A vehicle queuing control method, device, electronic equipment, and storage medium.

Abstract

This invention relates to a vehicle queuing control method, device, electronic device, and storage medium, belonging to the field of intelligent transportation technology. The method includes: constructing an objective function based on the first state of a target vehicle at the current moment, the first state including the target vehicle's position, speed, and acceleration; solving the objective function using an MPC algorithm to obtain a control signal; inputting the first state and the control signal into a PINN model to obtain the second state of the target vehicle at the next moment; feeding the second state back into the MPC algorithm to reacquire the control signal and iterating through all vehicles in the queue to achieve real-time control of the vehicle queuing. This invention primarily addresses situations where the precise dynamics of vehicles are unknown and only partial physical information is available. It achieves real-time control of the vehicle queuing while reducing data requirements, thus effectively solving the problem of high data demands in existing technologies.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation technology, and in particular to a vehicle queuing control method, device, electronic device, and storage medium. Background Technology

[0002] Intelligent vehicle control methods based on physical models require accurate dynamic models. However, factors such as vehicle heterogeneity, uncertainties in the operating environment, and uncertainties in vehicle parameters make obtaining accurate dynamic models costly.

[0003] Data-driven model predictive control methods approximate real vehicle dynamics models using machine learning models. For example, neural network-based model predictive control (NN-MPC) combines neural networks (NNs) and model predictive control (MPC). In this approach, the neural network is used to learn or approximate a dynamic system model, which is then used in the MPC algorithm to optimize future control actions. Neural networks, especially deep neural networks, possess powerful learning and representation capabilities, enabling them to handle complex nonlinear relationships and high-dimensional data. Through the training process, neural networks can automatically learn the mapping relationship between input and output data and use it to approximate the dynamic behavior of the actual system. This allows NN-MPC to handle complex nonlinear systems that are difficult for traditional MPC to process.

[0004] Model-driven control (MPC) is a model-based control method that achieves optimal control by predicting the future behavior of a system and optimizing control actions. In neural network-based MPC, MPC uses a model learned by a neural network to make predictions and calculate the system state and output over a future time period. Then, by searching for the optimal control sequence among these predictions, MPC calculates and implements the optimal sequence of control actions to achieve the best control effect for the system. The advantage of NN-MPC lies in its ability to combine the learning capabilities of neural networks with the optimization control capabilities of MPC, thereby improving the system's control performance. It is applicable to control problems of various nonlinear, time-varying, and constrained systems and has broad application prospects in fields such as industrial automation, process control, and traffic management. However, data-driven control methods, represented by NN-MPC, also face some challenges, such as the need for large amounts of data to train machine learning models and the potential for overfitting and generalization issues. Summary of the Invention

[0005] In view of this, it is necessary to provide a vehicle queuing control method to solve the problem of the large data requirements of existing technologies.

[0006] To address the above problems, the present invention provides a vehicle queuing control method, comprising:

[0007] The objective function is constructed based on the first state of the target car at the current moment. The first state includes the position, speed and acceleration of the target car.

[0008] The control signal is obtained by solving the objective function based on the MPC algorithm;

[0009] The first state and control signals are input into the PINN model to obtain the second state of the target car at the next moment;

[0010] The second state is fed back into the MPC algorithm to reacquire the control signal and traverse all cars in the queue to achieve real-time control of the car queue.

[0011] In one possible implementation, the control signals include the vehicle's accelerator and brake opening / closing angles.

[0012] In one possible implementation, constructing the objective function based on the first state of the target vehicle includes:

[0013] Obtain the following trajectory of the target car, the following trajectory including the position and speed of the preceding vehicle;

[0014] The following error is calculated based on the first state and the following trajectory.

[0015] Based on the following error, an objective function for solving the control signal is constructed;

[0016] The objective function includes:

[0017]

[0018] in, J () indicates the difference between the vehicle's state and the reference trajectory. This indicates the preset prediction range. k Indicates time, j Indicates the first j Target car R Indicates weight, u Indicates control signal, Indicates the first j Control signals for the target vehicle Indicates the first j The target car k+i Timing control signals, This indicates the following error.

