Pipeline-based robust data-driven control method and device for mixed vehicle platoons

Abstract

The present application relates to a pipeline-based hybrid vehicle fleet robust data-driven control method and device. It includes: obtaining a target input data sequence and a target output data sequence based on the hybrid vehicle fleet to be controlled; creating an over-approximation system matrix set based on the target input and output data sequences, and determining a data-driven minimum robust positive invariant set based on the over-approximation system matrix set; determining a pipeline-based robust data-driven predictive control strategy based on the minimum robust positive invariant set, and then calculating the optimal nominal control input and error feedback control input of the intelligent connected vehicle in combination with a preset rolling horizon control framework; obtaining a final control input based on the optimal nominal control input and error feedback control input, so that the controller controls the hybrid vehicle fleet to be controlled based on the final control input. Thus, the problems of low computational efficiency and insufficient robustness of existing control strategies are solved, safe control of a hybrid vehicle fleet is achieved, and driving safety is improved.

Description

Technical Field

[0001] The present application relates to the field of vehicle control technology, and in particular to a pipeline-based hybrid vehicle platoon robust data-driven control method and device. Background Art

[0002] With the rapid development of intelligent and connected vehicles (ICVs) technology, future intelligent transportation systems are expected to operate in mixed traffic environments, where ICVs and human-driven vehicles (HDVs) coexist, forming mixed fleets. These mixed fleets have significant potential to improve traffic stability and efficiency, spurring the development of advanced control strategies for mixed fleets.

[0003] In related technologies, strategies such as distributed data-driven model predictive control and data-supported predictive control have been proposed, and control verification of mixed fleets has been carried out in various scenarios such as simulation research, hardware-in-the-loop experiments, and driver-in-the-loop experiments.

[0004] However, most of these methods are verified under idealized conditions, which limits their application in the real world, leading to problems such as low efficiency and insufficient robustness, which need to be urgently addressed. Summary of the Invention

[0005] The present application provides a pipeline-based robust data-driven control method and device for a mixed vehicle platoon, to address the problems of low computational efficiency and insufficient robustness of existing control strategies, thereby achieving safe control of mixed vehicle fleets and improving driving safety.

[0006] To achieve the above objectives, a first embodiment of the present application proposes a pipeline-based hybrid vehicle platoon robust data-driven control method, comprising the following steps:

[0007] obtaining a target input data sequence and a target output data sequence based on the mixed vehicle queue to be controlled;

[0008] Creating an over-approximation system matrix set based on the target input data sequence and the target output data sequence, and determining a data-driven minimum robust positive invariant set according to the over-approximation system matrix set;

[0009] Determining a pipeline-based robust data-driven predictive control strategy that integrates safety and performance constraints based on the minimum robust positive invariant set, and calculating optimal nominal control inputs and error feedback control inputs of the intelligent connected vehicle using a preset receding horizon control framework and the pipeline-based robust data-driven predictive control strategy;

[0010] A final control input of the intelligent connected vehicle is obtained based on the optimal nominal control input and the error feedback control input, so that a controller controls the mixed vehicle queue to be controlled based on the final control input.

[0011] According to one embodiment of the present application, determining a data-driven minimum robust positive invariant set based on the over-approximation system matrix set includes:

[0012] determining a subsystem matrix set according to the over-approximation system matrix set;

[0013] Based on a preset probability measure, determining a finite sample set from the subsystem matrix set, solving a linear matrix inequality using each sample pair in the finite sample set to obtain a first optimization variable and a second optimization variable, and determining a stable feedback control law based on the first optimization variable and the second optimization variable;

[0014] The stable feedback control law is used to calculate the data-driven minimum robust positive invariant set.

[0015] According to one embodiment of the present application, the pipeline-based robust data-driven predictive control strategy is:

[0016]

[0017] Subject to:

[0018]

[0019] ∈ z (i|k)=∈(k);

[0020] x z (0|k)=x z (k);

[0021]

[0022] Among them, u(k) is the control input data sequence, x(k) is the state output data sequence, r(i|k) is the expected reference state at time step k+i, Q is the penalty state deviation matrix, R is the control input matrix, x z (i|k) is the state of the nominal system, u z (i|k) is the control input of the nominal system, ∈ z (i|k) is the disturbance input of the nominal system, x z (0|k) is the initial state of the nominal system, is the reachable set of the (i+1|k)th step of the nominal system, is the reachable set of the (i|k)th step of the nominal system, is the initial reachable set of the nominal system, is the set of over-approximation system matrices, is the state constraint, is the control input constraint, K is the stable feedback control law, is the minimum robust positive invariant set, k is the time step, i is a natural number, and N is an integer.

[0023] According to one embodiment of the present application, the final control input is:

[0024]

[0025] in, is the first control input, u e (k) is the error feedback control input, K is the stable feedback control law, x e (k) is the state of the error system.

[0026] According to one embodiment of the present application, obtaining a target input data sequence and a target output data sequence based on the mixed vehicle queue to be controlled includes:

[0027] Applying a preset control input, a preset interference input, and a preset noise input to the mixed vehicle queue to be controlled to obtain an initial input data sequence and an initial output data sequence;

[0028] The initial input data sequence and the initial output data sequence are respectively recombined to obtain the target input data sequence and the target output data sequence.

