connected through an intermediate medium, and may be the internal communication of two components or the interaction relationship between two components, unless otherwise clearly defined. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0055] In addition, the technical solutions between the various embodiments of the present invention can be combined with each other, but it must be based on the ability of those of ordinary skill in the art to implement. When the combination of technical solutions results in contradictions or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0056] Please refer to Figures 1-16 , the present invention proposes a multi-vehicle queue cooperative control method based on distributed model predictive control, including the following steps:

[0057] S100. Construct a multi-vehicle queue system and a dynamic model, where the dynamic model includes a system state vector, a motion state equation of a single vehicle, and a queue length state equation;

[0058] S200. Establish a unidirectional communication topology structure, where the communication topology structure includes an intra-queue communication topology and an inter-queue communication topology. The intra-queue communication topology includes a directed spanning tree with the leading vehicle as the root node, and the inter-queue communication topology includes a directed spanning tree with the leading vehicle of the system's first queue as the root node;

[0059] S300. Based on the communication topology structure, define the coupling constraints of vehicles and queues, and construct a distributed model predictive control optimization problem. The distributed model predictive control optimization problem includes a local optimization problem for vehicle nodes, a global optimization problem for queues, and a multi-objective optimization problem that integrates the local optimization problem for vehicle nodes and the global optimization problem for queues. The local optimization problem for vehicle nodes is used to solve the local optimal control input for each vehicle node, and the global optimization problem for queues is used to solve the coordinated control input between queues for each queue node using the multi-objective optimization problem;

[0060] S400. Solve the optimization control problem for each vehicle node, and adjust the control input and motion state of each vehicle according to the optimization results to achieve cooperative control within and between queues.

[0061] In the technical solution of the present invention, the unidirectional communication topology design: establish the communication structure within and between queues. The intra-queue adopts a directed tree structure with the leading vehicle as the root node, and the inter-queue adopts a directed graph structure with the leading vehicle of the system's first queue as the root node to achieve low-latency and efficient information transfer. The distributed model predictive control method: by introducing the coupling constraints within and between queues, construct local optimization problems, and combine with the global coordination mechanism to achieve dynamic synchronization and stable control between vehicles and between queues.

[0062] In another embodiment of the present invention, the steps of S100 include:

[0063] S110. Define the vehicle state and the update rules of vehicle state variables. The vehicle state vector includes position, speed, acceleration, and torque, and use the discrete-time model to construct the dynamic equation of a single vehicle;

[0064] S120. Define the queue length as the distance between the leading vehicle and the trailing vehicle, and establish the dynamic change equation of the queue length.

[0065] Specifically, define a multi-vehicle queue system

[0066] Consider a multi-queue system composed of vehicle queues (including a leading queue LP and M following queues FP). The queue numbers range from 0 to M, where number 0 is the leading queue of the queue system, and the remaining following queue numbers are . Each queue k contains A vehicle (including a leading vehicle LV and a number of following vehicles FV), the vehicle numbers range from 0 to , where the number 0 is the leading vehicle of the queue, and the number is the following vehicle, as Figure 1 shown. Define the set of vehicles in the queue k as: . The set of all vehicles in the system is , which can be expressed as:

[0067] ;

[0068] Construct the dynamic model of the multi - queue system:

[0069] Define the state vector of the system as .

[0070] Among them , , and respectively represent the position, speed, acceleration, and driving force or braking torque of the k th vehicle in the i th queue at time t .

[0071] The present invention assumes that is the discrete time interval, then the discrete - time model of the k th vehicle in the i th queue can be expressed as:

[0072] ;

[0073] Among them: , , , respectively represent the total force, vehicle mass, control input, and inertial lag of longitudinal dynamics acting on the k th vehicle in the i th queue per unit discrete time.

[0074] Therefore, the motion state of the k th vehicle in the i th queue can be expressed as:

[0075] ;

[0076] ;

[0077] .

[0078] Among them, , respectively represent the k th queue's i vehicle's state and the weight matrix of the control input, represents the tire radius, is the mechanical efficiency of the powertrain.

[0079] The queue system is composed of multiple queues. The relative positions, speeds, and dynamic behaviors of the queues will affect each other. The present invention introduces to describe the length or density state of the queue, and defines as the distance from the leading vehicle to the last vehicle in the queue, that is , where , respectively represent the k leading vehicle and the position of the last vehicle of the

[0080] th queue.

[0081] ;

[0082] Therefore, the change in the length state of the queue can be expressed as:

[0083] .

