Carry-scraper path tracking control method and device based on improved NMPC control strategy

Abstract

The invention discloses a carry-scraper path tracking control method and device based on an improved NMPC control strategy. The method comprises the steps that a kinematics control system model of an underground carry-scraper is established; according to the kinematic control system model, designing a non-linear model predictive controller (NMPC); dynamically optimizing time domain parameters of a nonlinear model predictive controller (NMPC) by using a particle swarm optimization (PSO) algorithm based on the path reference point; the optimized time domain parameters are transmitted into the non-linear model prediction controller NMPC, and the control quantity is output to act on a steering mechanism of the carry-scraper to complete path tracking. The method optimizes and improves the time domain parameters of the NMPC by using the particle swarm optimization algorithm, can be widely applied to the path tracking control process of the articulated vehicle, effectively avoids steering delay, snakelike displacement and the like of the vehicle, greatly improves the path tracking control precision, is particularly suitable for the control process of an underground carry-scraper, and has a wide application prospect. And reliable technical support is provided for unmanned driving of the carry-scraper.

Description

Technical Field

[0001] The present invention relates to the field of path tracking control of articulated vehicle robots, and more particularly to a path tracking control method and device for scraper robots based on an improved NMPC control strategy. Background Art

[0002] Currently, underground mining relies heavily on advanced mining equipment, the traditional operation of which mostly relies on manual control. With the rapid development of intelligent device technology, intelligent control has become a critical component of these devices. The mining industry is currently undergoing a trend of transitioning from manual operation to automation and intelligent control, a transformation crucial for improving the efficiency and safety of mining operations. Against this backdrop, the LHD (Land Loader)—an indispensable intelligent device in underground mining—and path-following control, at its core, are particularly important for research.

[0003] Unmanned driving is a prerequisite for unmanned mining, and path-following control is key to achieving autonomous operation for vehicles like loaders. While numerous studies have explored path-following control technologies, including pure tracking algorithms, PID control, linear quadratic regulator (LQR) control, and model predictive control (MPC) algorithms, most research is based on ground passenger vehicles, with limited research on articulated vehicles such as LHDs.

[0004] Among them, although new and effective path tracking methods (such as nonlinear model predictive control (NMPC)) have made significant progress in the research of articulated vehicles such as underground scrapers, they still need to be further studied. They have not been fully and seriously considered in both research and practical applications. Problems such as scrapers being unable to adjust the vehicle angle in time when turning on high-curvature paths have occurred, resulting in excessive deviation from the reference trajectory and collision with the roadway. These problems are particularly prominent when LHD uses an NMPC controller to track the reference path. Summary of the Invention

[0005] In light of this, there is currently a lack of systematic and in-depth research on key issues such as steering lag, prone to roadway wall collisions, and serpentine movement during scraper path tracking control. Therefore, the present invention provides a scraper path tracking control method and device based on an improved NMPC control strategy. This invention aims to further improve the NMPC control method and proposes a new controller optimization strategy that optimizes and improves the time-domain parameters of the NMPC using a particle swarm optimization (PSO) algorithm to achieve higher adaptability and better tracking results.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a path tracking control method for a scraper based on an improved NMPC control strategy, comprising the following steps:

[0008] S10. Establish a kinematic control system model for the underground scraper;

[0009] S20, designing a nonlinear model predictive controller (NMPC) according to the kinematic control system model;

[0010] S30, based on the path reference point, using the particle swarm optimization algorithm PSO to dynamically optimize the time domain parameters of the nonlinear model predictive controller NMPC;

[0011] S40 , transferring the optimized time domain parameters to the nonlinear model predictive controller NMPC, and outputting a control variable to act on the steering mechanism of the scraper to complete path tracking.

