Model prediction unmanned mine car path tracking control method considering delay compensation

Abstract

The invention discloses a model prediction unmanned mine car path tracking control method considering delay compensation. The method comprises the following steps: establishing a path tracking error dynamic model of an unmanned mine car; state augmentation is carried out on the path tracking error dynamic model, communication delay between a vehicle controller and an actuator and response hysteresis of a steering actuator are represented as new state variables, and an augmented state space model reflecting system delay characteristics is obtained; constructing a prediction model of model prediction control based on the augmented state space model, and designing an objective function and constraint conditions to convert a path tracking control problem into a quadratic programming problem; and solving the quadratic programming problem on line, and acting the obtained optimal control quantity on a mine car steering system. Defects of a traditional external delay compensation mode are overcome fundamentally, and path tracking precision and control stability of the unmanned mine car in a complex mine environment are improved.

Description

Technical Field

[0001] This invention relates to the field of unmanned mining truck technology, and in particular to a model prediction-based path tracking control method for unmanned mining trucks that takes into account delay compensation. Background Technology

[0002] Open-pit mines face complex production environments and high-intensity operations. Traditional manually driven mining trucks suffer from high labor costs, significant safety risks, and low handover efficiency. With the advancement of smart mine construction, unmanned driving technology has become an important way to solve these problems. Path tracking control, as a core component of unmanned driving systems, directly affects the safety, stability, and accuracy of mining truck operation.

[0003] However, due to their large size, heavy weight, and high inertia, mining trucks suffer from communication delays and sluggish actuator response during actual operation. In particular, the hydraulic actuators of the steering system respond slowly, which seriously affects the accuracy and real-time performance of path tracking.

[0004] Currently, path tracking methods based on Model Predictive Control (MPC) are widely used in the field of autonomous driving. However, they are usually based on ideal assumptions and do not fully consider the impact of system delay on control performance. Existing technologies often employ external compensation strategies, such as adding local controllers or introducing lead compensation mechanisms, to mitigate control deviations caused by delay. However, these methods can only make coarse corrections from outside the system and cannot fundamentally model the delay dynamics, leading to problems such as non-optimal control variables, complex parameter tuning, and poor adaptability. Especially in highly dynamic and unstructured environments such as mines, the control accuracy and robustness of traditional methods are insufficient to meet the needs of practical engineering.

[0005] Therefore, there is an urgent need for a high-precision path tracking control method that can inherently handle communication and execution delays in order to improve the tracking performance and control reliability of unmanned mining vehicles under complex working conditions. Summary of the Invention

[0006] This invention aims to solve the problem of precise path tracking control for mining trucks, so that the lateral control accuracy of the mining trucks can reach a technical specification of no more than 0.5m.

[0007] To address the aforementioned technical problems, this invention provides a model-based prediction-based path tracking control method for unmanned mining trucks that considers delay compensation, comprising the following steps: Establish a path tracking error dynamic model for unmanned mining vehicles; The path tracking error dynamics model is augmented with a state, and the communication delay between the vehicle controller and the actuator and the response hysteresis of the steering actuator are characterized as new state variables, resulting in an augmented state space model that reflects the system delay characteristics. Based on the augmented state-space model, a predictive model for model predictive control is constructed, and an objective function and constraints are designed to transform the path tracking control problem into a quadratic programming problem. The quadratic programming problem is solved online, and the obtained optimal control quantity is applied to the mine car steering system.

[0008] Furthermore, prior to constructing the prediction model, the following steps are also included: Discretize the augmented state-space model to obtain the state-space equations of the discrete-time system.

[0009] Furthermore, the process of constructing the prediction model includes: Based on the state-space equation of the discrete-time system, a prediction equation for the future state of the system with respect to the current state and future control input is derived in the prediction time domain.

[0010] Furthermore, the state augmentation of the path tracking error dynamics model specifically includes: Based on the analysis of the response characteristics of the steering system, the response hysteresis of the steering actuator is simplified to a first-order inertial element; real-time steering angle is introduced into the state variables for the first augmentation, resulting in the first augmented state space model. The communication delay between the vehicle controller and the actuator is modeled as a pure delay element, and its delayed instruction is used as a new state variable to perform a second augmentation on the first augmented state space model to obtain the augmented state space model.