[0029] According to the pipeline-based robust data-driven control method for mixed vehicle platoons proposed in an embodiment of the present application, the target input data sequence and target output data sequence of the mixed vehicle platoon to be controlled are obtained, and a set of over-approximation system matrices is constructed to determine the minimum robust positive invariant set. Based on this, a pipeline-based robust data-driven predictive control strategy is formulated, combining safety and performance constraints, and the optimal control inputs for the intelligent connected vehicle are calculated. Ultimately, the controller can use these control inputs to manage the mixed vehicle platoon to be controlled. This solves the problems of low computational efficiency and insufficient robustness of existing control strategies, achieving safe control of mixed vehicle fleets and improving driving safety.

[0030] To achieve the above objectives, a second embodiment of the present application provides a pipeline-based hybrid vehicle platoon robust data-driven control device, comprising:

[0031] an acquisition module, configured to acquire a target input data sequence and a target output data sequence based on a mixed vehicle queue to be controlled;

[0032] a determination module, configured to create an over-approximation system matrix set based on the target input data sequence and the target output data sequence, and determine a data-driven minimum robust positive invariant set according to the over-approximation system matrix set;

[0033] a calculation module, configured to determine a pipeline-based robust data-driven predictive control strategy integrating safety and performance constraints based on the minimum robust positive invariant set, and calculate optimal nominal control inputs and error feedback control inputs of the intelligent connected vehicle using a preset receding horizon control framework and the pipeline-based robust data-driven predictive control strategy;

[0034] A control module is configured to obtain a final control input of the intelligent connected vehicle based on the optimal nominal control input and the error feedback control input, so that a controller controls the mixed vehicle queue to be controlled based on the final control input.

[0035] According to one embodiment of the present application, the determining module is specifically configured to:

[0036] determining a subsystem matrix set according to the over-approximation system matrix set;

[0037] Based on a preset probability measure, determining a finite sample set from the subsystem matrix set, solving a linear matrix inequality using each sample pair in the finite sample set to obtain a first optimization variable and a second optimization variable, and determining a stable feedback control law based on the first optimization variable and the second optimization variable;

[0038] The stable feedback control law is used to calculate the data-driven minimum robust positive invariant set.

[0039] According to one embodiment of the present application, the pipeline-based robust data-driven predictive control strategy is:

[0040]

[0041] Subject to:

[0042]

[0043] ∈ z (i|k)=∈(k);

[0044] x z (0|k)=x z (k);

[0045]

[0046] Among them, u(k) is the control input data sequence, x(k) is the state output data sequence, r(i|k) is the expected reference state at time step k+i, Q is the penalty state deviation matrix, R is the control input matrix, x z (i|k) is the state of the nominal system, u z (i|k) is the control input of the nominal system, ∈ z (i|k) is the disturbance input of the nominal system, x z (0|k) is the initial state of the nominal system, is the reachable set of the (i+1|k)th step of the nominal system, is the reachable set of the (i|k)th step of the nominal system, is the initial reachable set of the nominal system, is the set of over-approximation system matrices, is the state constraint, is the control input constraint, K is the stable feedback control law, is the minimum robust positive invariant set, k is the time step, i is a natural number, and N is an integer.

[0047] According to one embodiment of the present application, the final control input is:

[0048]

[0049] in, is the first control input, u e (k) is the error feedback control input, K is the stable feedback control law, x e (k) is the state of the error system.

[0050] According to one embodiment of the present application, the acquisition module is specifically configured to:

[0051] Applying a preset control input, a preset interference input, and a preset noise input to the mixed vehicle queue to be controlled to obtain an initial input data sequence and an initial output data sequence;

[0052] The initial input data sequence and the initial output data sequence are respectively recombined to obtain the target input data sequence and the target output data sequence.

[0053] According to the pipeline-based robust data-driven control device for a mixed vehicle fleet proposed in an embodiment of the present application, the target input data sequence and target output data sequence of the mixed vehicle fleet to be controlled are obtained, and a set of over-approximation system matrices is constructed to determine the minimum robust positive invariant set. Based on this, a pipeline-based robust data-driven predictive control strategy is formulated, incorporating safety and performance constraints, and the optimal control inputs for the intelligent connected vehicle are calculated. Ultimately, the controller can use these control inputs to manage the mixed vehicle fleet to be controlled. This solves the problems of low computational efficiency and insufficient robustness of existing control strategies, enabling safe control of mixed vehicle fleets and improving driving safety.

[0054] To achieve the above-mentioned objectives, the third aspect of the present application proposes an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the pipeline-based hybrid vehicle platoon robust data-driven control method as described in the above-mentioned embodiment.

[0055] To achieve the above-mentioned objectives, the fourth embodiment of the present application proposes a computer-readable storage medium on which a computer program is stored, which is executed by a processor to implement the pipeline-based hybrid vehicle platoon robust data-driven control method as described in the above-mentioned embodiment.

[0056] To achieve the above objectives, the fifth embodiment of the present application proposes a computer program product, which includes a computer program. When the computer program is executed by a processor, it is used to implement the pipeline-based hybrid vehicle queue robust data-driven control method as described in the above embodiment.