[0019] In one possible implementation, the MPC algorithm includes the following constraints:

[0020]

[0021] in, Indicates the lower bound of the control signal. Indicates the upper bound of the control signal. Indicates the first j The target car in the l- Control signal at time 1 Indicates the first j The position of the target vehicle at any given time within a preset range. Indicates the first j The speed of the target vehicle at any given time within a preset range. This indicates the preset safe distance between vehicles. This indicates the maximum number of vehicles following in the car queue.

[0022] In one possible implementation, the step of solving the objective function based on the MPC algorithm to obtain the control signal includes:

[0023] The MPC algorithm iteratively solves the objective function within a preset prediction range until the objective function reaches a global minimum, thereby obtaining a control signal sequence.

[0024] Output the first control signal in the control signal sequence.

[0025] In one possible implementation, the MPC algorithm iteratively solves the objective function within a preset prediction range, including:

[0026] The MPC algorithm solves the objective function based on the LQR regulator.

[0027] In one possible implementation, inputting the first state and control signal into the PINN model to obtain the second state of the target vehicle at the next moment includes:

[0028] An autoregressive prediction function is constructed based on the PINN model;

[0029] The first state and the control signal are input into the autoregressive prediction function to obtain the second state;

[0030] The autoregressive prediction functions include:

[0031]

[0032] in, Indicates the first j The target car in the l The state at any given moment, This indicates the following trajectory of the vehicle following the target car. Indicates the first j The target car in the l Location at any given moment Indicates the first j The target car in the l The speed of time, Indicates the first j The target car in the l acceleration at any moment Indicates the first j The initial state of the target car at time 0. Indicates the first j The target car in the k Timing control signals, Indicates the first j PINN model of the target car Indicates the first j An autoregressive prediction model for a target vehicle. This represents hyperparameters.

[0033] The present invention also provides a vehicle queuing control device, comprising:

[0034] The control signal module is used to construct the objective function based on the first state of the target vehicle and solve for the control signal based on the MPC algorithm.

[0035] The state prediction module is used to input the first state and control signals into the PINN model to obtain the second state of the target car at the next moment.

[0036] The control module is used to feed back the second state to the MPC algorithm, reacquire the control signal, and traverse all cars in the queue to achieve real-time control of the car queue.

[0037] The present invention also provides an electronic device, comprising:

[0038] Memory, used to store programs;

[0039] A processor, coupled to the memory, is configured to execute the program stored in the memory to implement the steps of the vehicle queuing control method described in any of the preceding method items.

[0040] The present invention also provides a storage medium for storing a computer-readable program or instruction, which, when executed by a processor, can implement the steps of the vehicle queuing control method described in any one of the above-described methods.

[0041] The beneficial effects of this invention are as follows: This invention provides a vehicle queuing control method that combines PINN with distributed model predictive control (MPC). Based on the MPC algorithm, at each time step, an objective function is constructed according to the current state of the vehicle, and the objective function is iteratively solved to make the predicted state of the vehicle as close as possible to a preset reference trajectory, thereby obtaining the optimal control signal (throttle, brake opening angle, etc.). After obtaining the control signal within a finite time range, the vehicle's state at that time step is obtained based on the control signal, and the vehicle's state at that time step is fed back to the MPC. The above process is repeated to achieve real-time vehicle control. This invention is mainly aimed at situations where the precise dynamics of the vehicle are unknown and only partial physical information is available. It only requires obtaining the vehicle's speed, position, and acceleration information to achieve real-time control of the vehicle queuing while reducing the amount of data required, thus effectively solving the problem of large data requirements in existing technologies. Attached Figure Description

[0042] Figure 1 A flowchart illustrating an embodiment of the vehicle queue control method provided by the present invention;

[0043] Figure 2 for Figure 1 A flowchart of a method according to an embodiment of step S101;

[0044] Figure 3 for Figure 1 A flowchart of a method according to an embodiment of step S102;

[0045] Figure 4 for Figure 1 A flowchart of a method according to an embodiment of step S103;

[0046] Figure 5 This is a schematic diagram of the control strategy flow provided by the present invention;

[0047] Figure 6 This is a schematic diagram of an embodiment of the vehicle queue control device provided by the present invention;

[0048] Figure 7 A schematic diagram of the structure of an embodiment of the electronic device provided by the present invention. Detailed Implementation

[0049] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not intended to limit the scope of the present invention.