[0011] Furthermore, the first augmentation step includes: The steering angle command issued by the vehicle controller is defined as the system input; The real-time turning angle and its derivative relationship are introduced into the state equation of the path tracking error dynamics model to form the first augmented state space model containing the original state variables and the real-time turning angle state.

[0012] Furthermore, the second augmentation step includes: The communication delay between the vehicle controller and the actuator is defined as an integer multiple of the system sampling period, d. The d historical control commands prior to the current time are augmented into new state variables, and the state-space equation of the discrete-time system is transformed into the augmented state-space model in a delay-free form.

[0013] Furthermore, the objective function is:

[0014] In the formula, Indicates the number of steps. To predict the time domain, To control the time domain, , The weight matrix is ​​positive definite. These are the weighting coefficients. As a relaxation factor, Indicates the angle increment. Indicates the first One cycle, This represents the state vector.

[0015] Furthermore, the constraints include at least one of steering angle constraints and steering angle change rate constraints, and the constraints are softened by introducing the relaxation factor to ensure the feasibility of the optimization problem.

[0016] Furthermore, if the quadratic programming problem fails to be solved online, the controller will use the control quantity from the previous sampling time as the control quantity for the current time.

[0017] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: This invention internalizes system communication delay and steering actuator response hysteresis, representing them as state variables in an augmented state-space model. This constructs a predictive model that accurately reflects the system's true dynamics, and then uses this model for model predictive control optimization. This fundamentally overcomes the shortcomings of traditional external delay compensation methods, improving the path tracking accuracy and control stability of unmanned mining vehicles in complex mining environments. Lateral errors can be controlled within 0.5 meters, while simultaneously reducing the system's dependence on additional compensation parameter calibration, enhancing the algorithm's robustness and engineering applicability. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is the overall flowchart disclosed in this invention; Figure 2 This invention discloses a two-degree-of-freedom dynamic model for a mining truck. Figure 3 This is a flowchart of the online solution to the quadratic programming problem disclosed in this invention. Detailed Implementation

[0020] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] This invention aims to provide a model predictive path tracking control method for unmanned mining trucks that considers delay compensation. By performing two state augmentations, the system communication delay and actuator response hysteresis are internalized into the predictive model, and a predictive model that accurately reflects the real dynamics of the system is constructed. Based on this model, model predictive control optimization is performed to solve the shortcomings of traditional external delay compensation methods.

[0022] Please see Figure 1 A model-based path tracking control method for unmanned mining trucks that considers delay compensation mainly includes the following steps: S1. Establish a path tracking error dynamic model for unmanned mining vehicles.

[0023] In solving the path tracking problem of unmanned mining trucks, a dynamic model of the mining truck using state variables such as position and direction errors relative to the road surface is very effective. Establishing a path tracking error dynamic model derived from the two-degree-of-freedom dynamic model of the mining truck allows for the mathematical expression of the vehicle's tracking error state variables and the applied control variables.

[0024] In this scheme, the state variable is defined as follows: ,in, These represent lateral error, heading angle error, rate of change of lateral error, and rate of change of heading angle error, respectively. According to Figure 2 The state-space representation of the tracking error variables for the shown mine car dynamics model is as follows: (1) in, Defined system state variables; This is the corner input, i.e., the system control quantity; The curvature at the reference path projection point is the measurable disturbance of the system. , , In the formula These represent the lateral stiffness of the front wheel of the mine car, the lateral stiffness of the rear wheel, the longitudinal speed of the mine car, the mass of the mine car, the yaw moment of inertia of the mine car, the distance from the center of mass of the mine car to the front axle, and the distance from the center of mass to the rear axle, respectively.

[0025] S2. The path tracking error dynamics model is augmented with state variables. The communication delay between the vehicle controller and the actuator and the response hysteresis of the steering actuator are represented as new state variables, resulting in an augmented state-space model that reflects the system delay characteristics.