[0012] Furthermore, the step S10 includes:

[0013] S101. Establish a coordinate system and set χ = ​​[x, y, θ, γ]T and u = [v, ω]T. The kinematic control system model is expressed as:

[0014]

[0015] Where χ represents the vehicle state, u represents the vehicle control variable, x, y represent the horizontal and vertical coordinates of the LHD reference point, θ, γ represent the LHD heading angle and articulation angle, v, ω represent the LHD speed and angular velocity, respectively; f(·) represents the kinematic state transfer function;

[0016] S102. Use Euler integrator to discretize the continuous nonlinear system:

[0017] χ(k+1)=χ(k)+Tf(χ(k),Δu(k))

[0018] u(k)=u(k-1)+Δu(k)

[0019] Where T represents the discretized time step, Δu(k) represents the increment of the control input, and k represents the sequence number of the discretized time step, that is, the index value.

[0020] Furthermore, the step S20 includes:

[0021] S201, design a nonlinear model predictive controller NMPC whose objective function is:

[0022]

[0023] Among them, t represents the current time; represents the predicted state of the vehicle based on the current measured state; ref represents the reference state quantity; ref (t+k|t)=[x ref (t+k|t) y ref (t+k|t) θ ref (t+k|t) 0] T Indicates the status information of the reference path; N p represents the prediction time domain, N c represents the control time domain; Q represents the weight matrix of tracking error; R represents the weight matrix of control input increment; represents the terminal cost, and P represents the weight matrix of the terminal cost;

[0024] S202, NMPC optimization problem is solved by sequential quadratic programming algorithm, dealing with nonlinear constraints:

[0025] γ min ≤γ(t+k|t)≤γ max ,k=1,2,3,...,N p

[0026] v min ≤v(t+k|t)≤v max ,k=1,2,3,...,N c

[0027] ω min ≤ω(t+k|t)≤ω max ,k=1,2,3,...,N c

[0028] Among them, γ min and γ max Indicates the minimum and maximum values ​​of the physical limit of the articulation angle; v min 、v max Indicates the minimum and maximum longitudinal speed limits; ω min 、ω max Indicates the minimum and maximum values ​​for the steering rate limit.

[0029] Furthermore, the step S30 includes:

[0030] S301, initialize the particle swarm, each particle represents a set of prediction time domain and control time domain parameters;

[0031] S302, define the fitness function:

[0032]

[0033] Among them, N simRepresents the number of samples, which is also the cyclic ordinal number.

[0034] S303 , based on the path reference point, the nonlinear model predictive controller NMPC is inputted, and the particle position and velocity are updated iteratively to obtain the optimal time domain parameters.

[0035] Furthermore, the particle position and velocity are iteratively updated to obtain the optimal time domain parameters, including:

[0036] The update speed and v of the i-th particle i (t) Position x i (t) is expressed by the following equation:

[0037] v i (t+1)=wv i (t)+c1rand(p b -x i (t))+c2Rand(g b -x i (t))

[0038] x i (t+1)=x i (t)+v i (t+1)

[0039] Among them, v i (t) and x i (t) represents a vector in multidimensional space; p b , g b They represent the local optimal position and the global optimal position respectively; the particle inertia weight is represented by w; the particle cognitive acceleration and social acceleration are represented by c1 and c2, which are defined as positive constants; rand and Rand are two random parameters in [0, 1].

[0040] In a second aspect, an embodiment of the present invention further provides a scraper path tracking control device based on an improved NMPC control strategy, using the scraper path tracking control method based on the improved NMPC control strategy as described in any one of the first aspects, the device comprising:

[0041] Kinematic modeling module, used to establish the kinematic control system model of the underground scraper;

[0042] An NMPC controller module is used to design a nonlinear model predictive controller (NMPC) based on the kinematic control system model;

[0043] The PSO optimization module uses the particle swarm optimization algorithm (PSO) to dynamically optimize the time domain parameters of the nonlinear model predictive controller (NMPC) based on the path reference point;

[0044] The execution module transmits the optimized time domain parameters to the nonlinear model predictive controller NMPC, and outputs the control quantity to act on the steering mechanism of the scraper to complete the path tracking.