[0050] In the description of the embodiments of the present invention, unless otherwise stated, "multiple" means two or more. "And / or" describes the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0051] The terms "first," "second," etc., used in the embodiments of this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a technical feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature.

[0052] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0053] To solve the above problems, such as Figure 1 As shown, the present invention provides a vehicle queuing control method, comprising:

[0054] S101. Construct the objective function based on the first state of the target car;

[0055] Preferably, the first state includes the target vehicle's position, speed, and acceleration.

[0056] S102. The control signal is obtained by solving the objective function based on the MPC algorithm;

[0057] Preferably, the control signals include: the accelerator and brake opening / closing angles of the vehicle.

[0058] S103. Input the first state and control signal into the PINN model to obtain the second state of the target car at the next moment;

[0059] S104. Feed the second state back to the MPC algorithm, reacquire the control signal, and traverse all cars in the queue to achieve real-time control of the car queue.

[0060] It should be noted that this invention is mainly aimed at situations where the precise dynamics of a vehicle are unknown and only partial physical information is available. It provides a Physics-informed Machine Learning (PINN) network to combine limited observation data with some known physical information. After obtaining the PINN model, the PINN is combined with distributed model predictive control (MPC) for heterogeneous vehicle control.

[0061] PINN is a type of neural network based on physical information. By introducing physical equations as constraints, the fitted results are made to better conform to physical laws.

[0062] Combining PINN with MPC, PINN-MPC fully leverages PINN's capabilities in physical law modeling and MPC's advantages in optimized control. This approach not only handles complex nonlinear dynamic systems but also ensures that the control strategy satisfies physical constraints, thereby improving system performance and stability and achieving a more efficient and reliable control strategy.

[0063] The PINN-MPC method outperforms traditional data-driven MPC (NN-MPC) without physical constraints. Experiments show that PINN-MPC has great potential for heterogeneous vehicle fleet control, reducing data acquisition costs and requiring only partial knowledge of physics.

[0064] Compared with existing technologies, the present invention provides a vehicle queuing control method that combines PINN with distributed model predictive control (MPC). Based on the MPC algorithm, at each time step, an objective function is constructed according to the current state of the vehicle, and the objective function is iteratively solved to make the predicted state of the vehicle as close as possible to a preset reference trajectory, thereby obtaining the optimal control signal (throttle, brake opening angle, etc.). After obtaining the control signal within a finite time range, the vehicle state at that time step is obtained based on the control signal, and the vehicle state at that time step is fed back to the MPC. The above process is repeated to achieve real-time vehicle control. The adaptive control strategy of MPC reduces the requirement for model accuracy. The accurate model in MPC can reduce controller oscillation and promote faster convergence. Real-time control of the vehicle queuing is achieved while reducing the amount of data required, thus effectively solving the problem of large data requirements in existing technologies.

[0065] like Figure 2 In one possible implementation, step S101 includes:

[0066] S201. Obtain the following trajectory of the target car;

[0067] Preferably, the following trajectory includes the position and speed of the preceding vehicle;

[0068] It is understood that in this embodiment, the first car in the car queue is the lead car, and its trajectory is predetermined, denoted as . The second car in the queue (i.e. the following car) uses the trajectory of the lead car as a reference and obtains the control signal for the next moment in real time based on the method given in this invention, so as to ensure that it can maintain a safe distance from the lead car while achieving real-time following. Similarly, the third car in the queue uses the trajectory of the second car as a reference and continues to obtain the control signal, and so on, until all cars in the queue are traversed, thus completing the real-time control of the queue.

