A monotonically convergent iterative learning control method for vehicle systems based on full output feedback

By adopting the iterative learning control method with full output feedback in the vehicle system, designing the iterative learning control gain matrix and updating the control input, the problem that traditional methods cannot make the output trajectory of the vehicle system monotonically converge, and high-precision desired trajectory tracking is achieved.

CN115586723BActive Publication Date: 2024-12-06XIDIAN UNIV
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202211297316.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-10-21
Publication Date
2024-12-06
Estimated Expiration
2042-10-21

AI Technical Summary

Technical Problem

The traditional iterative learning control method is difficult to make the actual output trajectory of a vehicle system with time-varying characteristics monotonically converge to the desired trajectory, and the optimal control effect cannot be achieved.

Method used

The monotonic convergence iterative learning control method of vehicle system based on full output feedback is adopted. By setting the desired trajectory and designing the iterative learning control gain matrix of the iterative learning control law of the full output feedback iterative learning control law, the tracking error and output deviation are obtained, and the control input is updated based on these data. Until the tracking error converges monotonically to 0.

Benefits of technology

The actual output trajectory of the vehicle system is realized to converge monotonically to the desired trajectory, overcome the influence of time-varying characteristics, and improve the control accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115586723B_ABST
    Figure CN115586723B_ABST
Patent Text Reader

Abstract

The present invention proposes a vehicle system monotonically convergent iterative learning control method based on full output feedback, including: obtaining the desired trajectory of the vehicle system; obtaining the control gain of the vehicle system; obtaining the position of the vehicle system in the current iteration, thereby calculating the tracking error of the vehicle system in the current iteration; updating the control input of the vehicle system in the next iteration according to the control input, tracking error and control gain of the vehicle system in the current iteration, thereby controlling the vehicle system. The present invention can make the actual output trajectory of the vehicle system monotonically converge to the desired trajectory, and the control accuracy is relatively high.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of automatic control technology, to vehicle automatic driving control, and in particular to a vehicle system monotonic convergence iterative learning control method based on full output feedback. Background Art

[0002] At present, issues related to vehicle autonomous driving have received widespread attention in real life. With the rapid development of the times, the accuracy requirements for autonomous control of vehicle systems are getting higher and higher. In real life, the precise model of the vehicle system is unknown, which makes some traditional control methods unable to meet the task of high-precision real-time tracking of the desired trajectory. The iterative learning control method is simple and does not require an accurate mathematical model of the dynamic system. Therefore, it is very suitable for controlling the above-mentioned vehicle system to ensure that the vehicle system can achieve the goal of high-precision tracking of the desired trajectory.

[0003] However, in real life, the system matrix corresponding to the vehicle system often changes over time. Traditional iterative learning control methods often cannot make the actual output trajectory of the above-mentioned vehicle system with time-varying characteristics converge monotonically to the desired trajectory. Therefore, it has limitations and cannot achieve the best control effect. Summary of the invention

[0004] In order to solve the above problems of the prior art, the present invention proposes a vehicle system monotonically convergent iterative learning control method based on full output feedback, which can make the actual output trajectory of the vehicle system monotonically converge to the desired trajectory.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] According to one aspect of the present invention, a vehicle system monotone convergence iterative learning control method based on full output feedback is proposed, comprising the following steps:

[0007] A: Setting the desired trajectory of the vehicle system;

[0008] B: Design the iterative learning control gain matrix of the full output feedback iterative learning control law;

[0009] C: obtaining input and output data of the vehicle system under the system input drive generated by the full output feedback iterative learning control law, and then calculating the tracking error and output deviation of the vehicle system in the current iteration;

[0010] D: based on the control input, output deviation and iterative learning control gain matrix of the vehicle system in the current iteration, using the full output feedback iterative learning control law to update the control input of the vehicle system when it is running next time;

[0011] E: Return to step C and continue until the desired trajectory is tracked, that is, the tracking error converges monotonically to 0.

[0012] Furthermore, the vehicle system has the following characteristics:

[0013]

[0014] Where t∈{0,1,…,T-1} is the running time, T is the running cycle, and k∈{1,2,…} is the number of iterations; and are the n-dimensional state, q-dimensional input and p-dimensional output of the vehicle system at time t in the kth iteration; A(t), B(t) and C(t) are the state matrix, input matrix and output matrix of the vehicle system at time t, which are independent of each other; x k (0) = x 0 Indicates that the initial state of the system can be reset to a fixed state x 0 .

