A multi-vehicle cooperative control method and system considering time-varying directed communication topology

Abstract

The application discloses a kind of multi-vehicle cooperative control method and system considering time-varying directed communication topology, belong to vehicle control technical field.The method first models unreliable communication topology structure, in order to describe the randomness of communication topology, uses continuous-time Markov chain to represent unreliable communication topology structure;Then, respectively to the following vehicle and the longitudinal dynamics model of pilot vehicle is established;Next, a kind of fully distributed adaptive observer is designed to obtain the state estimation information of pilot vehicle;Finally, the state of pilot vehicle and the front vehicle state information observed by self-vehicle observer are used to construct an optimal control problem, the constraint condition including string stability constraint is introduced, the optimal control problem is solved using DMPC framework, and the optimal control quantity is obtained, to realize the tracking control of pilot vehicle state.

Description

Technical Field

[0001] This invention belongs to the field of vehicle control technology, specifically relating to a multi-vehicle cooperative control method and system that considers time-varying directed communication topology. Background Technology

[0002] With the acceleration of global urbanization, the rapid increase in the number of vehicles has led to increasingly serious problems such as traffic congestion and frequent traffic accidents, posing a severe challenge to road traffic. In order to solve these problems, improve road capacity, enhance driving comfort and safety, and reduce energy consumption, multi-vehicle platoon cooperative control technology has emerged.

[0003] The core of multi-vehicle platoon cooperative control technology lies in achieving coordinated operation of vehicles within a platoon through communication and collaboration. This technology can improve traffic flow stability, reduce the number of stops, and shorten the safe distance between vehicles, thereby improving overall traffic efficiency. Furthermore, by optimizing platoon operation, it can effectively reduce vehicle fuel consumption and environmental pollution.

[0004] In the implementation of multi-vehicle platoon cooperative control technology, time-varying communication topology is a key factor. As the relative positions and speeds of vehicles constantly change during operation, communication conditions and quality also fluctuate over time, leading to changes in the communication links between vehicles and resulting in a time-varying communication topology. This is an unreliable communication condition, and the switching of the topology poses challenges to information exchange between vehicles and the implementation of cooperative control strategies. Therefore, researching how to achieve effective platoon cooperative control under unreliable communication conditions has become an important research direction in this field.

[0005] Distributed Model Predictive Control (DMPC), as an advanced control strategy, is playing an increasingly important role in this field. Model Predictive Control (MPC) is a model-based control strategy that predicts the future behavior of a system by optimizing the control input and can explicitly consider the constraints of the control input and the system. DMPC combines the optimization capabilities of MPC with the flexibility of distributed control, making it an ideal choice for handling complex dynamic systems, and therefore suitable for studying multi-vehicle platooning systems. In platooning control, the state error of any vehicle in the platoon can affect other vehicles and may even be amplified downstream of the traffic flow. Therefore, a crucial issue in platooning control is ensuring the chordal stability of the platoon, meaning that disturbances upstream of the traffic flow can be mitigated downstream. DMPC's advantages make it suitable for solving vehicle platooning control problems, but existing research using DMPC to study the chordal stability of vehicle platoons remains limited.

[0006] Existing technology 1:

[0007] Technical approach one considers the case of time-varying undirected communication topologies and proposes a network-physical level security control framework for connected automated vehicle (CAV) queuings. First, at the network level, a continuous-time Markov chain is used to simulate the unreliable communication topology, where the topology is time-varying and global information is difficult to obtain. Then, a fully distributed adaptive observer is designed for each following vehicle to obtain the state information of the lead vehicle. Finally, at the physical level, an adaptive fault-tolerant controller is designed to track the observer state, ultimately achieving a consistent control objective.

[0008] The disadvantage of the existing technology is that it uses an undirected communication topology, which requires the Laplace matrix of the queue system to be symmetric, that is, the CAVs in the queue must be interconnected in pairs. In practice, in order to ensure information security or in the case of limited communication, only unidirectional information transmission between vehicles in the queue is allowed. At this time, the designed observer will not be able to meet the observation requirements. Furthermore, the influence of string stability factors is not considered in the controller. The vehicle spacing error will increase with the size of the vehicle queue, which will have a great impact on the stability of the queue.