[0045] It can be seen from the above technical solutions that compared with the prior art, the present invention has the following technical advantages:

[0046] This paper aims to further improve the NMPC control method and proposes a new controller optimization strategy. By utilizing the particle swarm optimization algorithm (PSO) to optimize and improve the time-domain parameters of the NMPC, this approach provides a novel and improved nonlinear model predictive controller (PT-NMPC) to achieve greater adaptability and better tracking performance. This method can be widely applied to the path tracking control process of articulated vehicles, effectively avoiding vehicle steering delays and serpentine movements, significantly improving path tracking control accuracy. It is particularly suitable for the control process of underground scrapers, providing reliable technical support for unmanned scraper operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0048] Figure 1 This is a flow chart of the scraper path tracking control method based on the improved NMPC control strategy provided by the present invention.

[0049] Figure 2 This is a flow chart of particle swarm parameter optimization provided by the present invention.

[0050] Figure 3 Schematic diagram of the nonlinear model predictive control strategy based on time domain optimization provided by the present invention.

[0051] Figure 4 This is a block diagram of the scraper path tracking control device based on the improved NMPC control strategy provided by the present invention. DETAILED DESCRIPTION

[0052] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0053] The embodiment of the present invention discloses a path tracking control method for a scraper based on an improved NMPC control strategy, comprising the following steps:

[0054] S10. Establish a kinematic control system model for the underground scraper;

[0055] S20, designing a nonlinear model predictive controller (NMPC) according to the kinematic control system model;

[0056] S30, based on the path reference point, using the particle swarm optimization algorithm PSO to dynamically optimize the time domain parameters of the nonlinear model predictive controller NMPC;

[0057] S40 , transferring the optimized time domain parameters to the nonlinear model predictive controller NMPC, and outputting a control variable to act on the steering mechanism of the scraper to complete path tracking.

[0058] This invention establishes and discretizes the kinematic model of the scraper and designs a NMPC controller with terminal constraints. It innovatively introduces a particle swarm optimization (PSO) algorithm to dynamically optimize the prediction and control time domains, evaluating tracking error and control cost by inputting path reference points. The optimal values ​​in the control and prediction time domains are obtained, and these optimal values ​​are fed into the NMPC to generate control instructions, achieving path tracking control. This method can address steering delay issues during high-curvature turns, reduce lateral tracking error, eliminate serpentine trajectories, and improve control smoothness. It is suitable for use in narrow environments such as underground tunnels, enhancing the safety of unmanned driving.

[0059] The present invention is effectively applied to the path tracking control process of underground scraper. Specifically, referring to Figure 1 As shown, the following steps are included:

[0060] Step 1: Establish the kinematic control system model of the scraper: Establish a suitable coordinate system. During the operation of the scraper, the front vehicle is used as the reference point. At this time, x = x f , y=y f .

[0061] Assume χ = [x, y, θ, γ] T and u=[v,ω] T ,The system kinematics can be simply expressed as:

[0062]

[0063] Among them, χ represents the vehicle state, u represents the vehicle control quantity, and x f and y fare the horizontal and vertical coordinates of the front body of the scraper, x and y represent the horizontal and vertical coordinates of the scraper reference point, v and ω represent the scraper velocity and angular velocity, respectively; θ and γ represent the scraper heading angle and articulation angle, respectively. f(·) represents the kinematic state transfer function.

[0064] In order to facilitate the implementation of NMPC on a computer, the present invention uses the Euler integrator to discretize the continuous nonlinear system:

[0065] χ(k+1)=χ(k)+Tf(χ(k),Δu(k))

[0066] u(k)=u(k-1)+Δu(k)

[0067] Where T is the discretized time step, and Δu(k) is the increment of the control input, which ensures that the constraints are maintained during the solution. k represents the index of the discretized time step.

[0068] Step 2: Design a nonlinear model predictive controller: The NMPC controller aims to ensure that the vehicle can accurately follow the desired trajectory. Therefore, the error between the predicted state and the reference path should be as small as possible. In addition, smooth control must be guaranteed, which means that changes in the control input should be minimized. In the controller design of this invention, a sequential quadratic programming algorithm is introduced to address the possible impact of nonlinear factors on the convex optimization problem. The objective function of the NMPC controller can be expressed as:

[0069]

[0070] Among them, t represents the current time, represents the predicted state of the vehicle based on the current measurement state, χ ref (t+k|t)=[x ref (t+k|t) y ref (t+k|t) θ ref (t+k|t) 0] T is the state information of the reference path, N p is the prediction time domain, N c is the control time domain. Q is the weight matrix of the tracking error, and R is the weight matrix of the control input increment, both of which are positive definite matrices.