[0069] S202. The following error is calculated based on the first state and the following trajectory;

[0070] Preferably, the following error includes: the interval error between vehicles and the speed error.

[0071] S203. Construct an objective function for solving the control signal based on the following error;

[0072] Specifically, the objective function includes:

[0073] (1)

[0074] In equation (1), J () represents the difference between the predicted state of the vehicle and the reference trajectory. This indicates the preset prediction range. k Indicates time, j Indicates the first j Target car R Indicates weight, u Indicates control signal, Indicates the first j Control signals for the target vehicle Indicates the first j The target car k+i Timing control signals, This indicates the following error.

[0075] In one possible implementation, the MPC algorithm includes the following constraints:

[0076] (2)

[0077] In equation (2), Indicates the lower bound of the control signal. Indicates the upper bound of the control signal. Indicates the first j The target car in the l- Control signal at time 1 Indicates the firstj The position of the target vehicle at any given time within a preset range. Indicates the first j The speed of the target vehicle at any given time within a preset range. This indicates the preset safe distance between vehicles. This indicates the maximum number of vehicles following in the car queue.

[0078] It should be noted that the constraints consist of two parts: 1) the control signal boundary, which represents the vehicle's acceleration and deceleration capabilities; and 2) the constraint on the vehicle's terminal state at the end of the prediction range, to ensure that the vehicle state converges to the desired safe region. and It is the terminal state predicted in the Local Linear Quadratic Regulator (LQR), whose constraints are determined by the previous vehicle. j- 1. Confirmed. It is important to note that the control signal boundary is a hard constraint that must be satisfied during LQR optimization. However, the termination constraint may not achieve the desired termination state, but it can approximate it.

[0079] Furthermore, having defined the control objectives and constraints, a proposed physical and data-driven MPC framework was developed, as shown below:

[0080] (3)

[0081] In equation (3), This indicates the reference trajectory of the following vehicle. Indicates the first j- The reference trajectory of a car Indicates the first j- The reference trajectory of a car represents the first... j- One car in the first l The reference trajectory at time 1, i.e., the 1st j The reference trajectory of the car following the car is such that, except for the lead car at the very beginning and the following car at the very end, any car in the queue can be either the "follower" car of the car in front or the "leader" car of the car behind.

[0082] In implementation, the distributed PINN-MPC optimal control framework can be implemented using MATLAB. The function is solved numerically.

[0083] Although the adaptive control strategy of MPC alleviates the need for model accuracy, an accurate model in MPC can reduce controller oscillations and promote faster convergence.

[0084] like Figure 3 In one possible implementation, step S102 includes:

[0085] S301. Based on the MPC algorithm, iteratively solve the objective function within the preset prediction range until the objective function reaches the global minimum value, and obtain the control signal sequence.

[0086] It should be noted that since the MPC algorithm solves the objective function within a given time range, it will obtain several control signals at different times. Arranging these control signals in chronological order will yield a control signal sequence.

[0087] In one possible implementation, the MPC algorithm solves the objective function based on an LQR regulator.

[0088] S302, the first control signal in the output control signal sequence.

[0089] Understandably, given the current state of the vehicle and the reference trajectory (the previous vehicle) predicted by PINN in an autoregressive manner, the optimal control signal is obtained through local LQR. Within the prediction range, the optimal control signal solved for each time instance is... However, only the first control signal... It is applied to vehicles to propagate the vehicle state from the current time step to the next time step.

[0090] like Figure 4 In one possible implementation, step S103 includes:

[0091] S401. Construct an autoregressive prediction function based on the PINN model;

[0092] S402. Input the first state and control signal into the autoregressive prediction function to obtain the second state;

[0093] Specifically, the autoregressive prediction function includes:

[0094] (4)

[0095] In equation (4), Indicates the first j The target car in the l The state at any given moment, This indicates the following trajectory of the vehicle following the target car. Indicates the first j The target car in the l Location at any given moment Indicates the first j The target car in the l The speed of time, Indicates the first j The target car in the l acceleration at any moment Indicates the first j The initial state of the target car at time 0. Indicates the first j The target car in the k Timing control signals, Indicates the first j PINN model of the target car Indicates the first j An autoregressive prediction model for a target vehicle. This represents hyperparameters.