[0009] Existing technology 2:

[0010] A hierarchical cooperative control scheme for heterogeneous connected vehicles is proposed. It uses a hierarchical control framework that includes an upper-level navigator state observer and a lower-level state tracking controller. The closed-loop stability is considered under conditions of no communication delay and with communication delay. A finite-time convergent observer is proposed to handle the string stability under general communication topologies.

[0011] The drawbacks of Existing Technology Two: Technology Two assumes that the communication links between vehicles in the queue are valid. However, in practical applications, communication between vehicles may be subject to interference, with the possibility of communication link failure or packet loss. Furthermore, the queue communication topology may switch over time. Additionally, in this scheme, adjusting the control gain of the lower-level controller requires experience, and the robustness of the tracking control algorithm is difficult to guarantee. Moreover, in practical applications, both control inputs and states are limited to certain ranges, and the lower-level controller in this scheme cannot handle these constraints.

[0012] Existing technology three:

[0013] The third technical solution addresses the communication delay problem in vehicle platooning. Under the lead vehicle-following communication topology, it applies a constant spacing strategy and proposes an improved DMPC method to enhance the control robustness of vehicle platooning under communication delay. Furthermore, by introducing robust constraints and chord stability constraints into the optimization problem, the reliability of platoon control is guaranteed.

[0014] The drawback of existing technology three is that it requires a specific communication topology to achieve platooning coordination control. In this structure, each following vehicle only communicates with the preceding and lead vehicles. While this simplifies the communication mode, it limits the system's flexibility and scalability. In practical applications, due to unreliable communication and limitations in communication distance, it may be impossible to maintain communication with the lead vehicle, which can significantly impact the overall platooning control.

[0015] Existing technology four:

[0016] The fourth technical solution constructs a queue system model that considers the Bernoulli packet loss model and designs a distributed controller based on a directed spanning tree communication topology, suitable for a lead vehicle as the root node.

[0017] The disadvantage of the existing technology 4: The solution of technology 4 uses the Bernoulli packet loss model to simulate the packet loss phenomenon in unreliable communication. This is a simplified random packet loss model. It assumes that the packet loss event of each data packet is independent, that is, whether the previous data packet is lost does not affect the packet loss probability of the next data packet. However, in actual applications, communication packet loss has the characteristic of continuity, and the packet loss rate will change over time. The Bernoulli packet loss model cannot simulate the unreliable communication topology well. Summary of the Invention

[0018] The purpose of this invention is to provide a multi-vehicle cooperative control method considering time-varying directed communication topology. It constructs a vehicle queuing system by establishing an unreliable communication topology model and a vehicle longitudinal dynamics model, designs a fully distributed adaptive observer and a DMPC controller, and ensures the chordal stability of the queuing by imposing constraints on the optimization problem, so as to realize vehicle queuing cooperative control under time-varying directed communication topology, thereby solving at least one of the technical problems involved in the background art.

[0019] To solve the above-mentioned technical problems, the present invention is implemented as follows:

[0020] This invention provides a multi-vehicle cooperative control system considering a time-varying directed communication topology, including a communication module, an observation module, a computational optimization module, and an actuator module, wherein:

[0021] The communication module is used to establish an unreliable communication topology model based on a continuous-time Markov process to describe the communication topology of random switching based on vehicle-to-vehicle communication technology, and to receive the state estimation of the neighboring vehicle to the lead vehicle and the state information of the preceding vehicle and send them to the observation module.

[0022] The observation module is used to establish a longitudinal dynamic model of the lead vehicle to characterize the dynamic characteristics of the lead vehicle, and to design an adaptive observer to observe the state information of the lead vehicle based on the dynamic model of the lead vehicle and the unreliable communication topology model, generate the self vehicle's estimate of the lead vehicle's state, and send it together with the state of the preceding vehicle to the calculation and optimization module.

[0023] The computational optimization module is used to establish a longitudinal dynamics model of the following vehicle to characterize the dynamics of the following vehicle, and to design a DMPC tracking controller based on the longitudinal dynamics model of the following vehicle, the estimation of the ego vehicle's state of the lead vehicle, the state of the lead vehicle, and the ego vehicle's state, and to output the optimal expected acceleration obtained from solving the optimization problem to the actuator module.

[0024] The actuator module is used to convert the optimal desired acceleration into a direct control signal for the actuator, while simultaneously monitoring the vehicle's status and outputting it to the calculation and optimization module.