[0015] Furthermore, step A specifically includes:

[0016] According to the tracking requirements, the expected trajectory is given manually as follows:

[0017] y d (t),t∈{1,2,…,T}.

[0018] Furthermore, step B specifically includes:

[0019] Control gain Γ t,s The following monotonic convergence conditions are met:

[0020] ρ(I q -C(t+1)B(t)Γ t,t )<1,t∈{0,1,…,T-1},

[0021] ||I qT -HΓ|| ∞ ≤1,

[0022] Among them I q is the q×q identity matrix, I qT is the qT×qT identity matrix, ρ(I q -C(t+1)B(t)Γ t,t ) is the matrix I q -C(t+1)B(t)Γ t,t The spectral radius, ||I qT -HΓ|| ∞ is the matrix I qT -The infinite norm of HΓ, Γ t,sis the control gain, t, s represent subscripts, and t, s∈{0, 1, …, T-1},

[0023] It is an iterative learning control gain matrix, which consists of control gains.

[0024] Furthermore, step C specifically includes:

[0025] According to the output y of the vehicle system at the current iteration k (t) and the expected trajectory y d (t), calculate the output deviation e of the vehicle system in the current iteration k (t) and the system output tracking error in the infinite norm sense where e k (t) = y d (t)-y k (t), t∈{1,2,…,T}.

[0026] Further, step D specifically includes:

[0027] D1: Determine the control input of the vehicle system when it is running according to the control input, output deviation and iterative learning control gain matrix of the current iteration of the vehicle system:

[0028]

[0029] Among them, u k (t) is the control input of the vehicle system in the kth iteration, e k (s+1) is the output deviation of the vehicle system in the kth iteration, Γ t,s is the control gain of the vehicle system, and the iterative learning control gain matrix Γ is composed of the control gain Γ t,s composition.

[0030] D2: Controlling the vehicle system to execute the next iteration process according to the control input of the vehicle system in the next iteration.

[0031] According to another aspect of the present invention, a vehicle system device is provided, specifically comprising:

[0032] A first acquisition module, used to acquire a desired trajectory of the vehicle system;

[0033] A second acquisition module, used to acquire an iterative learning control gain matrix of the vehicle system;

[0034] A third acquisition module, used to acquire a tracking error and an output deviation of the vehicle system in a current iteration;

[0035] A control module is used to update the control input of the vehicle system in the next iteration according to the control input, output deviation and iterative learning control gain matrix of the vehicle system in the current iteration, and control the vehicle system until the tracking error monotonically converges to 0.

[0036] Furthermore, the control module includes:

[0037] Full output iterative learning control rate submodule: used to determine the control input of the vehicle system when it runs next time according to the control input, output deviation and iterative learning control gain matrix of the current iteration of the vehicle system:

[0038]

[0039] Among them, u k (t) is the control input of the vehicle system in the kth iteration, Γ t,s is the control gain of the vehicle system, e k (s+1) is the tracking error of the vehicle system in the kth iteration;

[0040] Control submodule: used to control the vehicle system to execute the next iteration process according to the control input when the vehicle system is running next time.

[0041] Compared with the prior art, the monotonic iterative learning control method for a vehicle system based on full output feedback provided by the present invention overcomes the influence of the time-varying characteristics of the vehicle system, and makes the actual output trajectory of the vehicle system converge monotonically to the desired trajectory, thereby achieving tracking of the desired trajectory with higher accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The drawings constituting a part of the present application are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0043] Figure 1 The figure is a flow chart of a monotone convergence iterative learning control method for a vehicle system based on full output feedback according to an embodiment of the present invention.

[0044] Figure 2 This is the dynamic evolution diagram of the tracking error of the vehicle system along the iteration axis. DETAILED DESCRIPTION

[0045] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0046] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. 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 creative work should fall within the scope of protection of the present invention.

[0047] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so as to describe the embodiments of the present invention described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0048] Figure 1 FIG. 1 is a flow chart of a monotonic iterative learning control method for a vehicle system based on full output feedback according to an embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:

[0049] Step S101: setting an expected trajectory of the vehicle system, wherein the expected trajectory is an expected running route set manually and is a reference trajectory for the final expected running of the vehicle system;

[0050] Specifically, according to the tracking requirements, the expected trajectory is given manually as follows:

[0051] y d (t),t∈{1,2,…,T}.