[0057] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0059] Figure 1 Flowchart of a pipeline-based hybrid vehicle platoon robust data-driven control method provided according to an embodiment of the present application;

[0060] Figure 2 is a schematic diagram of a mixed vehicle platoon according to one embodiment of the present application;

[0061] Figure 3 is a flowchart of offline data collection for a mixed vehicle queue according to one embodiment of the present application;

[0062] Figure 4is a flowchart of offline learning for a mixed vehicle platoon according to one embodiment of the present application;

[0063] Figure 5 is an online control flow chart for a mixed vehicle platoon according to one embodiment of the present application;

[0064] Figure 6 Schematic diagram of a block diagram of a pipeline-based hybrid vehicle platoon robust data-driven control device according to an embodiment of the present application;

[0065] Figure 7 A schematic diagram of the structure of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0066] The following describes in detail embodiments of the present application. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0067] The following describes a pipeline-based hybrid vehicle platoon robust data-driven control method and apparatus according to an embodiment of the present application with reference to the accompanying drawings.

[0068] Figure 1 The flowchart of a pipeline-based hybrid vehicle platoon robust data-driven control method according to an embodiment of the present application is shown.

[0069] Before introducing the pipeline-based hybrid vehicle platoon robust data-driven control method proposed in the embodiment of the present application, the relevant technical background is first introduced.

[0070] The presence of HDVs in mixed vehicle platoons introduces a human-in-the-loop system, which is inherently complex due to the unpredictable nature of human driving behavior. Current research primarily relies on model-based approaches to address the associated control challenges. These approaches typically involve modeling the behavior of HDVs using established driver behavior models, such as the Optimal Velocity Model (OVM) and the Intelligent Driver Model (IDM). The resulting system dynamics serve as the basis for designing controllers using techniques such as optimal control, robust control, and model predictive control (MPC). These controllers aim to generate control inputs for ICVs to stabilize and optimize the behavior of mixed vehicle platoons. However, the effectiveness of these model-based approaches is often limited by the inherent uncertainties of human driving behavior. These uncertainties, combined with the complexity of vehicle dynamics and chassis components, make it challenging to develop accurate models for mixed vehicle platoons based solely on first principles. Such modeling inaccuracies can degrade the performance of control algorithms, affect the stability of mixed vehicle platoons, exacerbate traffic fluctuations, and even pose safety risks.

[0071] To address the modeling challenges of controlling mixed vehicle platoons, data-driven control methods offer great potential as an innovative alternative. Compared to traditional model-based approaches, which rely on predefined system dynamics, data-driven approaches directly leverage input and output data to predict system behavior. Using input and output data, these methods can predict future system outputs and optimize control inputs to achieve specific performance goals, providing a flexible framework for managing complex systems.

[0072] In the field of mixed-vehicle platoon control, researchers have proposed a variety of data-driven strategies, including distributed data-driven model predictive control, data-supported predictive control, and their derivatives. These strategies have demonstrated promising results in various scenarios, including simulation studies, hardware-in-the-loop experiments, and driver-in-the-loop experiments, particularly for handling traffic waves. Despite these advances, most data-driven approaches have been validated under idealized conditions, limiting their real-world application. In practice, the formation and operation of mixed-vehicle platoons are significantly affected by communication networks, which are susceptible to noise. Furthermore, speed fluctuations of leading vehicles outside the platoon often introduce disturbances that propagate through the system. Current research has not adequately addressed how to enhance resistance to noise and disturbances in data-driven control frameworks. This oversight undermines the reliability and robustness of current approaches and could compromise their performance in real-world applications where noise and disturbances are prevalent.

[0073] Recent research progress has focused on improving the robustness of mixed vehicle platoon control systems. For example, a data-supported predictive control framework enhanced with min-max robust optimization has shown effectiveness in mitigating unknown disturbances from the lead vehicle. Another notable approach is fully symmetric polytope (zonotope) data-driven predictive control, which exploits linear reachability analysis to mitigate the effects of observation noise on data collection. Although these approaches represent significant progress, they also introduce significant limitations. For example, the use of min-max robust optimization involves solving a complex two-level optimization problem, which increases computational cost and is not suitable for real-time applications. Similarly, the zonotope data-driven predictive control approach requires multi-step forward reachability set computation, which requires extensive computational resources and high numerical accuracy. Therefore, addressing these computational inefficiencies is crucial for advancing robust data-driven control strategies in mixed vehicle platoon systems.

[0074] It is precisely based on the above problems that the embodiments of the present application propose a pipeline-based robust data-driven control method for a mixed vehicle queue. This method is based on pipeline robust data-driven predictive control (TRDDPC). By collecting input-output data from the mixed vehicle queue, a matrix zonotope is constructed to directly address system uncertainty, and then a data-driven minimum robust positive invariant set is calculated. The boundaries of system uncertainty and interference are clearly defined. Based on the minimum robust positive invariant set, a nominal TRDDPC optimization problem with integrated safety constraints is formulated and solved using a rolling horizon control framework to obtain the final control input.

[0075] Next, the pipeline-based hybrid vehicle platoon robust data-driven control method proposed in the embodiment of the present application will be described in detail.

[0076] For example, Figure 1 As shown, the pipeline-based hybrid vehicle platoon robust data-driven control method includes the following steps:

[0077] In step S101 , a target input data sequence and a target output data sequence are obtained based on a hybrid vehicle queue to be controlled.

[0078] Among them, such as Figure 2 As shown in FIG, the mixed vehicle queue to be controlled may be a mixed queue consisting of an intelligent connected vehicle (ICV) and multiple human-driven vehicles (HDVs), as well as a tracked vehicle outside the mixed vehicle queue.