[0025] Optionally, the randomly switched communication topology is derived from the randomly switched diagram. It means, and The communication topology representing the vehicle queue will randomly switch between l different graphs if and only if σ(t)=p∈{1,2,…,l}. Through the transition rate matrix Define an infinitesimal generator of σ(t) such that for any positive scalar Δt, as Δt→0, it has the following definition:

[0026]

[0027] In the formula, μ pq μ is the transition rate from state p to state q. If p ≠ q, then μ pq ≥0, if p=q then μ pp =-∑ p≠q μ pq o(Δt) represents an infinitesimal quantity of Δt, satisfying

[0028] Optionally, the longitudinal dynamics model of the following vehicle is expressed by the following equation:

[0029]

[0030] In the formula, p i (t), v i (t) and a i (t) represent the position, velocity, and acceleration of the following vehicle i, respectively; η T,i and r w,i T represents mechanical efficiency and tire radius, respectively; i (t) and T des,i (t) represents the actual and desired driving or braking torque of the following vehicle i, respectively; m i and g represent the mass of the following vehicle and the gravitational constant, respectively; C A,i and f r,i These refer to the lumped air drag coefficient and the rolling drag coefficient, respectively; τ i This represents the inertial time delay following the longitudinal dynamics of the vehicle, and satisfies...

[0031] Linearization is performed using feedback linearization techniques, with the desired driving or braking torque being:

[0032]

[0033] In the formula, For the new control input; thus, a third-order linear model for following the vehicle is established:

[0034]

[0035] Right now:

[0036]

[0037] In the formula, A i For the system matrix, B iGiven an input matrix, and:

[0038]

[0039] Optionally, the longitudinal dynamics model of the lead vehicle is expressed by the following equation:

[0040]

[0041] In the formula,

[0042] Optionally, the adaptive observer is represented by the following equation:

[0043]

[0044] In the formula, For observation error, define The observations of the observer, Estimate the position of the lead vehicle;

[0045] Estimate the speed of the lead vehicle; Estimate the acceleration of the lead vehicle; x0(t) represents the state of the lead vehicle; φ i (t) is the relative observation error, defined as:

[0046]

[0047] make Where ρ i (0)≥1, g i (ρ i ) is ρ i A nonlinear increasing nonnegative function.

[0048] The present invention also provides a multi-vehicle cooperative control method considering time-varying directed communication topology based on the aforementioned multi-vehicle cooperative control system, comprising the following steps:

[0049] Step 1: Perform initialization at t=0, setting the initial states x0(0) and x for all vehicles. i (0), i∈{1,2,…,N} and the initial value of the observer.

[0050] Step 2: Initialize the hypothetical trajectory of the following vehicle and set the hypothetical estimated state trajectory. Assuming control input trajectory Assuming a state trajectory Generated by the following formula:

[0051]

[0052] Step 3: At time t>0, broadcast the estimated state of the voluntary vehicle to the navigator vehicle. Assume state trajectories for neighboring vehicles and your own vehicle. Give it to the car behind;

[0053] Step 4: Receive the estimated state of the lead vehicle from neighboring vehicles. and the assumed state trajectory of the vehicle in front

[0054]

[0055] Step 5: Generate the estimated state trajectory of the autonomous vehicle relative to the navigator vehicle.

[0056]

[0057] Step Six: Solve the optimization problem Generate the optimal control input trajectory sequence and optimal state trajectory

[0058] Step 7: Extract the optimal control input trajectory sequence The first element Apply to the control actuator module;

[0059] Step 8: Update the hypothesis control input trajectory and the assumed state trajectory of the vehicle Updated by the following formula:

[0060]

[0061] Step 9: Update the vehicle observer state according to the update law designed for the adaptive observer.

[0062] Step 10: Determine whether the formation driving task has been completed. If it has been completed, end the algorithm process; otherwise, return to Step 3.

[0063] Compared with the prior art, the advantages of this invention are as follows:

[0064] 1. Compared with the prior art, the adaptive observer designed in this invention can be applied to a one-way random switching communication topology with the lead vehicle as the root node, further relaxing the restrictions on the communication topology conditions. It can accurately estimate the state of the lead vehicle without the need for information from the following vehicles, and can achieve chordal stability of the vehicle queue through constraints.