[0052] Step S102: designing an iterative learning control gain matrix of a full output feedback iterative learning control law;

[0053] The full-output feedback iterative learning control rate can enable the vehicle system to generate a control input for the next iteration, thereby controlling the vehicle system; the iterative learning control gain matrix is ​​the core part of the full-output feedback iterative learning control law, which is used to ensure that the full-output feedback iterative learning control law can enable the vehicle system to track the desired trajectory.

[0054] Specifically, for a vehicle system with the following time-varying characteristics:

[0055]

[0056] Where t∈{0,1,…,T-1} is the running time, T is the running cycle, and k∈{1,2,…} is the number of iterations; and are the n-dimensional state, q-dimensional input and p-dimensional output of the vehicle system at time t in the kth iteration; A(t), B(t) and C(t) are the state matrix, input matrix and output matrix of the vehicle system at time t, which are independent of each other; x k (0) = x 0 Indicates that the initial state of the system can be reset to a fixed state x 0 .

[0057] Control gain Γ t,s The following conditions must be met:

[0058] ρ(I q -C(t+1)B(t)Γ t,t )<1,t∈{0,1,…,T-1},

[0059] ||I qT -HΓ|| ∞ ≤1,

[0060] Among them I q is the q×q identity matrix, I qT is the qT×qT identity matrix, ρ(I q -C(t+1)B(t)Γ t,t ) is the matrix I q -C(t+1)B(t)Γ t,t The spectral radius, ||I qT -HΓ|| ∞ is the matrix I qT -The infinite norm of HΓ, Γ t,s Control gain, t, s represent subscripts, and t, s∈{0, 1, …, T-1},

[0061] It is an iterative learning control gain matrix, which consists of control gains.

[0062] Step S103: obtaining input and output data of the vehicle system under system input drive generated by the full output feedback iterative learning control law, and then calculating the tracking error and output deviation of the vehicle system in the current iteration;

[0063] Specifically, the system input generated by the full output feedback iterative learning control law refers to the system input of the previous iteration, that is, u k-1 (t); the input data driven by the system input refers to the system input of the current iteration, that is, u k(t); the output data driven by the system input is the output y of the current iteration k (t); The above data can be obtained by installing corresponding position sensor devices in the vehicle system.

[0064] According to the output y of the vehicle system at the current iteration k (t) and the expected trajectory y d (t), calculate the output deviation e of the vehicle system in the current iteration k (t) and the system output tracking error in the infinite norm sense where e k (t) = y d (t)-y k (t), t∈{1,2,…,T}.

[0065] Step S104: updating the control input of the vehicle system when it is running next time by using the full output feedback iterative learning control law according to the control input, output deviation and iterative learning control gain matrix of the vehicle system in the current iteration;

[0066] Specifically, the iterative learning control rate of the vehicle system is first determined, and then the control input of the next iteration is updated according to the determined control rate, thereby controlling the vehicle system. In an embodiment of the present invention, the control input of the vehicle system during operation is determined according to the control input, output deviation and iterative learning control gain matrix of the current iteration of the vehicle system:

[0067]

[0068] Among them, u k (t) is the control input of the vehicle system in the kth iteration, which can be obtained by a sensor device installed on the vehicle system, Γ t,s is the control gain of the vehicle system, e k (s+1) is the output deviation of the vehicle system in the kth iteration; It means that each update needs to utilize the output deviation information at all times, which can improve learning performance.

[0069] The vehicle system is controlled to execute a next iteration process according to a control input of the vehicle system in the next iteration.

[0070] Step S105: Return to step S103 and continue until the tracking of the desired trajectory is completed, that is, the tracking error monotonically converges to 0.

[0071] Monotonic convergence means that the tracking error ||e k (t)|| ∞As the number of iterations k increases, the tracking error ||e k (t)|| ∞ It tends to 0. Monotonic convergence can prevent overshoot in the control process. Excessive overshoot may damage the system hardware.

[0072] Tracking the desired trajectory can enable the vehicle system to control the vehicle to move along a given route by a given iterative learning control rate without the need for human control.

[0073] The monotone convergence iterative learning control method for a vehicle system based on full output feedback of the present invention is described in detail below in conjunction with the embodiments, but it should not be understood as limiting the protection scope of the present invention.