[0071] In order to achieve higher tracking accuracy and improve the stability of the closed-loop system, a terminal cost is proposed based on the above NMPC controller as shown below:

[0072]

[0073] Among them, χ refrepresents the reference state quantity, and P is the weight matrix of the terminal cost.

[0074] In general, the path tracking problem can be transformed into an MPC program, such as the following optimization problem:

[0075]

[0076]

[0077]

[0078] u(t|t)=u(t-1|t)+Δu(t|t)

[0079] γ min ≤γ(y+k|t)≤γ max ,k=1,2,3,...,N p

[0080] v min ≤v(t+k|t)≤v max ,k=1,2,3,...,N c

[0081] ω min ≤ω(t+k|t)≤ω max ,k=1,2,3,...,N c

[0082] where γ min and γ max is the physical limit of the articulation angle, v min 、v max is the longitudinal speed limit, ω min 、ω max is the turning rate limit.

[0083] After the optimal solution is obtained, a series of control outputs can be obtained as follows:

[0084] Δu * (t)=[Δu * (t|t),Δu * (t+1|t),…,Δu * (t+N c |t)] T

[0085] Applying the first group to vehicles, it can be expressed as follows:

[0086] u(t)=Δu * (t|t)+u(t-1)

[0087] Step 3: Improved nonlinear model predictive control strategy using PSO: The NMPC controller encounters a series of challenges in the actual operation of the scraper, especially when turning on a high-curvature path, it is easy to fail to adjust the vehicle angle in time, resulting in a large deviation from the reference trajectory and collision with the roadway. These problems are particularly prominent when using the NMPC controller to track the reference path of the LHD. For NMPC, the prediction domain refers to the future time range considered by the controller during the optimization process, while the control domain is the time range in which the controller actually applies the control input. Designing an efficient NMPC controller is a challenging task, and the optimization algorithm provides an efficient and accurate solution that can efficiently optimize the time domain. The present invention proposes to improve the efficiency of LHD in underground roadway path tracking by optimizing the prediction domain and control domain of the NMPC controller. The particle swarm optimization (PSO) algorithm can be used to find the optimal values ​​of these two parameters, so that the LHD can more effectively achieve its path tracking goals.

[0088] PSO optimization: Particle swarm optimization algorithm was first proposed, and the process can be referred to Figure 2 , has been successfully applied to solve problems with nonlinearity, nondifferentiability, and multiple optimality. Compared with other random methods, PSO can generate high-quality solutions with more stable and faster convergence characteristics, and has a shorter calculation time. For the standard PSO at time t, the update speed of the i-th particle and v i (t) Position x i (t) is expressed by the following equation:

[0089] v i (t+1)=wv i (t)+c1rand(p b -x i (t))+c2Rand(g b -x i (t))

[0090] x i (t+1)=x i (t)+v i (t+1)

[0091] where v i (t) and x i (t) is a vector in multidimensional space. p b , g b Denote the local optimal position and the global optimal position, respectively. The particle inertia weight is represented by w. The particle cognitive acceleration and social acceleration are represented by c1 and c2, which are defined as positive constants. rand and Rand are two random parameters in [0, 1].

[0092] NMPC control strategy based on time domain optimization (PT-NMPC): Reference Figure 3 Optimizing the prediction and control time domains in NMPC using the PSO algorithm is an effective method for improving control system performance. The PSO algorithm can adjust these NMPC parameters, thereby improving the tracking performance of LHD devices on different paths, especially in terms of efficient and accurate tracking of reference paths with large curvature radius and timely turning. Furthermore, the PSO algorithm's adaptive adjustment capabilities enable the NMPC controller to dynamically adjust to the system's real-time requirements.