[0084] Among them, represents the speed of the leading vehicle at time t of the kth queue, represents the speed of the last vehicle at time t of the kth queue.

[0085] For each vehicle i in the queue and the previous vehicle, coupling control is achieved through a fixed distance . At the same time, in order to ensure the cooperative control between multiple queues, it is defined that each queue is coupled with the adjacent queue through a fixed distance constraint .

[0086] Hypothetical state output definition. Based on the optimal solution at the previous moment, a hypothetical state output is generated to predict the system behavior at the next moment.

[0087] ;

[0088] ;

[0089] Among them: represents t the optimal state sequence at time represents the hypothetical state sequence at the next moment;

[0090] Terminal constraint design:

[0091] Terminal state constraint design, which defines the terminal state constraints for all vehicles, ensures that the state at the end of the prediction horizon can meet the requirements of platoon synchronization, and guarantees recursive feasibility.

[0092] ;

[0093] In the formula: represents the terminal predicted state of the k th i vehicle at the t moment, represents the terminal assumed state of the k th i vehicle's leading vehicle at the t moment, represents the minimum safe distance between vehicles.

[0094] Prediction horizon setting. According to the vehicle dynamics characteristics and the platoon dynamic behavior, determine the length of the prediction horizon so that the controller can optimize for a sufficiently long future time period.

[0095] Cost function design:

[0096] State error cost function design. Set the state error cost function to control the state error, control input, assumed state error, and leading vehicle distance error respectively, to ensure the stability and coordination of the vehicle in the platoon, punish the error of the vehicle deviating from the desired state, and optimize the state consistency inside and outside the platoon during the path tracking process.

[0097] 1. Control state error:

[0098] ;

[0099] 2. Control input magnitude:

[0100] ;

[0101] 3. Assumed state error

[0102] ;

[0103] 4. Leading vehicle distance error

[0104] ;

[0105] In the formula: represents the predicted state sequence, represents the assumed state sequence, represents the assumed state sequence of the leading vehicle, represents the desired state sequence, represents the predicted control input sequence, represents the vehiclei Distance error from the vehicle ahead.

[0106] Control input constraint definition. Incorporate the physical limits of the control input (such as maximum acceleration, maximum steering angle) to ensure that the solution of the optimization problem can meet the operating conditions of the actual vehicle.

[0107] ;

[0108] In the formula: Denotes the minimum control input, Denotes the maximum control input.

[0109] In another embodiment of the present invention, the local optimization problem of the vehicle node includes a vehicle local cost function and vehicle constraint conditions. The vehicle constraint conditions include vehicle prediction model constraints, the constraint that the initial prediction state is equal to the actual state at time t, vehicle state constraints, vehicle control input constraints, vehicle position constraints, and constraints of recursive feasibility and stability.

[0110] The expression of the local optimization problem of the vehicle node includes:

[0111] ;

[0112] Specifically, in the formula: Denotes k The vehicle node in the queue i The cost function of the local optimization problem; Denotes the prediction horizon; Denotes k The vehicle node in the queue i At t The state at time, And Respectively denote the predicted state and predicted control input sequence of vehicle node i at t Time, Denotes the assumed state of vehicle node i at time t, Denotes the assumed state of the vehicle ahead of vehicle node i at time t; , Denotes k The vehicle node in the queue i At t The control input at time.

[0113] The constraint formulas respectively represent the constraints of the prediction model of vehicle i ; The constraint that the initial prediction state is equal to the actual state at time t ; The state constraint of vehicle k In the queue i ; The state constraint of vehicle k In the queue iInput constraints; for vehicles i Position trajectory Set to ensure that the spatial formation constraints within the queue are met; constraints to ensure recursive feasibility and stability.

[0114] More specifically, the control objective within the queue is defined. The control objective for each vehicle is set to follow the vehicle in front, maintain a fixed safe distance, and gradually approach the speed of the leading vehicle while achieving stability within the queue.

[0115] ;

[0116] Where: Represents the safe distance between vehicles.

[0117] In another embodiment of the present invention, the local cost function includes:

[0118] ;

[0119] Where: And Respectively represent the predicted state and predicted control input sequence of vehicle node i at time t, Represents the assumed state of vehicle node i at time t; Represents the assumed state of the vehicle in front of vehicle node i at time t, Represents the distance between vehicle node i and the vehicle in front; the weight matrix Is set as a symmetric non - negative definite matrix, respectively representing the control state error, control input, assumed state error, and distance error of the vehicle in front.