[0065] 2. Compared with the prior art 2, the adaptive observer designed in this invention has stronger robustness and can be applied to situations with random switching of communication topologies. At the same time, the designed DMPC controller is an optimal control algorithm that can explicitly handle constraints and obtain the optimal control quantity.

[0066] 3. Compared with the existing technology 3, the present invention can be applied to multiple communication topologies, enhancing the flexibility and scalability of the queue system.

[0067] 4. Compared with the prior art, the present invention uses Markov chains to simulate unreliable communication situations, including packet loss, which can adapt to the working conditions where network conditions are constantly changing and packet loss is continuous.

[0068] 5. This invention uses a continuous-time Markov chain to simulate unreliable communication topology, which can adapt to the working conditions where network conditions are constantly changing and packet loss is continuous. Compared with the method of modeling communication packet loss as a Bernoulli model, this method also regards communication packet loss as a communication topology structure, which can be included in a unified research framework and enhances the flexibility and scalability of the queuing system.

[0069] 6. The fully distributed adaptive observer designed in this invention is suitable for time-varying directed communication topologies. Compared with time-varying undirected communication topologies, it reduces the amount of information required, relaxes the restrictions on communication topology conditions, and can accurately estimate the state of the lead vehicle without the need for information from the following vehicle, thus enhancing the robustness of the system under unreliable communication conditions.

[0070] 7. This invention designs a DMPC controller that uses information from the upper-level observer and the preceding vehicle to construct a cost function and constraints. Compared with non-optimal control algorithms, it can explicitly handle constraints and obtain the optimal control quantity, thereby improving the accuracy and efficiency of control. Furthermore, by adding chord stability constraints, it can ensure that the spacing error in the queue system will not be amplified during the propagation process downstream along the queue, thus improving the stability and safety of the queue. Attached Figure Description

[0071] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, wherein:

[0072] Figure 1 A schematic diagram of the framework of a multi-vehicle cooperative control system considering a time-varying directed communication topology provided by the present invention;

[0073] Figure 2This invention provides a possible communication topology diagram for unreliable communication scenarios.

[0074] Figure 3 This is a flowchart illustrating the multi-vehicle cooperative control method considering time-varying directed communication topology provided by the present invention. Detailed Implementation

[0075] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0076] The terms "first," "second," etc., used in the specification and claims of this invention are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0077] Please see Figure 1 As shown, this embodiment of the invention provides a multi-vehicle cooperative control system that considers time-varying directed communication topology, including a communication module, an observation module, a calculation and optimization module, and an actuator module.

[0078] The communication module is used to establish an unreliable communication topology model based on a continuous-time Markov process to describe the communication topology of random switching based on vehicle-to-vehicle communication technology, and to receive the state estimates of the neighboring vehicle to the lead vehicle and the state information of the preceding vehicle and send them to the observation module.

[0079] It should be further explained that the communication topology of a queue system is the information interaction relationship between CAVs (Cars, Vehicles, and Axes) in the queue, which can be represented using graph theory. Graphs can be divided into directed graphs and undirected graphs. In a directed graph, information can only be transmitted unidirectionally between two nodes, while in an undirected graph, nodes can exchange information bidirectionally. The communication topology of a queue system consisting of front-vehicle and rear-vehicle pairs is represented by a directed graph. In other words, in the formula, It is a point set, representing the set of vehicles in the queuing system. It is an edge set, representing unidirectional information transmission between vehicles in the queue system, where N is the total number of vehicles following in the queue system.

[0080] Furthermore, if there exists a containing The tree-like subgraph of all nodes is called a It contains a directed spanning tree. Furthermore, if the directed spanning tree starts with the navigator as its initial node, then it is called... It contains a directed spanning tree rooted at the lead vehicle. Assume the following vehicles i∈{1,2,…N}, and define the adjacency matrix. When car i can receive information from car j, a ij =1, otherwise a ij =0. Assume no self-loops, i.e., a ii =0. Define the traction matrix. When car i can receive information from the lead car, ω i0 =1, otherwise ω i0 =0. Define the Laplace matrix. When i = j When i ≠ j, l ij =-a ij Define the information flow matrix. This invention relies solely on unidirectional information to achieve queue collaborative control, which is an asymmetric communication topology. The Laplace matrix of this type of communication topology is also asymmetric.