[0026] In a further proposed approach, the path tracking error dynamics model is augmented with state extensions, specifically including: S21. In reality, the actuators of a vehicle steering system exhibit response delays. Based on the analysis of the steering system's response characteristics, the response hysteresis of the steering actuator is simplified to a first-order inertial element; real-time steering angle is introduced into the state variables for the first augmentation, resulting in the first augmented state-space model.

[0027] In this scheme, the first augmentation steps include: Define the steering angle command issued by the vehicle controller as the system input; introduce the real-time steering angle and its derivative relationship into the state equation of the path tracking error dynamics model to form the first augmented state space model containing the original state variables and the real-time steering angle state.

[0028] Specifically, based on the state-space equations, real-time rotation angles are introduced into the state variables. Based on the analysis of the steering system, the steering angle command Represented as: (2) The state space will be expanded for the first time as follows: .

[0029] Combining equation (1), we get: (3) To facilitate controller design, discretization methods such as zero-order hold or forward Euler can be used to fix the time step of the state-space equation (3) of the continuous system. Transformed into the state-space equations of a discrete system: (4) In the formula, These are the coefficient matrices of a discrete system considering the effects of steering hysteresis. Indicates the number of steps.

[0030] S22. The communication delay between the vehicle controller and the actuator is modeled as a pure delay element, and its delayed instruction is used as a new state variable to augment the first augmented state space model a second time to obtain the second augmented state space model (the second augmented state space model is the augmented state space model obtained in step S2, and is named the second augmented state space model here in order to be consistent with the first augmented state space model).

[0031] In this scheme, the second augmentation step includes: The communication delay between the vehicle controller and the actuator is defined as an integer multiple of the system sampling period d; the d historical control commands before the current time are augmented into new state variables, and the state-space equation of the discrete-time system is transformed into an augmented state-space model in a delay-free form.

[0032] Specifically, due to the communication link between the mine car domain controller and the steering system actuator, a pure delay occurs between the controller issuing the cornering command and the actuator receiving the cornering command. Referring to the optimal control theory of discrete-time delay systems, by performing a second augmentation on the system's state vector and using the delay command as a new state variable, the state-space expression of the discrete-time delay-free system is obtained, as shown in formula (5). Here, to simplify the formula derivation, the input delay is considered to be a fraction of the system sampling period. Times: (5) make The state-space model that comprehensively considers input delay and turn delay can be simplified as follows: (6) In the formula, The coefficient matrix of a discrete system considering the effects of input delay and steering hysteresis.

[0033] S3. Construct a predictive model for model predictive control based on an augmented state-space model, and design the objective function and constraints to transform the path tracking control problem into a quadratic programming problem.

[0034] In this scheme, before constructing the prediction model, the augmented state-space model is discretized to obtain the state-space equation of the discrete-time system.

[0035] The process of constructing the prediction model involves deriving the prediction equation for the future state of the system with respect to the current state and future control inputs within the prediction time domain, based on the state-space equation of the discrete-time system.

[0036] Specifically, the prediction model is as follows: (7) In the formula, To predict the time domain, To control the time domain, the other matrices are defined as follows:

[0037] , ,

[0038]

[0039]

[0040] Based on the above prediction model, the following objective function is designed: (8) In the formula, Indicates the number of steps. To predict the time domain, To control the time domain, , The weight matrix is ​​positive definite. These are the weighting coefficients. As a relaxation factor, Indicates the angle increment. Indicates the first One cycle, This represents the state vector.

[0041] To transform the original problem into a quadratic programming optimization problem, the objective function needs to be transformed into a standard quadratic form: (9) In the formula, , , It is a quadratic Hessian matrix. It is the gradient vector of the quadratic form.

[0042] The final quadratic optimization problem can be expressed as:

[0043]

[0044]

[0045]

[0046]

[0047] At this point, the secondary optimization problem has been constructed.

[0048] In a further embodiment, the constraints include at least one of steering angle constraints and steering angle change rate constraints, and the constraints are softened by introducing the relaxation factor to ensure the feasibility of the optimization problem.

