Position type full-format generic model driven ocean robot model-free adaptive motion control method, program, equipment and storage medium

Abstract

The invention belongs to the technical field of robot motion control, and particularly relates to a position type full-format generic model driven ocean robot model-free self-adaptive motion control method, program and equipment and a storage medium. According to the invention, a position type full-format generic model is constructed for an ocean robot motion control system, a position type full-format pseudo gradient estimation algorithm is obtained through a pseudo gradient estimation resolving criterion function, and a position type full-format pseudo gradient reset condition and a pseudo gradient reset algorithm are designed. And a position type full-format model-free adaptive control law is obtained by controlling an input solution criterion function. The problems that when a model-free adaptive control method is directly applied to a non-self-balancing motion control system of a type of ocean robots, due to the fact that a generic model is not matched with a controlled system, system output oscillation divergence is caused, and the response speed is low are solved, the oscillation problem can be effectively solved, and the higher response speed is achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of robot motion control, and particularly relates to a model-free adaptive motion control method, program, device and storage medium for an underwater robot driven by a position-type full-format general model. Background Art

[0002] During the operation of an underwater robot, the maneuvering response model changes with the wetted area and speed of the underwater robot, and is also affected by environmental disturbances. Therefore, control methods designed based on mathematical models often cannot achieve good results. The model-free adaptive control method (MFAC) can adaptively adjust control parameters online without the need for an accurate model of the system, thereby effectively controlling the system output, and is applicable to systems that cannot be accurately modeled.

[0003] However, in the field of underwater robot motion control, there is a class of non-self-regulating systems, and the standard full-format general model does not match such non-self-regulating systems. Direct application of the standard FFDL-MFAC will cause severe oscillation problems in the system output. The heading control system is a typical non-self-regulating system. Taking the heading control system as an example, the control input u is the rudder angle command, and the system output y is the heading. According to maneuvering common sense, the change in heading Δy mainly depends on the rudder angle command u itself, rather than the change in the rudder angle command Δu. However, the standard full-format general model directly constructs the relationship between Δy(k + 1) and Δu(k), Δu(k - 1), …, Δu(k - L u + 1), which is the main reason why the standard FFDL-MFAC algorithm cannot converge the heading to the expected value.

[0004] A patent document with the publication date of April 15, 2022, the application patent number CN202110248550.5, and the title "Full-format model-free adaptive control method for a class of non-self-regulating systems" provides an improved MFAC method, which solves the problem of system output oscillation generated when the full-format model-free adaptive control method is applied to non-self-regulating systems such as the heading control system by adding a term related to the output change amount to the standard control input criterion function to reduce the integral effect of the control input. However, this method still uses the standard FFDL-MFAC general model and does not fundamentally solve the problem of general model mismatch, which affects the response speed and anti-interference effect while increasing the difficulty of parameter adjustment.

[0005] A patent document with the publication date of January 7, 2025, the application patent number CN202411696975.2, and the title "Position-type compact-format model-free adaptive heading control method for underwater robots" designs a position-type compact-format general model Δy(k + 1) = φ c (k)u(k) that is more suitable for the heading control system, where φ c(k) is a bounded time-varying parameter called pseudo partial derivative (PPD) that is estimated in real time at time k. Based on this general model, a position-type compact-form model-free adaptive control law and a pseudo partial derivative estimation algorithm are designed, enabling this method to be used in the motion control system of an ocean robot. However, the position-type compact-form general model only attributes the heading change Δy(k + 1) to the driving of the rudder angle command u(k) at the previous moment, without considering the time-delay effect and inertia effect of the manipulation mechanism of the ocean robot in the design. Specifically, the heading change of the ocean robot at any moment is not only affected by the historical rudder angle commands within a certain time window but also significantly affected by the heading angular velocity of the robot itself. Especially when the heading angular velocity of the ocean robot is relatively large, even if the rudder is turned in the reverse direction, due to the action of inertial force, the direction of the heading change often remains unchanged. Therefore, the model-free adaptive control method designed based on this general model has poor applicability to large ocean robots with obvious inertial effects and slow servo response speeds. Summary of the Invention

[0006] The object of the present invention is to solve the problem of system output oscillation divergence when directly applying the model-free adaptive control method (MFAC) to the non-self-balancing motion control system of a class of ocean robots, as well as the problem that the existing position-type compact-form MFAC scheme is not applicable to large ocean robots with significant inertial effects, and to provide a position-type full-form general model-driven model-free adaptive motion control method, program, device, and storage medium for ocean robots.