[0096] It should be noted that, in this embodiment, except for the following vehicles whose reference trajectories are the trajectory of the lead vehicle, the reference trajectories of other following vehicles can all be predicted by this function, i.e., the first... j The reference trajectory of the target car is based on the first j- Predicted status of 1 vehicle.

[0097] Thus, the control strategy flow diagram of the present invention is as follows: Figure 5 As shown.

[0098] like Figure 6 The present invention also provides a vehicle queue control device 60, comprising:

[0099] The control signal module 610 is used to construct an objective function based on the first state of the target vehicle and solve for the control signal based on the MPC algorithm.

[0100] The state prediction module 620 is used to input the first state and control signals into the PINN model to obtain the second state of the target car at the next moment.

[0101] The control module 630 is used to feed back the second state to the MPC algorithm, reacquire the control signal, and traverse all cars in the queue to achieve real-time control of the car queue.

[0102] like Figure 7 As shown, the present invention also provides an electronic device 70, comprising:

[0103] Memory 710 is used to store programs;

[0104] The processor 720, coupled to the memory 710, is used to execute the program stored in the memory 710 to implement the steps in the vehicle queuing control method described in any of the above embodiments.

[0105] Figure 7 Only some components of the electronic device 70 are shown, but it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0106] In some embodiments, processor 720 may be a central processing unit (CPU), microprocessor, or other data processing chip for running program code stored in memory 710 or processing data, such as the car queue control method of the present invention.

[0107] In some embodiments, processor 720 may be a single server or a group of servers. The server group may be centralized or distributed. In some embodiments, processor 720 may be local or remote. In some embodiments, processor 720 may be implemented on a cloud platform. In one embodiment, the cloud platform may include a private cloud, public cloud, hybrid cloud, community cloud, distributed cloud, internal cloud, multi-cloud, etc., or any combination thereof.

[0108] In some embodiments, memory 710 may be an internal storage unit of electronic device 70, such as a hard disk or memory of electronic device 70. In other embodiments, memory 710 may also be an external storage device of electronic device 70, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. equipped on electronic device 70.

[0109] Furthermore, the memory 710 may include both internal storage units of the electronic device 70 and external storage devices. The memory 710 is used to store application software and various types of data installed on the electronic device 70.

[0110] In one embodiment, when the processor 720 executes the vehicle queue control program in the memory 710, the following steps can be implemented:

[0111] Construct the objective function based on the first state of the target car;

[0112] The control signal is obtained by solving the objective function based on the MPC algorithm;

[0113] The first state and control signals are input into the PINN model to obtain the second state of the target car at the next moment;

[0114] The second state is fed back into the MPC algorithm to reacquire the control signal and traverse all cars in the queue to achieve real-time control of the car queue.

[0115] Furthermore, the embodiments of the present invention do not specifically limit the type of electronic device 70 mentioned. Electronic device 70 can be a mobile phone, tablet computer, personal digital assistant (PDA), wearable device, laptop computer, or other portable electronic device. Exemplary embodiments of portable electronic devices include, but are not limited to, portable electronic devices running iOS, Android, Microsoft, or other operating systems. The aforementioned portable electronic device can also be other portable electronic devices, such as a laptop computer with a touch-sensitive surface (e.g., a touch panel). It should also be understood that in some other embodiments of the present invention, electronic device 70 may not be a portable electronic device, but rather a desktop computer with a touch-sensitive surface (e.g., a touch panel).

[0116] The present invention also provides a storage medium for storing a computer-readable program or instruction, which, when executed by a processor, can implement the steps of the vehicle queuing control method described in any of the above embodiments.

[0117] Those skilled in the art will understand that all or part of the processes of the methods described in the above embodiments can be implemented by a computer program instructing related hardware (such as a processor, controller, etc.), and the computer program can be stored in a computer storage medium. The computer storage medium can be a disk, optical disk, read-only memory, or random access memory, etc.