[0079] Specifically, specific data can be collected first to provide a basis for subsequent control strategies. During the data collection phase of the embodiment of the present application, the system of the hybrid vehicle platoon to be controlled can be stimulated by applying small control inputs to the intelligent connected vehicles in the platoon and interference (i.e., disturbance input) to the leading vehicle (i.e., the tracked vehicle), thereby collecting offline data, i.e., target input data sequences and target output data sequences.

[0080] The following further describes how to obtain a target input data sequence and a target output data sequence based on the mixed vehicle queue to be controlled.

[0081] As a possible implementation method, in some embodiments, obtaining a target input data sequence and a target output data sequence based on the mixed vehicle queue to be controlled includes: applying a preset control input, a preset interference input, and a preset noise input to the mixed vehicle queue to be controlled to obtain an initial input data sequence and an initial output data sequence; and recombining the initial input data sequence and the initial output data sequence respectively to obtain a target input data sequence and a target output data sequence.

[0082] Specifically, the state vector x(k) of the mixed vehicle platoon system to be controlled is mainly affected by the following factors: control input u(k) (i.e., preset control input), head vehicle speed ∈(k) (i.e., preset interference input), and random noise input ω(k) (i.e., preset noise input). During the data collection process, the three quantities u(k), ∈(k), and ∈(k) are measurable, while the preset noise input ω(k) cannot be directly measured. However, in the embodiment of this application, it is assumed that ω(k) is finite. Figure 3 As shown in Figure 1, to facilitate data collection, the preset control input u(k) and the preset interference input ∈(k) can be continuously applied to the hybrid vehicle platoon system to be controlled within a certain time period (e.g., time step T+1). This continuous excitation method helps collect a series of data (i.e., initial input data sequence and initial output data sequence), which can show the performance of the hybrid vehicle platoon system to be controlled under different control and interference conditions. The initial input data sequence includes the control input U and the interference input E, and the initial output data sequence is the corresponding state output X, namely:

[0083] U=[u(1),u(2),…,u(T+1)]∈R 1×(T+1) ;

[0084] E=[∈(1),∈(2),…,∈(T+1)]∈R 1×(T+1) ;

[0085] X=[x(1),x(2),…,x(T+1)]∈R 2n×(T+1) ;

[0086] Furthermore, in order to facilitate offline learning and directly use the above data sequence, the embodiment of the present application recombines the initial input data sequence and the initial output data sequence to obtain the target input data sequence and the target output data sequence as follows:

[0087] U - =[u(1),u(2),…,u(T)]∈R 1×T ;

[0088] E - =[ε(1),ε(2),…,∈(T)]∈R 1×T ;

[0089] X - =[x(1),x(2),…,x(T)]∈R 2n×T ;

[0090] X + =[x(2),x(3),…,x(T+1)]∈R 2n×T ;

[0091] Among them, U - is the reorganized control input, E - is the interference input after reorganization, X - is the state of the previous system after reorganization, X + is the state of the system after the reorganization.

[0092] To facilitate subsequent derivation, the target input data sequence also includes noise input, which can be expressed as:

[0093] W - =[ω(1),ω(2),…,ω(T)]∈R 2n×T ;

[0094] Although ω(k) cannot be measured directly, it is a key component in analyzing the behavior of mixed vehicle platooning systems to be controlled.

[0095] In step S102 , an over-approximation system matrix set is created based on the target input data sequence and the target output data sequence, and a data-driven minimum robust positive invariant set is determined according to the over-approximation system matrix set.

[0096] Specifically, after obtaining the key target output data sequence and target input data sequence in the data collection phase, the offline learning phase is entered. In the offline learning phase of the embodiment of the present application, since the system model matrix A (system state matrix), B (control input matrix) and H (disturbance input matrix) in the hybrid vehicle platoon to be controlled are unknown, in order to solve this problem, the pre-collected target output data sequence and target input data sequence can be used to calculate an over-approximated system matrix The over-approximation system matrix set, in the form of a matrixzonotope, systematically encapsulates all feasible system matrices consistent with the observed data, taking into account the effects of noise. This set encompasses the uncertain and unknown dynamic behavior of the mixed vehicle platoon system to be controlled. By leveraging the over-approximation system matrix set, a data-driven minimum robustly positively invariant set (RPI) can be further determined. This minimum robustly positively invariant set explicitly defines the robustness bounds of the system in the face of uncertainty and external disturbances.

[0097] Furthermore, if Figure 4 As shown, the over-approximation system matrix set created based on the target input data sequence and the target output data sequence is All possible system models represented as [ABH] can be over-approximated. This construction process ensures that the resulting over-approximation system matrix set Keeping it consistent with the noise data provides a solid foundation for subsequent analysis and processing. The specific construction process is as follows:

[0098] Given the target input data sequence and target output data sequence U obtained by the hybrid vehicle platoon system to be controlled - 、E - 、X - 、X + , if the matrix With full row rank, the set of all possible matrices [ABH] (i.e., the set of over-approximation system matrices ) can be expressed as:

[0099]

[0100] in, is the Moore-Penrose pseudoinverse of the matrix, is the noise matrix band set, which can be expressed as:

[0101]

[0102] in, for The central matrix of for The generator matrix of is the number of generated vectors.