[0093] First, the PSO algorithm simulates swarm intelligence behavior. Based on the state and control variables after the initial scraper path tracking control, each particle is assigned prediction and control time domain parameters. Next, the fitness of each particle is evaluated based on system performance metrics such as tracking error and control cost. The individual optimal value obtained by each particle is then passed to the NMPC for scraper path tracking control, and the local and global optimal solutions are recorded. During the iteration process, the PSO algorithm leverages information sharing between particles to dynamically adjust particle positions and velocities to approximate the optimal prediction and control time domain parameters. This approach enables the NMPC system to adapt to diverse dynamic changes and environmental conditions, improving tracking accuracy and robustness. Finally, simulations and experiments verify that the NMPC method combined with PSO significantly improves trajectory tracking accuracy compared to traditional methods. By adjusting time domain parameters to adapt to complex or changing environments, it achieves superior control results in practical applications.

[0094] In this invention, the fitness function we designed is mainly based on the sum of the squares of the system state deviation and the control quantity to reflect the degree of deviation of the system state and the control cost. After considering the state quantity and other relevant parameters during the LHD operation process, as well as the requirements of the objective function, the present invention constructs a suitable fitness function as follows:

[0095]

[0096] Among them, N sim It is the number of samples and also the cyclic ordinal number.

[0097] The PSO algorithm can also handle nonlinear optimization problems in NMPC, including constraints, by using the sequential quadratic programming (SQP) algorithm to improve the efficiency and robustness of solving non-convex problems and ensure that the global optimal solution is obtained in convex problems. Within each sampling period, the PSO-optimized NMPC controller can solve the constrained optimization problem, which involves minimizing the tracking error and the change of the control input while ensuring smooth control. This approach allows the controller to update the control input in each iteration to respond to immediate changes in the system.

[0098] In summary, the application of PSO algorithm in NMPC provides a flexible and effective method to optimize the prediction time domain and control time domain, significantly improving the performance and adaptability of the control system.

[0099] Step 4: Based on the pre-set local segment reference path of the real tunnel, the path reference point is passed to the nonlinear model predictive controller to perform the initial path tracking task, and the path tracking error and control cost of the scraper in this round are calculated. The optimal value of each particle is obtained through particle swarm iterative update optimization, and the local optimal value is finally obtained through the cycle.

[0100] Step 5: Obtain the optimal value of the time domain parameters: Under the PT-NMPC control strategy, the particle swarm continuously iterates to obtain the local optimum according to the state quantity of the scraper's multiple path tracking tasks. By comparing the fitness values, the optimal time domain parameters in the scraper control process are finally obtained, that is, the global optimal value of the particle swarm.

[0101] Step 6: The optimal time domain parameters are passed into the controller to enable the scraper to complete the path tracking task: After step 5, the global optimal time domain parameters are obtained, which are passed into the NMPC controller, and then the optimal control quantity is output. After receiving it, the scraper performs a high-precision path tracking control task.

[0102] The above steps solve the problems of steering delay and return lag in underground scrapers with large curvature turning radius, which lead to easy collision with the tunnel, and greatly improve the accuracy of the scraper's path tracking control.

[0103] The following is an extended example of the technical solution of the present invention, using a specific underground scraper operation scenario to explain in detail the implementation process and effects of PT-NMPC (nonlinear model predictive control based on particle swarm optimization). This example is based on a 90° sharp bend tunnel environment, and the scenario is set as follows:

[0104] Laneway structure:

[0105] Straight track length: 20m → 90° sharp bend (radius of curvature R = 4m) → straight track length: 15m;

[0106] Tunnel width: 3.5m (the width of the scraper is 2.8m, and the safety margin is only 0.35m / side);

[0107] Scraper parameters:

[0108] Articulation angle range: Maximum turning rate: ω max =15° / S; running speed: v=1.5m / s;

[0109] Control objective: lateral tracking error <0.2m, avoiding serpentine trajectory and collision.