[0081] The application of V2V communication can improve road safety of vehicle platooning systems and reduce fuel consumption. However, in practical applications, V2V communication is subject to interference from various factors, such as channel fading and communication congestion, which can make V2V communication unreliable. This unreliability may lead to packet loss and switching of communication topology, posing a potential threat to the safety of vehicle platooning systems. In order to describe the unreliability and randomness of actual communication, a continuous-time Markov model is used to characterize the time-varying communication topology. Considering the simplicity and uniformity of modeling, this invention also models communication packet loss as a type of communication topology with partial link failure.

[0082] The communication topology between CAVs is represented by a randomly switching diagram. express, The communication topology representing the vehicle queue will randomly switch between l different graphs if and only if σ(t)=p∈{1,2,…,l}. Here It is directional. This is achieved through the transition matrix. Define an infinitesimal generator of σ(σ) such that for any positive scalar Δt, as Δt→0, it has the following definition:

[0083]

[0084] In the formula, η pq η is the transition rate from state p to state q. If p ≠ q, then η pq ≥0; if p = q then μ pp =-∑ p≠q μ pq o(Δt) represents an infinitesimal quantity of Δt, satisfying

[0085] This unreliable communication topology model posits that the switching process of the communication topology is governed by a continuous-time Markov process {σ(t), t≥0}, where the transition rate matrix μ is ergodic. This means that regardless of the initial state, the process will eventually visit all its possible states with equal probability, and its statistical characteristics are stable over long periods.

[0086] For a continuous-time Markov process, there exists a probability vector π = [π1, π2, ..., π]. l ] T ,satisfy And π q ≥0, n=1,2,…,l. This describes the probability of the process reaching various states after running for a long time. This probability vector is fixed and unchanging, i.e., a steady-state distribution. No matter what state the process starts from, it will eventually tend to this probability distribution.

[0087] In current research on vehicle platoon control, the most commonly used unidirectional communication topologies are precursor following (PF) and precursor-leader following (PLF). However, considering unreliable communication scenarios, some links may fail, resulting in packet loss. Figure 2 The diagram shown is a schematic of the time-varying communication topology under unreliable communication as considered in this invention. The switching of these four communication topology diagrams is modeled as a continuous-time Markov process, and each diagram is a state of this process.

[0088] The observation module is used to establish a longitudinal dynamic model of the lead vehicle to characterize the dynamic characteristics of the lead vehicle, and to design an adaptive observer to observe the state information of the lead vehicle based on the dynamic model of the lead vehicle and the unreliable communication topology model, generate the self vehicle's estimate of the lead vehicle's state, and send it together with the state of the preceding vehicle to the calculation and optimization module.

[0089] The computational optimization module is used to establish a longitudinal dynamics model of the following vehicle to characterize the dynamics of the following vehicle, and to design a DMPC tracking controller based on the longitudinal dynamics model of the following vehicle, the estimation of the ego vehicle's state of the lead vehicle, the state of the lead vehicle, and the state of the ego vehicle, and to output the optimal expected acceleration obtained from solving the optimization problem to the actuator module.

[0090] It should be further explained that, for the longitudinal dynamics modeling of the lead vehicle and following vehicles, this invention considers a vehicle platoon composed of CAVs, consisting of one lead vehicle and N following vehicles, labeled 0, 1, ..., N. All CAVs are coupled into a platoon system through interconnected sensors and a V2V communication network. Each CAV adjusts its own speed and acceleration in real time using vehicle status information broadcast by neighboring CAVs and estimated information about the lead vehicle. The control objective of the vehicle platoon system is to adjust the vehicle spacing based on a certain spacing strategy and make the speed and acceleration error with the lead vehicle zero.

[0091] This invention considers the longitudinal movement of vehicle platoons with heterogeneous parameters in a highway scenario, and makes the following assumptions and simplifications to the vehicle dynamics model:

[0092] 1) Only consider the longitudinal vehicle movement. The vehicle travels in a straight direction and ignores the movement in other directions. That is, there is no vertical bumping and no lateral deviation.