[0007] A position-type full-form general model-driven model-free adaptive motion control method for an ocean robot includes the following steps:

[0008] Obtain the expected motion control system output and the initial motion control system output of the ocean robot;

[0009] Preset an initial pseudo-gradient estimation vector, calculate the control input according to the position-type full-form model-free adaptive control law; transmit the control input to the motion control system, and obtain the output and output change amount of the motion control system;

[0010] Construct a sliding time window for the control input and the system output; form a sliding time window vector by all the control inputs within the control input sliding time window and all the output change amounts within the system output sliding time window;

[0011] Update the pseudo-gradient estimation vector according to the sliding time window vector and the output change amount of the motion control system; after updating the pseudo-gradient estimation vector, determine whether the pseudo-gradient reset condition is satisfied; if satisfied, reset the updated pseudo-gradient estimation vector to the initial pseudo-gradient estimation vector;

[0012] Repeat the above steps until the deviation between the output of the motion control system and the desired motion control system output meets the requirements.

[0013] Further, obtain the desired motion control system output y of the marine robot * and the output y0 of the initial motion control system; preset the initial pseudo-gradient estimation vector

[0014] where, is the i-th element of y +L u ; i = 1, 2,..., L y ; L u is the motion control system output linearization length constant; L u and L y are both positive integers, and L u ≥ 1, L y ≥ 1;

[0015] If the motion control system output increases when the control input is positive, then If the motion control system output increases when the control input is negative, then

[0016] Further, calculate the control input u(k) according to the position-type full-format model-free adaptive control law;

[0017]

[0018] where, is the i-th element of i ; ρ i is the step size factor, ρ

[0019] ∈ (0, 1]; λ is the weight coefficient, λ > 0;

[0020] Input the control input u(k) into the motion control system to obtain the output y(k + 1) of the motion control system and the output change Δy(k + 1) = y(k + 1) - y(k); * If the deviation between y(k + 1) and y

[0021] is less than the preset threshold, then end the motion control. u + 1, k] of all control inputs and the system output sliding time window [k - L y + 1, k] of all output changes to form a sliding time window vector

[0022]

[0023] Further, in the iterative calculation process, if k < j, then Δy(k - j) = 0 and u(k - j) = 0; when k = 1, Δy(1) = 0 and y(1) = y0.

[0024] Further, according to the sliding time window vector and the output change Δy(k + 1) of the motion control system, update the pseudo-gradient estimation vector

[0025]

[0026] where η is the step size factor, η ∈ (0, 2]; μ is the weight coefficient, μ > 0.

[0027] Further, the pseudo-gradient reset condition is:

[0028] If or or then reset the pseudo-gradient estimation vector, and let

[0029] A computer device / system, including a memory, a processor, and a computer program stored on the memory, where the processor executes the computer program to implement the steps of the above position-type full-format universal model-driven model-free adaptive motion control method for marine robots.

[0030] A computer-readable storage medium, on which a computer program / instructions are stored, and when the computer program / instructions are executed by a processor, the steps of the above position-type full-format universal model-driven model-free adaptive motion control method for marine robots are implemented.

[0031] A computer program product, including a computer program / instructions, and when the computer program / instructions are executed by a processor, the steps of the above position-type full-format universal model-driven model-free adaptive motion control method for marine robots are implemented.

[0032] The beneficial effects of the present invention are as follows:

[0033] The present invention proposes a position-type full-format universal model, and then designs a position-type full-format model-free adaptive control law, a pseudo-gradient estimation algorithm, and a pseudo-gradient reset mechanism, forming a position-type full-format universal model-driven model-free adaptive motion control method for marine robots. The present invention solves the problems of system output oscillation divergence and slow response speed caused by the mismatch between the universal model and the controlled system when directly applying the model-free adaptive control method to a class of non-self-equilibrating motion control systems of marine robots. Description of the Drawings

[0034] Figure 1 This is the overall architecture diagram of the present invention.

[0035] Figure 2 This is the control schematic diagram of the present invention.

[0036] Figure 3 This is the comparison diagram of the heading control simulation test between the present invention and other control methods. Detailed implementation manners

[0037] The present invention will be further described below with reference to the accompanying drawings.

[0038] For a common class of non-self-balancing motion control systems of marine robots, the inputs are acceleration information such as thrust and rudder angle, and the outputs are position information such as coordinates and heading. It can be expressed as the following second-order system:

[0039]

[0040] where y is the system output, u is the control input, b is the input gain, and g(...) is an unknown function.

[0041] For the above-mentioned class of non-self-balancing motion control systems of marine robots, the present invention provides a model-free adaptive motion control method for marine robots driven by a position-type full-format universal model.

[0042] Select an appropriate control input linearization length constant L u ≥1 and a system output linearization length constant L y ≥1. Then, in the model-free adaptive control method, construct a position-type full-format universal model for the motion control system of the marine robot;

[0043]

[0044] where Δy(k + 1) = y(k + 1) - y(k) is the output change of the motion control system in the (k + )-th control.