[0120] Specifically, And Respectively represent the predicted state and predicted control input sequence of vehicle node i at time t, Represents the assumed state of vehicle node i at time t; Represents the assumed state of the vehicle in front of vehicle node i at time t, Represents the distance between vehicle node i and the vehicle in front.

[0121] In another embodiment of the present invention, the queue global optimization problem includes a queue global cost function and queue constraint conditions. The queue constraint conditions include communication topology constraints, queue position constraints, and queue length change constraints. The communication topology constraints are used to restrict the state update of the leading vehicle to depend on the state information of other leading vehicles defined by the communication topology, ensuring that the behavior of the leading vehicle is restricted by the communication graph structure and can only obtain information from neighbors in the communication topology. The queue position constraints are set for the position trajectories of the leading vehicle and the trailing vehicle in the queue to ensure that the spatial constraints between queues are met.

[0122] The cost function of the queue global optimization problem includes:

[0123] ;

[0124] Where: For queue k The amount of information received by the pilot car

[0125] Integration of multi-objective optimization problems. Integrate local optimization goals with global optimization goals into multi-objective optimization problems to ensure the global coordination of the system.

[0126] ;

[0127] Where: are the weight coefficients of the coordinated control of node i and queue k, represents the cost function of the local optimization problem of queue k, Cost function of the local optimization problem of vehicle nodes.

[0128] In another embodiment of the present invention, the controller is modularly designed, and the distributed model predictive controller is divided into a local optimization module, an information sharing module and a state update module, which is convenient for actual system implementation.

[0129] Specifically, the control objectives between queues are defined to ensure that adjacent pilot vehicles between queues maintain a fixed safety distance and synchronize their speeds to avoid collisions and incoordination between queues.

[0130] ;

[0131] In another embodiment of the present invention, the queue global cost function includes:

[0132] ;

[0133] Where: represents the predicted state sequence of the pilot vehicle; represents the predictive control input sequence of the pilot vehicle, represents the distance error between adjacent pilot vehicles in the queue, represents the distance error between the leading vehicle in queue k and the last vehicle in the previous queue, represents the k-1 queue length; represents the distance error between adjacent pilot vehicles in the queue, represents the distance between the leading vehicle in queue k and the last vehicle in the previous queue; It is set as a symmetric non-negative definite matrix, which controls the state error, control input, assumed state error and leading vehicle distance error respectively to ensure the stability and coordination of vehicles in the queue.

[0134] Weight matrix is set as a symmetric non - negative definite matrix, which respectively controls the state error, control input, assumed state error and leading vehicle distance error to ensure the stability and coordination of vehicles in the queue.

[0135] In another embodiment of the present invention, the steps of S400 include:

[0136] S410. Initialize the vehicle state, input state and queue length;

[0137] S420. For each queue, generate the desired trajectory of the leading vehicle within the prediction horizon;

[0138] S430. At each moment, solve the optimal control problem for each vehicle node, and according to the optimization result, obtain the optimal control input sequence to achieve the cooperative control within and between queues.

[0139] Specifically, initialization

[0140] Vehicle state initialization: Set the initial state of each vehicle, including the specific states of the leading vehicle and following vehicles. At the initial time , the state of each vehicle is expressed as:

[0141] ;

[0142] Among them, . Assume that the initial speed of each vehicle in the queue is , ensure that the vehicle distance meets the minimum safety distance , and the queue distance meets the minimum safety distance .

[0143] S410. Input state initialization: Initialize the assumed input sequence and assumed state sequence of each node for predicting future inputs and states.

[0144] Queue length initialization: Define the queue length as the distance between the leading vehicle and the trailing vehicle. At the initial moment: , where: represents the number of vehicles in the queue, and the queue distance meets the minimum safety distance.

[0145] S420. Desired trajectory generation:

[0146] For each queue k , generate the desired trajectory of the leading vehicle within the prediction horizon

[0147] ;

[0148] wherein, , represents the expected position of the platoon leader vehicle, represents the expected speed of the platoon leader vehicle, ensuring that the platoon leader vehicle can maintain synchronization according to the distance between platoons.

[0149] S430. Solving the optimal control problem

[0150] At each moment t , for each vehicle node i solve the optimal control problem to obtain the optimal control input sequence , wherein .