[0093] 2) The vehicle is considered as a rigid body with bilateral symmetry, and the load transfer effect of the suspension system is ignored;

[0094] 3) Assuming the vehicle is driven on a flat, good road, the tires will not slip longitudinally;

[0095] 4) Assume that the input-output characteristics of the vehicle's power system can be described by a first-order inertial element.

[0096] When the assumptions are met, the simplified longitudinal dynamics model of the vehicle is analyzed. According to Newton's second law, the longitudinal dynamics model of the following vehicle i can be obtained:

[0097]

[0098] In the formula, p i (t), v i (t) and a i (t) represent the position, velocity, and acceleration of the following vehicle i, respectively; η T,i and r w,i T represents mechanical efficiency and tire radius, respectively; i (t) and T des,i (t) represents the actual and expected driving or braking torque following vehicle i; m iand g represent the mass of the following vehicle and the gravitational constant, respectively; C A,i and f r,i These refer to the lumped air drag coefficient and the rolling drag coefficient, respectively; τ i This represents the inertial time delay following the longitudinal dynamics of the vehicle, satisfying...

[0099] To facilitate the analysis of the control system, the commonly used feedback linearization technique is employed to linearize the nonlinear equations, and the desired driving or braking torque becomes:

[0100]

[0101] In the formula, This becomes the new control input. Therefore, the nonlinear model is transformed into a third-order linear model:

[0102]

[0103] The expression can be rewritten as:

[0104]

[0105] In the formula, A i For the system matrix, B i Given an input matrix, and:

[0106]

[0107] The dynamic model of the virtual pilot vehicle is represented as follows: in The status of the virtual navigator vehicle can only be received by a subset of the following vehicles via V2V communication.

[0108] Based on the established communication and dynamic models, the objective of queue control can be expressed as follows:

[0109]

[0110] In the formula, d i0 d represents the distance between the following car i and the lead car. i0 = i·d0, where d0 is the desired distance between the preceding and following vehicles. Let ||·|| denote the mathematical expectation, and let ||·|| denote the Euclidean norm.

[0111] The adaptive observer includes an upper-layer fully distributed adaptive observer and a lower-layer distributed model prediction controller.

[0112] For the aforementioned upper-layer fully distributed adaptive observer, considering that following vehicles downstream in the vehicle queue cannot directly obtain information from the lead vehicle, but can only estimate it through the observation information of neighboring vehicles, a fully distributed adaptive observer is designed at the upper network layer to directly obtain the state of the lead vehicle or indirectly estimate it from the observation information of neighboring vehicles in the network. The upper-layer fully distributed observer is designed in the following form:

[0113]

[0114] in, For observation error, define The observations of the observer, Estimate the position of the lead vehicle. Estimate the speed of the lead car. Estimate the acceleration of the lead car, x0(t) represents the state of the lead car, and φ i (t) is the relative observation error, defined as:

[0115]

[0116]

[0117] make Where ρ i (0)≥1, g i (ρ i ) is ρ i A nonlinear increasing nonnegative function.

[0118] The update law of this observer can be used to make the observed state converge to the state of the lead vehicle in a mean square manner: Right now:

[0119]

[0120] For the lower-level distributed model predictive controller (DMPC), the core idea is to decompose the entire system into multiple subsystems, each with its own MPC controller. These subsystems are then connected via a communication protocol to achieve overall system control. In a queuing system, each following vehicle is a subsystem. MPC is an optimal control method, which typically requires finding a control law or control trajectory sequence that optimizes the objective function within the permissible control set. The core idea of ​​MPC is to construct an optimization problem based on the established system model, the system's control objective, and the system's constraints. At each time step, an optimization problem in the prediction time domain is solved to obtain the optimal control trajectory sequence. The first term of the control trajectory sequence is then applied to the current system. In the next time step, the optimization process iterates forward, continuously generating new control trajectory sequences and state trajectories.

[0121] Define sampling time The observations of the lead vehicle obtained by the upper-level observer are discretized and sent to the lower-level controller. The design optimization problem of the DMPC controller for discrete systems will be carried out by discretizing the vehicle dynamics model:

[0122]

[0123] In the formula, For the discretized system matrix, The discretized input matrices are defined as follows:

[0124]

[0125] In the prediction time domain N p Within, three types of trajectories are defined:

[0126] 1) and Predicting the state trajectory and predicting the control input trajectory are the quantities to be optimized in the optimization problem.