[0118] The vehicle queue control method, device, electronic device, and storage medium provided by the present invention have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A vehicle queuing control method, characterized in that, include: The objective function is constructed based on the first state of the target car at the current moment. The first state includes the position, speed and acceleration of the target car. The control signal is obtained by solving the objective function based on the MPC algorithm; The first state and control signals are input into the PINN model to obtain the second state of the target car at the next moment; The second state is fed back into the MPC algorithm to reacquire the control signal and traverse all cars in the queue to achieve real-time control of the car queue.

2. The vehicle queuing control method according to claim 1, characterized in that, The control signals include the accelerator and brake opening / closing angles of the vehicle.

3. The vehicle queuing control method according to claim 1, characterized in that, The objective function constructed based on the first state of the target vehicle includes: Obtain the following trajectory of the target car, the following trajectory including the position and speed of the vehicle preceding the target car; The following error is calculated based on the first state and the following trajectory. The objective function is constructed based on the following error; The objective function includes: in, J () represents the difference between the predicted state of the vehicle and the reference trajectory. This indicates the preset prediction range. k Indicates time, j Indicates the first j Target car R Indicates weight, u Indicates control signal, Indicates the first j Control signals for the target vehicle Indicates the first j The target car k+i Control signals at specific times, This indicates the following error.

4. The vehicle queuing control method according to claim 3, characterized in that, The MPC algorithm includes the following constraints: in, Indicates the lower bound of the control signal. Indicates the upper bound of the control signal. Indicates the first j The target car in the l- Control signal at time 1 Indicates the first j The position of the target vehicle at any given time within a preset range. Indicates the first j The speed of the target vehicle at any given time within a preset range. This indicates the preset safe distance between vehicles. This indicates the maximum number of vehicles following in the car queue.

5. The vehicle queuing control method according to claim 3, characterized in that, The method of solving the objective function based on the MPC algorithm to obtain the control signal includes: The MPC algorithm iteratively solves the objective function within a preset prediction range until the objective function reaches a global minimum, thereby obtaining a control signal sequence. Output the first control signal in the control signal sequence.

6. The vehicle queuing control method according to claim 5, characterized in that, The MPC algorithm iteratively solves the objective function within a preset prediction range, including: The MPC algorithm solves the objective function based on the LQR regulator.

7. The vehicle queuing control method according to claim 1, characterized in that, The step of inputting the first state and control signal into the PINN model to obtain the second state of the target vehicle at the next moment includes: An autoregressive prediction function is constructed based on the PINN model; The first state and the control signal are input into the autoregressive prediction function to obtain the second state; The autoregressive prediction functions include: in, Indicates the first j The target car in the l The state at any given moment, This indicates the following trajectory of the vehicle following the target car. Indicates the first j The target car in the l Location at any given moment Indicates the first j The target car in the l The speed of time Indicates the first j The target car in the l acceleration at any moment Indicates the first j The initial state of the target car at time 0. Indicates the first j The target car in the k Control signals at specific times, Indicates the first j PINN model of the target car Indicates the first j An autoregressive prediction model for a target vehicle. This represents hyperparameters.

8. A vehicle queuing control device, characterized in that, include: The control signal module is used to construct the objective function based on the first state of the target vehicle and solve for the control signal based on the MPC algorithm. The state prediction module is used to input the first state and control signals into the PINN model to obtain the second state of the target car at the next moment. The control module is used to feed back the second state to the MPC algorithm, reacquire the control signal, and traverse all cars in the queue to achieve real-time control of the car queue.

9. An electronic device, characterized in that, include: Memory, used to store programs; A processor, coupled to the memory, is configured to execute the program stored in the memory to implement the steps of the vehicle queuing control method according to any one of claims 1 to 7.

10. A storage medium, characterized in that, Used to store computer-readable programs or instructions, which, when executed by a processor, can implement the steps in the vehicle queuing control method according to any one of claims 1 to 7.

Images