[0103] Using the derived over-approximation system matrix set The over-approximation data-driven system model can be obtained as follows:

[0104]

[0105] in, For an over-approximated data-driven system model, The overapproximated reachable set of system states x(k) representing the system dynamics.

[0106] Next, we explain in detail how to determine the data-driven minimum robust positive invariant set based on the set of over-approximated system matrices.

[0107] As a possible implementation method, in some embodiments, a data-driven minimum robust positive invariant set is determined based on an over-approximation system matrix set, including: determining a subsystem matrix set based on the over-approximation system matrix set; determining a finite sample set from the subsystem matrix set based on a preset probability measure, solving a linear matrix inequality using each sample pair in the finite sample set to obtain a first optimization variable and a second optimization variable, and determining a stable feedback control law based on the first optimization variable and the second optimization variable; and calculating the data-driven minimum robust positive invariant set using the stable feedback control law.

[0108] Specifically, to ensure that the system matrix set is over-approximated To determine the stability of all potential configurations, the embodiment of the present application can first determine the subsystem matrix set It contains all possible realizations of [AB], and its form can be expressed as:

[0109]

[0110] Among them, I is the identity matrix and O is the zero matrix, and the dimensions of both are consistent with the system space.

[0111] In order to calculate the stable feedback control law K∈R 1×2n , the embodiment of the present application can adopt a probability-based method, that is, in a preset probability measure Next, from the subsystem matrix set Get a finite sample set For each sample pair (A, B) in the finite sample set, solve the following linear matrix inequality (LMI):

[0112]

[0113] Among them, P is the first optimization variable and Z is the second optimization variable.

[0114] After obtaining P and Z, the stable feedback control law K is calculated as follows:

[0115] K=ZP -1 .

[0116] Using this representation, a data-driven stable feedback control law K is derived to ensure Captures the stability of all potential system configurations.

[0117] Next, the stable feedback control law can be used to calculate the minimum robust positive invariant set of data drive. That is, the embodiment of the present application first converts the over-approximated data drive system model into Decoupled into two components: the nominal system and the error system, the dynamic equations of these two systems can be expressed as:

[0118]

[0119] in, is the reachable set of the k+1th step of the nominal system, is the reachable set of the k-th step of the nominal system, u z (k) is the control input of the nominal system, ε z (k) is the prediction error, is the reachable set of the k+1th step of the error system, is the reachable set of the k-th step of the error system, u e (k) is the feedback control input of the error system, ∈ e (k) is the predicted value of the leading vehicle speed, which is set as a constant.

[0120] In the embodiment of the present application, the prediction error ∈ z (k) Limited and meeting the following conditions:

[0121] |ε e (k)| ∞ ≤ε max ;

[0122] Among them, ε max ∈ e (k) upper limit.

[0123] Furthermore, u z (k) and ε z (k), and u e (k),∈ e (k) and original system variables u(k) and ε(k) are intrinsically linked, namely:

[0124]

[0125] u(k)=u z (k)+u e (k);

[0126] ε(k)=∈ z (k)+ε e (k);

[0127] It should be noted that the noise term ω(k) here only affects the dynamics of the error system, but not the dynamics of the nominal system.

[0128] For the defined error system, the embodiment of the present application utilizes a stable feedback control law K to stabilize the error system dynamics, thereby generating the following closed-loop error system:

[0129]

[0130] Among them, the matrix and The construction is as follows:

[0131]

[0132] In order to ensure the boundedness of the closed-loop error system, the embodiment of the present application derives the minimum robust positive invariant set of the closed-loop error system, which is denoted as Defined as a bounded input and The minimal set of all reachable states is given by:

[0133]

[0134] The infinite sum is Minkowski addition. The infinite sum can be calculated only when the matrix is nilpotent. To effectively calculate this set in the non-nilpotent case, the embodiment of the present application proposes an approximate method to calculate the minimum robust positive invariant set, namely:

[0135]

[0136]

[0137] In step S103, a pipeline-based robust data-driven predictive control strategy integrating safety and performance constraints is determined based on a minimum robust positive invariant set, and the optimal nominal control input and error feedback control input of the intelligent connected vehicle are calculated using a preset rolling horizon control framework and a pipeline-based robust data-driven predictive control strategy.

[0138] Specifically, if Figure 5 As shown, after the data collection phase and the offline learning phase, the next phase is the online control phase for the mixed vehicle platoon to be controlled. In this phase, the present embodiment adopts the TRDDPC framework (a time-recursive dynamic programming control framework) in combination with a preset rolling horizon control framework to find the optimal solution through step-by-step backward calculation, thereby calculating the optimal nominal control input for the intelligent connected vehicle.

[0139] The TRDDPC framework integrates two basic components: a nominal controller and an error controller. The error controller plays a crucial role in compensating for disturbances and noise, ensuring that the system state remains within the bounds set by the minimum robust positive invariant set. The error controller's design accounts for the inevitable disturbances and noise encountered in real-world operation, dynamically adjusting the control input to maintain the stability and robustness of the system state. Meanwhile, the nominal controller controls the ideal dynamics of the system, assuming the absence of any disturbances or uncertainties. It operates based on a nominal reachable set model. This model encapsulates the optimal trajectory of the system in the absence of disturbances, providing a blueprint for ideal system operation. The primary goal of the nominal controller is to confine the system state within a constrained "pipeline," effectively limiting the deviation between the actual operating trajectory and the desired nominal trajectory. By maintaining the nominal state within this pipeline, the framework ensures that the deviation remains within an acceptable range, thereby maintaining system stability and performance in the face of uncertainties in real-world operation. The TRDDPC optimization problem (i.e., a pipeline-based robust data-driven predictive control strategy) is formulated as follows:

[0140]

[0141] Among them, u(k) is the control input data sequence, x(k) is the state output data sequence, r(i|k) is the expected reference state at time step k+i, Q is the penalty state deviation matrix, R is the control input matrix, x z (i|k) is the state of the nominal system, u z (i|k) is the control input of the nominal system, k is the time step, i is a natural number, and N is an integer.