[0151] In another embodiment of the present invention, the steps of S430 include:

[0152] S431. The continuously looped platoon state update step, node state update step, and information exchange and synchronization step at each moment, gradually updating the control inputs and states of all platoons and vehicle nodes to achieve cooperative control within and between platoons;

[0153] S432. The platoon state update step includes: updating the state sequence of the leader vehicle to ensure following the expected trajectory; within the prediction horizon, updating the state of each vehicle according to the optimal control input sequence, and calculating and updating the next assumed input sequence and the assumed state sequence of the leader vehicle;

[0154] S433. The node state update step includes: updating the current node state using the optimal control input, calculating and updating the optimal state sequence and the node state within the prediction horizon, updating the assumed input sequence of the vehicle, and predicting the assumed state sequence based on the assumed input sequence;

[0155] S434. The information exchange and synchronization step includes: sending the assumed state sequence and the assumed input sequence of the platoon leader vehicle node to the adjacent platoon leader vehicle nodes, and sending the assumed state sequence and the assumed input sequence of the node to the adjacent follower vehicle nodes to achieve information sharing.

[0156] Specifically, the solution process:

[0157] S431. Loop iteration:

[0158] At each moment t , continuously loop from S432 (Step1) to S434 (Step3), gradually updating the control inputs and states of all platoons and vehicle nodes until the entire system reaches a synchronous and stable state.

[0159] S432 (Step1): Queue status update:

[0160] (1) Update the status sequence of the leading vehicle to ensure it follows the desired trajectory.

[0161] (2) Update the status of each vehicle within the prediction horizon according to the optimal input Update the status of each vehicle.

[0162] (3) Calculate the next assumed input sequence of the leading vehicle and update the assumed status sequence ; queue length sequence .

[0163] S433 (Step2): Node status update:

[0164] (1) Update the current node status using the optimal control input Update the current node status.

[0165] (2) Calculate the optimal status sequence within the prediction horizon and update the node status.

[0166] (3) Update the assumed input sequence of the vehicle .

[0167] (4) Predict the assumed status sequence based on the assumed input sequence Predict the assumed status sequence:

[0168] .

[0169] S434 (Step3): Information exchange and synchronization:

[0170] (1) Send the assumed status sequence and assumed input sequence of the queue leading vehicle node and the assumed input sequence to the neighboring leading vehicle nodes.

[0171] (2) Send the assumed status sequence and assumed input sequence of the node and the assumed input sequence to the neighboring following vehicle nodes to achieve information sharing and ensure the overall optimization effect.

[0172] In another embodiment of the present invention, after the steps of S200, it includes:

[0173] S210. Establish multiple communication strategies, where the communication strategies include lead vehicle to lead vehicle strategy, lead vehicle to trailing vehicle strategy, and lead vehicle to trailing vehicle strategy; the lead vehicle to lead vehicle strategy includes a communication method where information is passed step by step through the queue lead vehicles, the lead vehicle to trailing vehicle strategy includes a communication method where the lead vehicle directly receives the information of the trailing vehicle of the previous queue, and the lead vehicle to trailing vehicle strategy includes a communication method that allows the lead vehicle to receive the information of the lead vehicle and the trailing vehicle of the previous queue simultaneously;

[0174] S220. Select different communication strategies according to the queue length, response speed, and dynamic environment.

[0175] Specifically, multiple communication strategies adapt to diverse requirements: Provide lead vehicle to lead vehicle, lead vehicle to trailing vehicle, and comprehensive communication strategies, which are respectively applicable to application scenarios of medium scale, rapid response, and complex dynamic environment, providing a flexible solution for fleet control with different requirements.

[0176] Communication topology design (within the queue). Design the communication topology structure within a single queue, use the lead vehicle as the root node to construct a spanning tree to ensure that each vehicle within the queue can obtain the status information of its preceding vehicle through communication. Specifically: Establish a unidirectional communication topology structure with the lead vehicle of each queue as the core, including the intra-queue communication topology and the inter-queue communication topology

[0177] Within each queue k In the present invention, a directed graph is considered , where is the set of communication edges between different vehicles within the queue k . In the graph the edge indicates that vehicle i can receive the status information transmitted by vehicle j . The present invention defines as the set of following vehicles that the k nd following vehicle in the i th queue can receive information from. In addition, to consider the communication topology structure of the lead vehicle, the present invention defines a graph that includes a directed spanning tree with all lead vehicles as the root nodes, where is the set of lead vehicle nodes in the queue system, is the set of communication edges between queues in the queue system to ensure information transfer between queues.