[0110] The PT-NMPC control process is as follows:

[0111] Step 1: Establish kinematic model;

[0112] Step 2: Initialize the NMPC controller; fix the time domain parameters (traditional method): N p =20 (predict 2 seconds), N c =10 (control 1 second);

[0113] Step 3: PSO dynamically optimizes time domain parameters

[0114] PSO parameter settings:

[0115] parameter value illustrate Number of particles 30 Solution space coverage > 95% ω 0.8→0.4 linear decay Balancing global / local search <![CDATA[c1和c2]]> 1.5,1.5 Acceleration constant Number of iterations 15 Real-time guarantee (<0.2 seconds)

[0116] According to the definition of fitness function, determine N sim =10: Evaluate the performance within the next 1 second.

[0117] Optimization process:

[0118] When the scraper is 5m away from the curve entrance (t = 13.5s);

[0119] 1. PSO initializes the particle swarm:

[0120] Particle 1: (N p =15,N c =8);

[0121] Particle 2: (N p =15,N c =8) ...

[0123] 2. Each particle is passed into the NMPC simulation:

[0124] Particle 1 → prediction 1.5 seconds → maximum lateral error 0.25m;

[0125] Particle 2 → prediction time 0.8 seconds → maximum lateral error 0.12m (optimal);

[0126] 3. Update the global optimum: (N p =8,N c =4);

[0127] Step 4: PT-NMPC control effect

[0128]

[0129] It can be seen from the above embodiments that:

[0130] Traditional NMPC Implementation: Due to fixed prediction and control time-domain parameters, the traditional NMPC controller was unable to adjust the vehicle's turning angle in a timely manner on high-curvature turns. This resulted in significant deviations between the scraper's actual driving path and the reference trajectory, and significant lateral tracking errors, with a maximum error of 0.5186 meters. Stability: The vehicle exhibited snaking during turns, with large speed fluctuations and unstable angular velocity changes, which were prone to problems such as self-centering lag, affecting driving stability.

[0131] Implementation of the PT-NMPC system: By dynamically optimizing prediction and control parameters using a particle swarm optimization (PSO) algorithm, the PT-NMPC system can more precisely control the scraper's steering mechanism, significantly reducing lateral tracking error in high-curvature turns to within 0.1355 meters, effectively avoiding collisions with tunnel walls. Stability: The optimized controller ensures smoother vehicle travel, more uniform changes in speed and angular velocity, reduces snaking and lag, and improves driving comfort and safety.

[0132] Based on the same inventive concept, the embodiment of the present invention also provides a scraper path tracking control device based on an improved NMPC control strategy. Since the principle of the problem solved by the device is similar to that of the aforementioned scraper path tracking control method based on the improved NMPC control strategy, the implementation of the device can refer to the implementation of the aforementioned method, and the repeated parts will not be repeated. Figure 4 As shown, the device includes:

[0133] Kinematic modeling module, used to establish the kinematic control system model of the underground scraper;

[0134] An NMPC controller module is used to design a nonlinear model predictive controller (NMPC) based on the kinematic control system model;

[0135] The PSO optimization module uses the particle swarm optimization algorithm (PSO) to dynamically optimize the time domain parameters of the nonlinear model predictive controller (NMPC) based on the path reference point;

[0136] The execution module transmits the optimized time domain parameters to the nonlinear model predictive controller NMPC, and outputs the control quantity to act on the steering mechanism of the scraper to complete the path tracking.

[0137] The present invention solves the problems of the scraper's steering delay, which easily causes collision with the tunnel, and serpentine movement resulting in poor path tracking effect under the NMPC controller, and achieves high path tracking control accuracy. It is an efficient and robust solution.

[0138] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0139] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A path tracking control method for scraper based on an improved NMPC control strategy is characterized by: The following steps are involved: S10. Establish a kinematic control system model for the underground scraper; S20, designing a nonlinear model predictive controller (NMPC) according to the kinematic control system model; S30, based on the path reference point, using the particle swarm optimization algorithm PSO to dynamically optimize the time domain parameters of the nonlinear model predictive controller NMPC; S40 , transferring the optimized time domain parameters to the nonlinear model predictive controller NMPC, and outputting a control variable to act on the steering mechanism of the scraper to complete path tracking.