[0142] Subject to:

[0143]

[0144] ε z (i|k)=∈(k);

[0145] x z (0|k)=x z (k);

[0146]

[0147] Among them, ∈ z (i|k) is the disturbance input of the nominal system, x z (0|k) is the initial state of the nominal system, is the reachable set of the (i+1|k)th step of the nominal system, is the reachable set of the (i|k)th step of the nominal system, is the initial reachable set of the nominal system, is the set of over-approximation system matrices, is the state constraint, is the control input constraint, K is the stable feedback control law, is the minimum robust positive invariant set.

[0148] Control input data sequence u(k)={u z (k),u z (k+1),…,u z (k+N-1)}, the state output data sequence x(k)={x z (k),x z (k+1),…,x z (k+N-1)}, and optimize on the prediction layer N.

[0149] Furthermore, the desired reference state at time step k+i is:

[0150]

[0151] Among them, r i (k) is the target reference state, is the desired vehicle spacing, is the expected vehicle speed.

[0152] State Constraints Defined as:

[0153]

[0154] in, Limit the deviation of spacing and speed. Here, 1 n is a dimensional vector consisting of 1s, represents the Kronecker product.

[0155] Control input constraints is defined as:

[0156]

[0157] Among them, u max is the maximum input allowed.

[0158] By solving the optimization problem, the optimal nominal control input of the system is obtained:

[0159]

[0160] And the state sequence:

[0161]

[0162] In step S104 , a final control input of the intelligent connected vehicle is obtained based on the optimal nominal control input and the error feedback control input, so that the controller controls the mixed vehicle queue to be controlled based on the final control input.

[0163] Specifically, by taking the first control input of the system's optimal nominal control input With the error feedback control input u e (k) are combined to form the final control input for the intelligent connected vehicle:

[0164]

[0165] in, is the first control input, u e (k) is the error feedback control input, K is the stable feedback control law, x e (k) is the state of the error system.

[0166] This final control input u(k) ensures robust trajectory tracking and mitigates the adverse effects of noise and interference in mixed vehicle platoons.

[0167] In summary: (1) The embodiment of the present application discloses a pipeline-based robust data-driven control technology architecture for mixed vehicle queues. In view of the characteristics of the mixed vehicle queue system, the method explicitly considers the influence of noise and external interference. By adopting a data-driven method to calculate the robust positive invariant set, the control input constraints and safety constraints of the nominal data-driven predictive control problem can be tightened, thereby ensuring the safety and efficient operation of the system under various uncertainties. This solves the problem of insufficient robustness in the existing mixed queue controller design and provides greater possibilities for the efficient operation of the entire traffic system. In addition, it also solves the problem that the existing methods for mixed queue control cannot guarantee safety constraints, realizes the safe control of the mixed queue, and significantly improves driving safety.

[0168] (2) By directly using a data-driven approach to calculate the robust positive invariant set, the arduous system identification and modeling work of traditional methods is bypassed, solving the problem that existing methods rely heavily on models to calculate the robust positive invariant set, while also ensuring the accuracy of the robust positive invariant set calculation. This is crucial for improving the control performance of mixed platoons and also provides strong technical support for building intelligent transportation systems.

[0169] (3) Unlike traditional model-based pipeline model predictive control methods that rely on explicit system models, the framework of the present embodiment can learn the pipeline directly from data, eliminating the need for explicit system modeling. This data-driven approach not only simplifies the implementation process but also enhances adaptability to real-world scenarios.

[0170] (4) Unlike traditional methods, which usually assume that noise and interference can be ignored, the framework of the embodiment of the present application explicitly takes these uncertainties into account as zonotopes when constructing the minimum RPI set. This systematic consideration significantly enhances the robustness of the framework. Based on the derived minimum RPI set, the embodiment of the present application formulates the TRDDPC optimization problem to ensure strict compliance with safety constraints and achieve robust and optimal control of autonomous driving vehicles in a mixed vehicle queue. Finally, the control performance of the mixed queue is analyzed through performance indicators, and the energy consumption and speed fluctuation of the mixed vehicle queue during travel are quantified, thereby providing a strong supporting technology for achieving stable, safe and efficient control of the mixed vehicle queue.

[0171] According to the pipeline-based robust data-driven control method for mixed vehicle platoons proposed in an embodiment of the present application, the target input data sequence and target output data sequence of the mixed vehicle platoon to be controlled are obtained, and a set of over-approximation system matrices is constructed to determine the minimum robust positive invariant set. Based on this, a pipeline-based robust data-driven predictive control strategy is formulated, combining safety and performance constraints, and the optimal control inputs for the intelligent connected vehicle are calculated. Ultimately, the controller can use these control inputs to manage the mixed vehicle platoon to be controlled. This solves the problems of low computational efficiency and insufficient robustness of existing control strategies, achieving safe control of mixed vehicle fleets and improving driving safety.