[0049] S4. Solve the quadratic programming problem online and apply the obtained optimal control quantity to the mine car steering system.

[0050] like Figure 3 As shown, solving the above quadratic optimization problem yields the open-loop optimal solution. .Pick The first component Then It acts on the vehicle system.

[0051] When solving a quadratic optimization problem, there may be cases where the solution fails. In such cases, the controller will use the corner input from the previous time step as the corner input for the current time step, meaning the corner increment for the current time step is 0.

[0052] This invention directly considers the delay characteristics in the form of a state vector within the state space representation, and constructs an optimal control problem based on this state space representation. This allows for the direct acquisition of the optimal control quantity for the system, thus solving the problem of excessively large or small control quantities caused by external compensation methods. It omits the process of addressing system delay through external compensation, saves on the work of calibrating relevant parameters, and ensures optimal control quantity.

[0053] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A model predictive unmanned mine vehicle path following control method considering delay compensation, characterized in that, The method comprises the following steps: establishing a path tracking error dynamics model of the unmanned mine vehicle; state-augmenting the path tracking error dynamics model, characterizing the communication delay between the vehicle controller and the actuator and the response delay of the steering actuator as new state variables to obtain an augmented state space model reflecting the delay characteristics of the system; constructing a prediction model of model predictive control based on the augmented state space model, and designing an objective function and constraint conditions to convert the path tracking control problem into a quadratic programming problem; solving the quadratic programming problem online and applying the obtained optimal control quantity to the steering system of the mine vehicle. 2.The model predictive path tracking control method for delay-compensated unmanned mine vehicle according to claim 1, wherein, Before the prediction model is constructed, the method further comprises: discretizing the augmented state space model to obtain a state space equation of a discrete-time system. 3.The model predictive path tracking control method for delay-compensated unmanned mine vehicle according to claim 2, characterized in that, The construction process of the prediction model comprises: according to the state space equation of the discrete-time system, deriving a prediction equation of the future state of the system with respect to the current state and future control input within a prediction time domain. 4.The model predictive path tracking control method for delay-compensated unmanned mine vehicle according to claim 2, wherein, The state-augmenting of the path tracking error dynamics model specifically comprises: based on analysis of the response characteristics of the steering system, simplifying the response delay of the steering actuator as a first-order inertia link; introducing a real-time steering angle into the state variables to perform first augmentation to obtain a first augmented state space model; modeling the communication delay between the vehicle controller and the actuator as a pure delay link, and taking the delayed command thereof as a new state variable to perform second augmentation to the first augmented state space model to obtain the augmented state space model. 5.The model predictive path tracking control method for delay-compensated unmanned mine vehicle according to claim 4, characterized in that, The first augmentation step comprises: defining the steering angle command issued by the vehicle controller as a system input; introducing the real-time steering angle and its derivative relationship into the state equation of the path tracking error dynamics model to form the first augmented state space model containing the original state variables and the real-time steering angle state. 6.The model predictive path tracking control method for delay-compensated unmanned mine vehicle according to claim 4, wherein, The second augmentation step comprises: defining the communication delay between the vehicle controller and the actuator as an integer d times of the system sampling period; augmenting a total of d historical control commands before the current time as new state variables, and converting the state space equation of the discrete-time system into a delay-free form of the augmented state space model.

7. The model predictive unmanned mine truck path following control method with delay compensation taken into account according to claim 1, characterized in that, The objective function is: wherein denotes the step number, is the prediction horizon, is the control horizon, , is a positive definite weight matrix, is a weight coefficient, is a relaxation factor, denotes the angle increment, denotes the th period, denotes the state vector. 8.The model predictive path tracking control method for delay-compensated unmanned mine vehicle according to claim 7, wherein, The constraint conditions include at least one of a steering angle constraint and a steering angle change rate constraint, and the constraint conditions are softened by introducing the relaxation factor to ensure the feasibility of the optimization problem. 9.The model predictive path tracking control method for delay-compensated unmanned mine vehicle according to claim 1, wherein, If the online solution of the quadratic programming problem fails, the controller will use the control quantity at the last sampling time as the control quantity at the current time.

Images