[0127] 2) and The optimal state trajectory and the optimal control input trajectory are the optimal solutions obtained by solving the local optimization problem.

[0128] 3) and The assumed state trajectory and the assumed control input trajectory are trajectories generated from the optimal trajectory.

[0129] The goal of the lower-level tracking controller is to follow the vehicle state so that it converges to the lead vehicle state estimated by the upper-level observer. Right now:

[0130]

[0131] Therefore, the open-loop optimization problem is defined.

[0132]

[0133] Constraints:

[0134]

[0135] Here, the cost function is defined as:

[0136]

[0137] and They are respectively by and The generated estimated states of the unicycle and the preceding vehicle relative to the lead vehicle, including the initial state. Depend on renew.

[0138] When the state of each following vehicle's fully distributed adaptive observer becomes consistent with that of the lead vehicle... have:

[0139]

[0140] In the formula, R i F i S i G i All are positive definite diagonal matrices, representing the weights of the corresponding terms; This indicates a penalty for controlling the input; This represents the penalty for deviation from the assumed trajectory of the vehicle at the previous moment; This indicates a penalty for deviation from the trajectory of the vehicle preceding it; This represents the penalty for deviations in the ego vehicle's estimation of the navigator's state.

[0141] Meanwhile, regarding the constraints, These are the initial state constraints. To follow the constraints of the vehicle dynamics model, To control the input trajectory constraints, where u lb For the minimum control quantity, u ub To be the maximum control quantity, For terminal constraints, let The string stability constraint can be written as:

[0142]

[0143] In the formula, β i ∈(0,1], γ=[1,0,0];||·|| ∞ Indicate l ∞ Norm; The purpose of string stability constraints is to ensure that spacing errors in a queue system are not amplified as they propagate downstream along the queue.

[0144] The actuator module is used to convert the optimal desired acceleration into a direct control signal for the actuator, while simultaneously monitoring the vehicle's status and outputting it to the calculation and optimization module.

[0145] Combined Figure 3 As shown, the present invention also provides a multi-vehicle cooperative control method considering time-varying directed communication topology based on the aforementioned multi-vehicle cooperative control system, comprising the following steps:

[0146] Step 1: Perform initialization at t=0, setting the initial states x0(0) and x for all vehicles. i (0), i∈{1,2,…,N} and the initial value of the observer.

[0147] Step 2: Initialize the hypothetical trajectory of the following vehicle and set the hypothetical estimated state trajectory. Assuming control input trajectory Assuming a state trajectory Generated by the following formula:

[0148]

[0149] Step 3: At time t>0, broadcast the estimated state of the voluntary vehicle to the navigator vehicle. Assume state trajectories for neighboring vehicles and your own vehicle. Give it to the car behind;

[0150] Step 4: Receive the estimated state of the lead vehicle from neighboring vehicles. and the assumed state trajectory of the vehicle in front

[0151]

[0152] Step 5: Generate the estimated state trajectory of the autonomous vehicle relative to the navigator vehicle.

[0153]

[0154] Step Six: Solve the optimization problem Generate the optimal control input trajectory sequence and optimal state trajectory

[0155] Step 7: Extract the optimal control input trajectory sequence The first element Apply to the control actuator module;

[0156] Step 8: Update the hypothesis control input trajectory and the assumed state trajectory of the vehicle Updated by the following formula:

[0157]

[0158] Step 9: Update the vehicle observer state according to the update law designed for the adaptive observer.

[0159]

[0160] Step 10: Determine whether the formation driving task has been completed. If it has been completed, end the algorithm process; otherwise, return to Step 3.

[0161] The multi-vehicle cooperative control method considering time-varying directed communication topology provided by this invention can achieve the vehicle queuing control objective, and the states of all following vehicles can converge to the desired state in a mean-square manner.

[0162]

[0163] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0164] Furthermore, it should be noted that the scope of the methods and systems in the embodiments of the present invention is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. In addition, features described with reference to certain examples may be combined in other examples.

[0165] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of the present invention.