2. The path tracking control method for scraper based on improved NMPC control strategy according to claim 1, characterized in that: The step S10 includes: S101. Establish a coordinate system and set χ = ​​[x, y, θ, γ] T and u=[v,ω] T , then the kinematic control system model is expressed as: Where χ represents the vehicle state, u represents the vehicle control variable, x, y represent the horizontal and vertical coordinates of the LHD reference point, θ, γ represent the LHD heading angle and articulation angle, v, ω represent the LHD speed and angular velocity, respectively; f(·) represents the kinematic state transfer function; S102. Use Euler integrator to discretize the continuous nonlinear system: χ(k+1)=χ(k)+Tf(χ(k),Δu(k)) u(k)=u(k-1)+Δu(k) Where T represents the discretized time step, Δu(k) represents the increment of the control input, and k represents the sequence number of the discretized time step, that is, the index value.

3. The path tracking control method for scraper based on improved NMPC control strategy according to claim 2, characterized in that: The step S20 includes: S201, design a nonlinear model predictive controller NMPC whose objective function is: Among them, t represents the current time; represents the predicted state of the vehicle based on the current measured state; ref represents the reference state quantity; ref (t+k|t)=[x ref (t+k|t)y ref (t+k|t)θ ref (t+k|t)0] T Indicates the status information of the reference path; N p represents the prediction time domain, N c represents the control time domain; q represents the weight matrix of tracking error; R represents the weight matrix of control input increment; represents the terminal cost, and P represents the weight matrix of the terminal cost; S202, NMPC optimization problem is solved by sequential quadratic programming algorithm, dealing with nonlinear constraints: γ min ≤γ(t+k|t)≤γ max ,k=1,2,3,...,N p v min ≤v(t+k|t)≤v max ,k=1,2,3,...,N c ω min ≤ω(t+k|t)≤ω max ,k=1,2,3,...,N c Among them, γ min and γ max Indicates the minimum and maximum values ​​of the physical limit of the articulation angle; v min 、v max Indicates the minimum and maximum longitudinal speed limits; ω min 、ω max Indicates the minimum and maximum values ​​for the steering rate limit.

4. The path tracking control method for scraper based on improved NMPC control strategy according to claim 3 is characterized in that: The step S30 includes: S301, initialize the particle swarm, each particle represents a set of prediction time domain and control time domain parameters; S302, define the fitness function: Among them, N sim Represents the number of samples, which is also the cyclic ordinal number. S303 , based on the path reference point, the nonlinear model predictive controller NMPC is inputted, and the particle position and velocity are updated iteratively to obtain the optimal time domain parameters.

5. The path tracking control method for scraper based on improved NMPC control strategy according to claim 4, characterized in that: Iteratively update particle positions and velocities to obtain optimal time domain parameters, including: The update speed and v of the i-th particle i (t) Position x i (t) is expressed by the following equation: v i (t+1)=wv i (t)+c1rand(p b -x i (t))+c2Rand(g b -x i (t)) x i (t+1)=x i (t)+v i (t+1) Among them, v i (t) and x i (t) represents a vector in multidimensional space; p b , g b They represent the local optimal position and the global optimal position respectively; the particle inertia weight is represented by w; the particle cognitive acceleration and social acceleration are represented by c1 and c2, which are defined as positive constants; rand and Rand are two random parameters in [0, 1].

6. The scraper path tracking control device based on the improved NMPC control strategy is characterized by: The scraper path tracking control method based on the improved NMPC control strategy according to any one of claims 1 to 5 is used, and the device includes: Kinematic modeling module, used to establish the kinematic control system model of the underground scraper; An NMPC controller module is used to design a nonlinear model predictive controller (NMPC) based on the kinematic control system model; The PSO optimization module uses the particle swarm optimization algorithm (PSO) to dynamically optimize the time domain parameters of the nonlinear model predictive controller (NMPC) based on the path reference point; The execution module transmits the optimized time domain parameters to the nonlinear model predictive controller NMPC, and outputs the control quantity to act on the steering mechanism of the scraper to complete the path tracking.

Images