[0178] Define as the set of vehicles that the queue lead vehicle can receive information from. The lead vehicles of each queue exchange information through and pass the received cross-queue information to the following vehicles in their own queues, thereby realizing cooperative control between multiple queues.

[0179] The communication topology inside the queue is a directed spanning tree, with the leading vehicle as the root node.

[0180] Let be the k adjacency matrix of the queue.

[0181] Among them, represents that the j th following vehicle can send information to the i th following vehicle.

[0182] All nodes start from the root node 0 to form a directed spanning tree, ensuring that all following vehicles can directly or indirectly obtain the status information of the leading vehicle.

[0183] Then the adjacency matrix can be expressed as:

[0184] ;

[0185] Define the Laplacian matrix k of the queue to represent the connection relationship between the vehicles inside the queue. It is constructed from the in-degree matrix and the adjacency matrix , and is expressed as , where represents the amount of information received by each node from other nodes.

[0186] Then the Laplacian matrix for the communication inside the queue is defined as:

[0187] ;

[0188] Among them, represents the amount of information received by the leading vehicle of the queue k .

[0189] Communication topology design (between queues):

[0190] Design the communication topology of the leading vehicles between queues, and use a directed graph to represent the information transfer path between the leading vehicles to ensure cross-queue cooperative control.

[0191] Specifically, the leading vehicle of each queue can receive specific information from the previous queue. For this purpose, for different strategies, define the communication matrix between queues, and at the same time define the system global adjacency matrix and the global Laplacian matrix .

[0192] System global state definition. The state set of the entire multi - queue system is defined as a global state vector, including the states (position, speed, acceleration) of all vehicles and the state coupling between queues.

[0193] Queue state update rule. In the system model, clarify the update rule of vehicle states within each time step to ensure that state updates can consider the dynamic impacts of communication topology and control inputs.

[0194] Information exchange rule between queues. Design an information exchange mechanism between leading vehicles to ensure that the leading vehicle of each queue can timely obtain relevant state information of other queues. Specifically, it includes the following:

[0195] Strategy a: Leading - vehicle - leading - vehicle tracking strategy. The leading vehicle of each queue only receives information from the leading vehicle of the previous queue. Then the communication matrix between queues can be expressed as:

[0196] ;

[0197] The adjacency matrix is:

[0198] ;

[0199] The Laplacian matrix is:

[0200] ;

[0201] Strategy b: Leading - vehicle - trailing - vehicle tracking strategy. Only the leading vehicle of each queue receives information from the trailing vehicle of the previous queue. Therefore, there is a communication connection between the leading vehicle and the trailing vehicle. Strategy b has the same structure for its adjacency matrix and Laplacian matrix as strategy a. The difference lies in the representation of its communication matrix between queues, which can be specifically expressed as:

[0202] ;

[0203] Strategy c: Combined strategy of leading - vehicle and leading - vehicle - trailing - vehicle. Combining the above two strategies, the leading vehicle can receive information from both the leading vehicle and the trailing vehicle of the previous queue. Therefore, its communication matrix between queues is expressed as:

[0204] ;

[0205] Leading - vehicle - to - leading - vehicle strategy (Strategy a): Information is passed step - by - step through the leading vehicles of queues, with high consistency and low communication load, and is suitable for medium - length queues. However, due to the characteristic of step - by - step information transmission, there may be a response lag in longer queues, which limits the dynamic response ability of the system.

[0206] Lead vehicle to trailing vehicle strategy (Strategy b): The lead vehicle directly receives information from the trailing vehicle of the previous queue, enhancing the system's dynamic response ability and being suitable for small-scale scenarios that require quick response. However, as the queue size increases, this strategy performs poorly in terms of stability.

[0207] Lead vehicle to trailing vehicle strategy (Strategy c): This strategy combines the previous two methods, allowing the lead vehicle to receive information from both the lead vehicle and the trailing vehicle of the previous queue simultaneously, thus taking into account both global and local state changes, balancing responsiveness and stability, and being very suitable for complex dynamic environments that require high consistency and quick response. However, its communication complexity is relatively high.

[0208] The method proposed in this application can effectively improve the global coordination and adaptability of the multi-queue system under complex dynamic conditions, providing a robust solution for realizing cooperative control of multi-vehicle queues.

[0209] In another embodiment of the present invention, there are executable instructions stored, and the executable instructions contain program codes for executing the cooperative control method of multi-vehicle queues based on distributed model predictive control as described in any one of the above when being executed.