[0074] Embodiment 1:

[0075] Consider a vehicle system with the following time-varying characteristics:

[0076]

[0077] The system matrix is:

[0078]

[0079] Consider the operation cycle as T = 20, and select the initial state of the vehicle system as x 0 =[0,0,0] T , the initial input is u0(t) = [0, 0] T , Perform the first iteration k=0;

[0080] S101: Determine the expected trajectory of the vehicle system as y d (t) = [sin(0.3t)e -0.05t , 2sin(0.3t)e -0.05t ] T , t∈{1, 2,…, 20}.

[0081] S102: Design an iterative learning control gain matrix of a full output feedback iterative learning control law;

[0082] The calculated matrix

[0083] Select the iterative learning control gain matrix:

[0084]

[0085] The control gain Γ can be obtained by iterative learning of the control gain matrix Γ t,s ,Right now

[0086]

[0087] And the selected iterative learning control gain matrix and control gain satisfy the following conditions:

[0088] ρ(I q -C(t+1)B(t)Γ t,t )<1,t∈{0,1,…,T-1},

[0089] ||I qT -HΓ|| ∞ ≤1,

[0090] Among them I q is the q×q identity matrix, I qT is the qT×qT identity matrix, ρ(I q -C(t+1)B(t)Γ t,t ) is the matrix I q -C(t+1)B(t)Γ t,t The spectral radius, ||I qT -HΓ|| ∞ is the matrix I qT -The infinity norm of HΓ.

[0091] S103: Obtaining input and output data of the vehicle system under system input drive generated by the full output feedback iterative learning control law, which are u k (t), t∈{0, 1, …, 19}, and y k (t), t∈{1, 2, ..., 20}, calculate the output deviation e under the current iteration k (t) = y d (t)-y k (t), t∈{1,2,…,T} and the system output tracking error in the infinite norm sense

[0092] S104: updating the control input of the vehicle system when it is running next time by using the full output feedback iterative learning control law according to the control input, output deviation and iterative learning control gain matrix of the vehicle system in the current iteration; specifically comprising the following steps:

[0093] First, the iterative learning control rate of the vehicle system is determined, and the control input of the vehicle system under operation is determined according to the control input, output deviation and iterative learning control gain matrix of the current iteration of the vehicle system:

[0094]

[0095] Among them, u k(t) is the control input of the vehicle system in the kth iteration, Γ t,s is the control gain of the vehicle system, e k (s+1) is the output deviation of the vehicle system in the kth iteration;

[0096] The vehicle system is controlled to execute a next iteration process according to a control input of the vehicle system in the next iteration.

[0097] S105: Return to step S103 and continue until the tracking of the desired trajectory is completed, that is, the tracking error monotonically converges to 0.

[0098] Figure 2 It is shown in Figure 2 that after 30 iterations, the tracking error of the vehicle system converges monotonically to 0. This figure shows that the vehicle system can perfectly track the desired output trajectory with high accuracy. Therefore, the monotonically convergent iterative learning control method for a vehicle system based on full output feedback provided by the present invention can overcome the influence of the time-varying characteristics of the vehicle system. For any given desired trajectory, the iterative learning control method can make the actual output trajectory of the vehicle system converge monotonically to the desired trajectory with high control accuracy.

[0099] The above specific implementation methods further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above are only specific implementation methods of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A monotonic convergence iterative learning control method for a vehicle system based on full output feedback, characterized in that: The following steps are involved: A: Setting the desired trajectory of the vehicle system; B: Design the iterative learning control gain matrix of the full output feedback iterative learning control law; C: obtaining input and output data of the vehicle system under the system input drive generated by the full output feedback iterative learning control law, and then calculating the tracking error and output deviation of the vehicle system in the current iteration; D: based on the control input, output deviation and iterative learning control gain matrix of the vehicle system in the current iteration, using the full output feedback iterative learning control law to update the control input of the vehicle system when it is running next time; E: Return to step C and continue until the desired trajectory is tracked, that is, the tracking error converges monotonically to 0; Wherein, the step B comprises: Control gain Γ t,s The following monotonic convergence conditions are met ρ(I q -C(t+1)B(t)Γ t,t )<1,t∈{0,1,…,T-1}, ||I qT -HΓ|| ∞ ≤1, Among them I q is the q×q identity matrix, I qT is the qT×qT identity matrix, ρ(I q -C(t+1)B(t)Γ t,t ) is the matrix I q -C(t+1)B(t)Γ t,t The spectral radius, ||I qT -HΓ|| ∞ is the matrix I qT -The infinite norm of HΓ, Γ t,s is the control gain, t, s represent subscripts, and t, s∈{0, 1, …, T-1}, To iteratively learn the control gain matrix, it consists of control gains; The step D specifically comprises: D1: Determine the control input of the vehicle system when it is running next time according to the control input, output deviation and control gain of the vehicle system in the current iteration: Among them, u k (t) is the control input of the vehicle system in the kth iteration, e k (s+1) is the output deviation of the vehicle system in the kth iteration, Γ t,s is the control gain of the vehicle system, and the iterative learning control gain matrix Γ is composed of the control gain Γ t,s composition; D2: Controlling the vehicle system to execute the next iteration process according to the control input of the vehicle system in the next iteration.