Claims

1. A multi-vehicle cooperative control system considering time-varying directed communication topology, characterized in that, It includes a communication module, an observation module, a computational optimization module, and an actuator module, among which: The communication module is used to establish an unreliable communication topology model based on a continuous-time Markov process to describe the communication topology of random switching based on vehicle-to-vehicle communication technology, and to receive the state estimation of the neighboring vehicle to the lead vehicle and the state information of the preceding vehicle and send them to the observation module. The observation module is used to establish a longitudinal dynamic model of the lead vehicle to characterize its dynamic features, and to design an adaptive observer based on the lead vehicle's dynamic model and unreliable communication topology model to observe the lead vehicle's state information. This adaptive observer generates an estimate of the lead vehicle's state from the vehicular vehicle's state, which, along with the state of the preceding vehicle, is sent to the computation and optimization module. The adaptive observer is expressed by the following formula: In the formula, For observation error, define , The observations of the observer, , Estimate the position of the lead vehicle; Estimate the speed of the lead vehicle; Estimate the acceleration of the lead vehicle; ; In navigator mode; The relative observation error is defined as: make ,in , for A nonlinear, increasing, nonnegative function; The computational optimization module is used to establish a longitudinal dynamics model of the following vehicle to characterize the dynamics of the following vehicle, and to design a DMPC tracking controller based on the longitudinal dynamics model of the following vehicle, the estimation of the ego vehicle's state of the lead vehicle, the state of the lead vehicle, and the ego vehicle's state, and to output the optimal expected acceleration obtained from solving the optimization problem to the actuator module. The actuator module is used to convert the optimal desired acceleration into a direct control signal for the actuator, while simultaneously monitoring the vehicle's status and outputting it to the calculation and optimization module.

2. The multi-vehicle cooperative control system considering time-varying directed communication topology according to claim 1, characterized in that, The communication topology with random switching is represented by the random switching diagram. express, This indicates that the communication topology of the vehicle queue will be randomized. Switching between different graphs if and only if hour, ; through the transition matrix definition An infinitesimal generator for any positive scalar ,when When, it is defined as follows: In the formula, From state Switch to state The transfer rate, if but ,like but , express an infinitesimal quantity that satisfies .

3. The multi-vehicle cooperative control system considering time-varying directed communication topology according to claim 2, characterized in that, The longitudinal dynamics model of the following vehicle is expressed by the following equation: In the formula, , and These respectively represent the following vehicles i Position, velocity, and acceleration; and These represent mechanical efficiency and tire radius, respectively. and These respectively represent the following vehicles i The actual and expected driving or braking torque; and These represent the mass of the following vehicle and the gravitational constant, respectively. and These refer to the lumped air drag coefficient and the rolling drag coefficient, respectively. This represents the inertial time delay following the longitudinal dynamics of the vehicle, and satisfies... ; Linearization is performed using feedback linearization techniques, with the desired driving or braking torque being: In the formula, For the new control input; thus, a third-order linear model for following the vehicle is established: Right now: In the formula, , For the system matrix, Given an input matrix, and:

4. The multi-vehicle cooperative control system considering time-varying directed communication topology according to claim 3, characterized in that, The longitudinal dynamics model of the pilot vehicle is expressed by the following equation: ； In the formula, 5. A multi-vehicle cooperative control method considering time-varying directed communication topology based on any one of claims 1-4, characterized in that, Includes the following steps: Step 1: In Initialization is performed continuously to set the initial state of all vehicles. and and observer initial values ; Step 2: Initialize the hypothetical trajectory of the following vehicle and set the hypothetical estimated state trajectory. Assuming the control input trajectory Assuming the state trajectory Generated by the following formula: Step 3: In At any given moment, the broadcast system estimates the status of the lead vehicle. Assume state trajectories for neighboring vehicles and your own vehicle. Give it to the car behind; Step 4: Receive the estimated state of the lead vehicle from neighboring vehicles. and the assumed state trajectory of the vehicle in front ; Step 5: Generate the estimated state trajectory of the autonomous vehicle relative to the navigator vehicle. : Step Six: Solve the optimization problem Generate the optimal control input trajectory sequence and optimal state trajectory ; Step 7: Extract the optimal control input trajectory sequence The first element It is applied to the control actuator module; Step 8: Update the hypothesis control input trajectory and the assumed state trajectory of the vehicle Updated by the following formula: Step 9: Update the vehicle observer state according to the update law designed for the adaptive observer. ; Step 10: Determine whether the formation driving task has been completed. If it has been completed, end the algorithm process; otherwise, return to Step 3.

Images