[0210] In another embodiment of the present invention, in order to verify the effectiveness of the present invention in the multi-vehicle queue system, the performance of the established model and algorithm is evaluated through numerical simulation.

[0211] The following are the specific verification analysis process and results:

[0212] Simulation background: The initial conditions are set such that all vehicles have no initial displacement error and speed error to ensure that the system is in a balanced state at the starting moment. The reference trajectory of the lead vehicle is designed as an acceleration-deceleration cycle to simulate the acceleration and braking behaviors of vehicles during actual operation. Specifically, the lead vehicle performs uniform acceleration motion with an acceleration of 2 m / s² within 1 to 2 seconds to increase the overall speed of the queue; and performs uniform deceleration motion with an acceleration of -2 m / s² within 3 to 4 seconds to reduce the vehicle speed. This reference trajectory provides a stable following benchmark for the subsequent vehicles, thus ensuring the response stability and queue consistency of the system under dynamic changing conditions.

[0213] Parameter settings: In this simulation, three queues (Red 9, Blue 7, Green 7) with a total of 23 vehicles are set. The vehicle mass is 1500 kg, the driving time delay is 0.7 s, the air resistance coefficient is 1, and the tire radius is 0.35 m.

[0214] The system time step is 0.1 s, the friction coefficient is 0.01, the transmission efficiency is 0.96, the gravitational acceleration is taken as 9.8 m / s², and the prediction horizon is 20 steps. The desired distance between vehicles is set to 10 m, and the desired distance between queues is 30 m. , .

[0215] The weight matrices of the remaining distributed model predictive control (DMPC) controllers are set as Figure 16 . These parameters ensure the effectiveness and accuracy of the multi-queue system simulation under different strategies.

[0216] Strategy a: Leader-follower tracking strategy ( Figure 1 ). The leader of each queue only receives the information of the leader of the previous queue. At this time , , the communication matrix between queues and the system's global adjacency matrix can be expressed as:

[0217] ;

[0218] ;

[0219] In strategy a, the communication topology has a hierarchical structure. The vehicles within a queue form a chain communication, while the cross-queue communication only occurs between the leaders. This design ensures that information is transmitted step by step from the leader of the front queue to the leader of the rear queue, thus constructing a multi-level control system with the leader as the core. From Figures 2-5 it can be seen that this communication structure effectively reduces the communication load of the system. At the same time, the step-by-step transmission method enhances the overall stability of the queue, making the system respond consistently in terms of speed, position, torque, and acceleration, and is suitable for the cooperative control of large-scale vehicle fleets with high consistency requirements.

[0220] Strategy b: Leader-trailer tracking strategy ( Figure 6 ). Only the leader of each queue receives the information of the trailer of the previous queue. There is , . Therefore, there is a communication connection between the leader and the trailer. Strategy b and strategy a have the same structure for their adjacency matrices. The difference lies in the representation of their inter-queue communication matrices, which can be specifically expressed as:

[0221] ;

[0222] In contrast, in strategy b, the leader obtains information from the trailer, enabling the next queue to respond more promptly to the state changes at the rear of the previous queue, thereby enhancing the safety and response speed of the system. Figures 7-10, it can be seen that this design improves the sensitivity of the queue to the movement changes of the leading vehicle. However, since the state of the trailing vehicle is easily affected by various factors, it increases the uncertainty and may have an adverse impact on the control stability of the leading vehicle in the next queue. Therefore, Strategy b is more suitable for scenarios with a small scale and a moderate queue length. In this case, the uncertainty of the trailing vehicle state will not be significantly amplified, and the information transmission delay is also small, which is conducive to the system quickly responding to the changes ahead and achieving high dynamic adaptability and safety.

[0223] Strategy c: The leading vehicle and the combination strategy of the leading vehicle - trailing vehicle ( Figure 11 ).

[0224] Strategy c combines the above two strategies. The leading vehicle can receive the information of the leading vehicle and the trailing vehicle in the previous queue simultaneously. At this time . Therefore, the communication matrix between queues is expressed as:

[0225] ;

[0226] In Strategy c, the leading vehicle can obtain the state information of the leading vehicle and the trailing vehicle in the previous queue simultaneously, so as to achieve double perception of global and local changes. From Figures 12-15 it can be seen that the communication structure of Strategy c improves the control accuracy of the queue, reduces the impact caused by information lag, and enables the system to respond to the dynamic changes of the previous queue in a timely manner. The results show that Strategy c performs well in terms of speed and position control. The speed curves of each queue are highly consistent, the position relationship is stable, and the changes in vehicle and queue spacing are small.