2. The vehicle system monotone convergence iterative learning control method based on full output feedback according to claim 1 is characterized in that: The vehicle system has the following characteristics: Where t∈{0,1,…,T-1} is the running time, T is the running cycle, and k∈{1,2,…} is the number of iterations; and are the n-dimensional state, q-dimensional input and p-dimensional output of the vehicle system at time t in the kth iteration; A(t), B(t) and C(t) are the state matrix, input matrix and output matrix of the vehicle system at time t, which are independent of each other; x k (0) = x 0 Indicates that the initial state of the system can be reset to a fixed state x 0 .

3. The vehicle system monotone convergence iterative learning control method based on full output feedback according to claim 1, characterized in that: The step A comprises: According to the tracking task requirements, the expected trajectory is set manually as follows: y d (t),t∈{1,2,…,T}。 4. The vehicle system monotone convergence iterative learning control method based on full output feedback according to claim 1, characterized in that: The step C comprises: According to the output y of the vehicle system at the current iteration k (t) and the expected trajectory y d (t), calculate the output deviation e of the vehicle system in the current iteration k (t) and the system output tracking error in the infinite norm sense where e k (t) = y d (t)-y k (t), t∈{1,2,…,T}.

5. A vehicle system device, characterized in that: include: A first acquisition module, used to acquire a desired trajectory of the vehicle system; A second acquisition module, used to acquire an iterative learning control gain matrix of the vehicle system; A third acquisition module, used to acquire a tracking error and an output deviation of the vehicle system in a current iteration; A control module, used for updating the control input of the vehicle system at the next iteration according to the control input, output deviation and iterative learning control gain matrix of the vehicle system at the current iteration, and controlling the vehicle system until the tracking error monotonically converges to 0; The second acquisition module controls the gain Γ t,s The following monotonic convergence conditions are met ρ(I q -C(t+1)B(t)Γ t,t )<1,t∈{0,1,…,T-1}, ||I qT -HΓ|| ∞ ≤1, Among them I q is the q×q identity matrix, I qT is the qT×qT identity matrix, ρ(I q -C(t+1)B(t)Γ t,t ) is the matrix I q -C(t+1)B(t)Γ t,t The spectral radius, ||I qT -HΓ|| ∞ is the matrix I qT -The infinite norm of HΓ, Γ t,s is the control gain, t, s represent subscripts, and t, s∈{0, 1, …, T-1}, To iteratively learn the control gain matrix, it consists of control gains; The control module comprises: Full output iterative learning control rate submodule: used to determine the control input of the vehicle system when it runs next time according to the control input, output deviation and iterative learning control gain matrix of the current iteration of the vehicle system: Among them, u k (t) is the control input of the vehicle system in the kth iteration, Γ t,s is the control gain of the vehicle system, e k (s+1) is the output deviation of the vehicle system in the kth iteration; Control submodule: used to control the vehicle system to execute the next iteration process according to the control input when the vehicle system is running next time.

Citation Information

Patent Citations

  • Advanced iterative learning control method for accurate target tracking

    CN101846979A

  • Iterative learning trajectory tracking control and robust optimization method for two-dimensional motion mobile robot

    CN105549598A

  • Robust monotonic convergence point-to-point iterative learning control method for single-axis feeding system

    CN113110063A

  • Real-time tracking iterative learning control system and method of time-varying vehicle system

    CN113625551A