[0172] Next, a pipeline-based hybrid vehicle platoon robust data-driven control device proposed according to an embodiment of the present application will be described with reference to the accompanying drawings.

[0173] Figure 6 4 is a block diagram of a pipeline-based hybrid vehicle platoon robust data-driven control device according to an embodiment of the present application.

[0174] like Figure 6 As shown, the pipeline-based hybrid vehicle platoon robust data-driven control device 10 includes: an acquisition module 100 , a determination module 200 , a calculation module 300 and a control module 400 .

[0175] The acquisition module 100 is used to acquire a target input data sequence and a target output data sequence based on the mixed vehicle queue to be controlled;

[0176] A determination module 200 is configured to create an over-approximation system matrix set based on a target input data sequence and a target output data sequence, and determine a data-driven minimum robust positive invariant set according to the over-approximation system matrix set;

[0177] A calculation module 300 is configured to determine a pipeline-based robust data-driven predictive control strategy that integrates safety and performance constraints based on a minimum robust positive invariant set, and to calculate optimal nominal control inputs and error feedback control inputs for the intelligent connected vehicle using a preset receding horizon control framework and the pipeline-based robust data-driven predictive control strategy;

[0178] The control module 400 is configured to obtain a final control input of the intelligent connected vehicle based on the optimal nominal control input and the error feedback control input, so that the controller controls the mixed vehicle queue to be controlled based on the final control input.

[0179] Furthermore, in some embodiments, the determination module 200 is specifically configured to:

[0180] Determine a subsystem matrix set according to the over-approximated system matrix set;

[0181] Based on a preset probability measure, a finite sample set is determined from the subsystem matrix set, a linear matrix inequality is solved using each sample pair in the finite sample set to obtain a first optimization variable and a second optimization variable, and a stable feedback control law is determined based on the first optimization variable and the second optimization variable;

[0182] The data-driven minimum robust positive invariant set is calculated using the stable feedback control law.

[0183] Furthermore, in some embodiments, the pipeline-based robust data-driven predictive control strategy is:

[0184]

[0185] Subject to:

[0186]

[0187] ∈ z (i|k)=∈(k);

[0188] x z (0|k)=x z (k);

[0189]

[0190] Among them, u(k) is the control input data sequence, x(k) is the state output data sequence, r(i|k) is the expected reference state at time step k+i, Q is the penalty state deviation matrix, R is the control input matrix, x z (i|k) is the state of the nominal system, u z (i|k) is the control input of the nominal system, ∈ z (i|k) is the disturbance input of the nominal system, x z (0|k) is the initial state of the nominal system, is the reachable set of the (i+1|k)th step of the nominal system, is the reachable set of the (i|k)th step of the nominal system, is the initial reachable set of the nominal system, is the set of over-approximation system matrices, is the state constraint, is the control input constraint, K is the stable feedback control law, is the minimum robust positive invariant set, k is the time step, i is a natural number, and N is an integer.

[0191] Furthermore, in some embodiments, the final control input is:

[0192]

[0193] in, is the first control input, u e (k) is the error feedback control input, K is the stable feedback control law, x e (k) is the state of the error system.

[0194] Furthermore, in some embodiments, the acquisition module 100 is specifically configured to:

[0195] Applying a preset control input, a preset interference input, and a preset noise input to the mixed vehicle queue to be controlled to obtain an initial input data sequence and an initial output data sequence;

[0196] The initial input data sequence and the initial output data sequence are recombined to obtain the target input data sequence and the target output data sequence.

[0197] It should be noted that the aforementioned explanation of the embodiment of the pipeline-based hybrid vehicle platoon robust data-driven control method is also applicable to the pipeline-based hybrid vehicle platoon robust data-driven control device of this embodiment, and will not be repeated here.

[0198] According to the pipeline-based robust data-driven control device for a mixed vehicle fleet proposed in an embodiment of the present application, the target input data sequence and target output data sequence of the mixed vehicle fleet to be controlled are obtained, and a set of over-approximation system matrices is constructed to determine the minimum robust positive invariant set. Based on this, a pipeline-based robust data-driven predictive control strategy is formulated, incorporating safety and performance constraints, and the optimal control inputs for the intelligent connected vehicle are calculated. Ultimately, the controller can use these control inputs to manage the mixed vehicle fleet to be controlled. This solves the problems of low computational efficiency and insufficient robustness of existing control strategies, enabling safe control of mixed vehicle fleets and improving driving safety.

[0199] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include:

[0200] A memory 701 , a processor 702 , and a computer program stored in the memory 701 and executable on the processor 702 .

[0201] When the processor 702 executes the program, the pipeline-based hybrid vehicle platoon robust data-driven control method provided in the above embodiment is implemented.

[0202] Furthermore, the electronic device further includes:

[0203] The communication interface 703 is used for communication between the memory 701 and the processor 702 .

[0204] The memory 701 is used to store computer programs that can be run on the processor 702 .

[0205] The memory 701 may include a high-speed RAM (Random Access Memory) memory, and may also include a non-volatile memory, such as at least one disk memory.