[0227] Generally speaking, Strategy c achieves an excellent balance between response speed and stability and is suitable for high-density and dynamically complex environments. Compared with Strategy a and b, Strategy c achieves a better balance in queue consistency and dynamic response, but also increases the communication complexity. Therefore, this strategy is particularly suitable for multi-queue cooperation scenarios with high consistency and fast response requirements.

[0228] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structural transformation made under the concept of the present invention by using the content of the specification and drawings of the present invention, or directly / indirectly applied in other related technical fields, is included in the patent protection scope of the present invention.

Claims

1. A multi-vehicle queue cooperative control method based on distributed model predictive control, characterized in that The method includes the following steps: Construct a multi-vehicle queue system and a dynamic model. The dynamic model includes a system state vector, the motion state equations of individual vehicles, and the queue length state equation. The multi-vehicle queue system includes a multi-queue system composed of vehicle queues. Each vehicle queue k contains vehicles. The vehicle set of queue k is: , and the set of all vehicles in the system is , which can be expressed as: ; Define the state vector of the system as , where , , and respectively represent the position, speed, acceleration, and driving force or braking torque of the i-th vehicle in the k-th queue at time t; the motion state equation of the k -th queue for the i -th vehicle includes: ； ， ； Among them, , respectively represent the weight matrix of the state and control input of the k th vehicle in the i th queue, , , , respectively represent the state, vehicle mass, control input at time t, inertial lag of longitudinal dynamics, k th vehicle in the i th queue at time t, tire radius, and is the mechanical efficiency of the driveline; ； Among them, represents the distance between the leading vehicle and the last vehicle in the queue, , respectively represent the positions of the leading vehicle and the last vehicle in the k th queue; Establish a unidirectional communication topology structure, which includes an intra-queue communication topology and an inter-queue communication topology. The intra-queue communication topology includes a directed spanning tree with the leading vehicle as the root node, and the inter-queue communication topology includes a directed spanning tree with the leading vehicle of the system's first queue as the root node; Based on the communication topology structure, define the coupling constraints of vehicles and queues, construct a distributed model predictive control optimization problem and a cost function. The distributed model predictive control optimization problem includes a local optimization problem of vehicle nodes, a global optimization problem of queues, and a multi-objective optimization problem that integrates the local optimization problem of vehicle nodes and the global optimization problem of queues. The local optimization problem of vehicle nodes is used to solve the local optimal control input for each vehicle node, and the global optimization problem of queues is used to solve the coordinated control input between queues for each queue node by using the multi-objective optimization problem; Solve the optimization control problem for each vehicle node, and adjust the control input and motion state of each vehicle according to the optimization results to achieve cooperative control within and between queues.

2. The multi-vehicle queue cooperative control method based on distributed model predictive control according to claim 1, wherein, The step of constructing a multi-vehicle queue system and a dynamic model, where the dynamic model includes a system state vector, a motion state equation of a single vehicle, and a queue length state equation includes: Define the vehicle state and the update rules of vehicle state variables. The vehicle state variables include position, speed, acceleration, and torque, and use a discrete-time model to construct the dynamic equation of a single vehicle; Define the queue length as the distance between the leading vehicle and the trailing vehicle, and establish a dynamic change equation of the queue length.

3. The multi-vehicle queue cooperative control method based on distributed model predictive control according to claim 1, characterized in that, The local optimization problem of vehicle nodes includes a vehicle local cost function and vehicle constraint conditions. The vehicle constraint conditions include vehicle prediction model constraints, the constraint that the initial prediction state is equal to the actual state at time t, vehicle state constraints, vehicle control input constraints, vehicle position constraints, and constraints of recursive feasibility and stability.

4. The multi-vehicle queue cooperative control method based on distributed model predictive control according to claim 3, characterized in that The vehicle local cost function includes: ； Wherein: and respectively represent the predicted state and the predicted control input sequence of vehicle node i at time t, represents the hypothesized state of vehicle node i at time t; represents the desired state of vehicle node i at time t; represents the hypothesized state of the leading vehicle of vehicle node i at time t, represents the distance between vehicle node i and the leading vehicle; The weight matrix is set as a symmetric non - negative definite matrix, and respectively represents the control state error, the control input, the hypothesized state error, and the leading vehicle distance error.