[0206] If the memory 701, processor 702, and communication interface 703 are implemented independently, the communication interface 703, memory 701, and processor 702 can be connected to each other via a bus and communicate with each other. The bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 7 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0207] Optionally, in a specific implementation, if the memory 701, the processor 702 and the communication interface 703 are integrated on a chip, the memory 701, the processor 702 and the communication interface 703 can communicate with each other through an internal interface.

[0208] The processor 702 may be a CPU (Central Processing Unit), or an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present application.

[0209] An embodiment of the present application further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above pipeline-based hybrid vehicle platoon robust data-driven control method.

[0210] An embodiment of the present application further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the computer program implements the above pipeline-based hybrid vehicle platoon robust data-driven control method.

[0211] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of such features. Throughout the description of this application, "plurality" means at least two, for example, two, three, etc., unless otherwise specifically defined.

[0212] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0213] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A pipeline-based robust data-driven control method for mixed vehicle platoons, characterized in that: The following steps are involved: obtaining a target input data sequence and a target output data sequence based on the mixed vehicle queue to be controlled; Creating an over-approximation system matrix set based on the target input data sequence and the target output data sequence, and determining a data-driven minimum robust positive invariant set according to the over-approximation system matrix set; Determining a pipeline-based robust data-driven predictive control strategy that integrates safety and performance constraints based on the minimum robust positive invariant set, and calculating optimal nominal control inputs and error feedback control inputs of the intelligent connected vehicle using a preset receding horizon control framework and the pipeline-based robust data-driven predictive control strategy; Obtaining a final control input for the intelligent connected vehicle based on the optimal nominal control input and the error feedback control input, so that a controller controls the mixed vehicle queue to be controlled based on the final control input; Wherein, the determining of the data-driven minimum robust positive invariant set based on the over-approximation system matrix set includes: determining a subsystem matrix set based on the over-approximation system matrix set, determining a finite sample set from the subsystem matrix set based on a preset probability measure, solving a linear matrix inequality using each sample pair in the finite sample set to obtain a first optimization variable and a second optimization variable, and determining a stable feedback control law based on the first optimization variable and the second optimization variable, and calculating the data-driven minimum robust positive invariant set using the stable feedback control law.

2. The method according to claim 1, characterized in that The pipeline-based robust data-driven predictive control strategy is: Subject to: ∈ z (i|k)=∈(k); x z (0|k)=x z (k); Among them, u(k) is the control input data sequence, x(k) is the state output data sequence, r(i|k) is the expected reference state at time step k+i, Q is the penalty state deviation matrix, R is the control input matrix, x z (i|k) is the state of the nominal system, u z (i|k) is the control input of the nominal system, ∈ z (i|k) is the disturbance input of the nominal system, x z (0|k) is the initial state of the nominal system, is the reachable set of the (i+1|k)th step of the nominal system, is the reachable set of the (i|k)th step of the nominal system, is the initial reachable set of the nominal system, is the set of over-approximation system matrices, is the state constraint, is the control input constraint, K is the stable feedback control law, is the minimum robust positive invariant set, k is the time step, i is a natural number, and N is an integer.

3. The method according to claim 1, characterized in that The final control input is: in, is the first control input, u e (k) is the error feedback control input, K is the stable feedback control law, x e (k) is the state of the error system.

4. The method according to claim 1, wherein The step of obtaining a target input data sequence and a target output data sequence based on the mixed vehicle queue to be controlled includes: Applying a preset control input, a preset interference input, and a preset noise input to the mixed vehicle queue to be controlled to obtain an initial input data sequence and an initial output data sequence; The initial input data sequence and the initial output data sequence are respectively recombined to obtain the target input data sequence and the target output data sequence.

5. A pipeline-based hybrid vehicle platoon robust data-driven control device, characterized in that: include: an acquisition module, configured to acquire a target input data sequence and a target output data sequence based on a mixed vehicle queue to be controlled; a determination module, configured to create an over-approximation system matrix set based on the target input data sequence and the target output data sequence, and determine a data-driven minimum robust positive invariant set according to the over-approximation system matrix set; a calculation module, configured to determine a pipeline-based robust data-driven predictive control strategy integrating safety and performance constraints based on the minimum robust positive invariant set, and calculate optimal nominal control inputs and error feedback control inputs of the intelligent connected vehicle using a preset receding horizon control framework and the pipeline-based robust data-driven predictive control strategy; a control module, configured to obtain a final control input of the intelligent connected vehicle based on the optimal nominal control input and the error feedback control input, so that a controller controls the mixed vehicle queue to be controlled based on the final control input; Among them, the determination module is specifically used to: determine the subsystem matrix set based on the over-approximation system matrix set, determine a finite sample set from the subsystem matrix set based on a preset probability measure, use each sample pair in the finite sample set to solve the linear matrix inequality to obtain a first optimization variable and a second optimization variable, and determine a stable feedback control law based on the first optimization variable and the second optimization variable, and use the stable feedback control law to calculate the data-driven minimum robust positive invariant set.

6. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the pipeline-based hybrid vehicle platoon robust data-driven control method according to any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the pipeline-based hybrid vehicle platoon robust data-driven control method according to any one of claims 1 to 4.

8. A computer program product, characterized in that The method comprises a computer program, which is used to implement the pipeline-based hybrid vehicle platoon robust data-driven control according to any one of claims 1 to 4 when the computer program is executed by a processor.

Images