5. The multi-vehicle queue cooperative control method based on distributed model predictive control according to claim 1, characterized in that The global optimization problem of queues includes a queue global cost function and queue constraint conditions. The queue constraint conditions include communication topology constraints, queue position constraints, and queue length change constraints. The communication topology constraints are used to restrict the state update of the leading vehicle to depend on the state information of other leading vehicles defined by the communication topology, ensuring that the behavior of the leading vehicle is restricted by the communication graph structure and can only obtain information from neighbors in the communication topology. The queue position constraints are set for the position trajectories of the leading vehicle and the trailing vehicle of the queue to ensure that the spatial constraints between queues are satisfied.

6. The multi-vehicle queue cooperative control method based on distributed model predictive control according to claim 5, characterized in that, The queue global cost function includes: ； wherein, represents the predicted state sequence of the leading vehicle; represents the predicted control input sequence of the leading vehicle, represents the assumed state of the leading vehicle in the k-th queue, represents the assumed state of the leading vehicle in the (k - 1)-th queue, represents the assumed state of the trailing vehicle in the (k - 1)-th queue, represents the length of the (k - 1)-th queue; represents the desired state of vehicle node i at time t, represents the desired state of the leading vehicle in the k-th queue at time t; represents the distance between the leading vehicle in queue k and the trailing vehicle of the previous queue; is set as a symmetric non - negative definite matrix, which respectively controls the state error, control input, assumed state error and leading vehicle distance error to ensure the stability and coordination of vehicles in the queue.

7. The multi-vehicle queue cooperative control method based on distributed model predictive control according to claim 1, characterized in that The step of adjusting the motion state of each vehicle according to the optimization results, where the motion state includes position, speed, and acceleration, to achieve cooperative control within and between queues includes: Initialize the vehicle state, input state, and queue length; For each queue, generate the desired trajectory of the leading vehicle within the prediction time domain; At each moment, solve the optimization control problem for each vehicle node, and obtain the optimal control input sequence according to the optimization results to achieve cooperative control within and between queues.

8. The multi-vehicle queue cooperative control method based on distributed model predictive control according to claim 1, wherein, The steps of solving the optimization control problem for each vehicle node at each moment and obtaining the optimal control input sequence to achieve the coordinated control within and between the queues include: The queue status update step, node status update step, and information exchange and synchronization step are continuously cycled at each moment, gradually updating the control input and status of all queues and vehicle nodes to achieve coordinated control within and between queues; The queue state updating step includes: updating the state sequence of the pilot vehicle to ensure that the desired trajectory is followed; updating the state of each vehicle according to the optimal control input sequence within the prediction time domain, and calculating the next step hypothetical input sequence of the pilot vehicle and updating the hypothetical state sequence; The node state updating step includes: updating the current node state using the optimal control input, calculating the optimal state sequence and updating the node state in the prediction time domain, updating the assumed input sequence of the vehicle, and predicting the assumed state sequence based on the assumed input sequence; The information exchange and synchronization steps include: sending the assumed state sequence and assumed input sequence of the queue leader vehicle node to the adjacent leader vehicle node, and sending the node assumed state sequence and assumed input sequence to the adjacent follower vehicle node to achieve information sharing.

9. The multi-vehicle queue cooperative control method based on distributed model predictive control according to claim 1, characterized in that After the step of establishing a unidirectional communication topology structure, wherein the communication topology structure includes an internal queue communication topology and an inter-queue communication topology, wherein the internal queue communication topology includes a directed spanning tree with the pilot vehicle as the root node, and the inter-queue communication topology includes a directed spanning tree with the pilot vehicle of the first queue of the system as the root node, the following steps are included: Establish multiple communication strategies, including a leader-to-leader strategy, a leader-to-last-car strategy, and a leader-to-last-car strategy; the leader-to-leader strategy includes a communication method in which information is transmitted step by step through the leader car of the queue, the leader-to-last-car strategy includes a communication method in which the leader car directly receives information from the last car of the previous queue, and the leader-to-last-car strategy includes a communication method in which the leader car is allowed to simultaneously receive information from the leader car and the last car of the previous queue; Select different communication strategies based on queue length, response speed, and dynamic environment.

10. A multi-vehicle queue cooperative control medium based on distributed model predictive control, characterized in that, Executable instructions are stored, and the executable instructions include program codes for executing the multi-vehicle platoon cooperative control method based on distributed model predictive control according to any one of claims 1 to 9 when executed.

Images