[0045]

[0046] is a bounded time-varying parameter vector called the pseudo-gradient (PG);

[0047] is the sliding time window vector composed of all control inputs within the control input sliding time window [k - L u + 1, k] and all output changes within the system output sliding time window [k - L y + 1, k];

[0048]

[0049] The pseudo-gradient estimation solution criterion function is as follows:

[0050]

[0051] where μ > 0 is the weight coefficient. Let and introduce the step-size factor η ∈ (0, 2], thus obtaining the position-type full-format pseudo-gradient estimation algorithm:

[0052]

[0053] The control input solution criterion function is as follows:

[0054]

[0055] where y * (k + 1) is the expected output of the system, and λ > 0 is the weight coefficient; let and introduce the step-size factor ρ i ∈ (0, 1], i = 1, 2, …, L, thus obtaining the position-type full-format model-free adaptive control law:

[0056]

[0057] The present invention constructs a position-type full-format general model for the motion control system of an underwater robot, obtains the position-type full-format pseudo-gradient estimation algorithm through the pseudo-gradient estimation solution criterion function; designs the position-type full-format pseudo-gradient reset condition and the pseudo-gradient reset algorithm; and obtains the position-type full-format model-free adaptive control law through the control input solution criterion function.

[0058] Based on the above content, a model-free adaptive motion control method for an underwater robot driven by a position-type full-format general model provided by the present invention specifically includes the following steps:

[0059] Step 1: Obtain the expected output y * of the motion control system of the underwater robot and the output y0 of the initial motion control system;

[0060] Step 2: Set the control input linearization length constant L u , the motion control system output linearization length constant L y , the step-size factors ρ i and η, the weight coefficients λ and μ; L u and L y are both positive integers, and L u ≥ 1, L y ≥ 1; ρ i ∈ (0, 1], i = 1, 2,..., L y + L u; η ∈ (0, 2]; λ > 0; μ > 0; Initialize the control count k = 1;

[0061] Initialize the pseudo-gradient estimation vector

[0062]

[0063] Wherein, represents the i-th element of; When k = 1, initialize L y + L u elements If the output of the motion control system increases when the control input is positive, then If the output of the motion control system increases when the control input is negative, then

[0064] Step 3: Calculate the control input u(k) in the current k-th control according to the position-type full-format model-free adaptive control law;

[0065]

[0066] Wherein, Δy(k) = y(k) - y(k - 1), Δy(1) = 0, y(1) = y0; If k < j, then Δy(k - j) = 0, u(k - j) = 0; That is

[0067] Step 4: Input the control input u(k) into the motion control system to obtain the output y(k + 1) and the output change Δy(k + 1) of the motion control system;

[0068] Step 5: Form a sliding time window vector from all the control inputs within the sliding time window [k - L u + 1, k] and all the output changes within the sliding time window [k - L y + 1, k]

[0069]

[0070] Step 6: According to the pseudo-gradient estimation vector the sliding time window vector and the output change Δy(k + 1) of the motion control system, update the pseudo-gradient estimation vector

[0071]

[0072] Step 7: If or or then reset the pseudo-gradient estimation vector, let

[0073] Step 8: If the deviation between y(k + 1) and y * is less than a preset threshold, end the motion control; otherwise, let k = k + 1 and return to Step 3.

[0074] Compared with the standard full - format universal model, the position - type full - format universal model adopted by the present invention is more suitable for the non - self - regulating motion control system. The universal model is the currently widely used standard full - format universal model, but this model does not match the non - self - regulating system.

[0075]

[0076] The output y of the heading control system is the heading, and the control input u is the rudder angle. According to the common sense of maneuvering, the change in heading Δy depends on the rudder angle u itself, rather than the change in rudder angle Δu. For example, when the rudder angle u is a non - zero constant, the marine robot will perform a steady turning motion. At this time, Δu = 0 and Δy is a non - zero constant. Therefore, Δy and Δu cannot form a linear relationship. However, the standard full - format universal model attempts to establish the relationship between Δy(k + 1) and Δu(k), Δu(k - 1), …, Δu(k - L u + 1). Therefore, the standard full - format universal model does not match the heading control system, resulting in severe heading oscillation. The position - type full - format universal model adopted by the present invention constructs the relationship between Δy(k + 1) and u(k), u(k - 1), …, u(k - L u + 1), which is more suitable for the heading control system and solves the problem of severe heading oscillation.

[0077] Compared with the existing position - type compact - format universal model, the position - type full - format universal model adopted by the present invention can capture the system dynamic characteristics based on historical input - output data of a certain dimension, can fully consider the inertia problem of the marine robot, and further improve the control performance. A position - type compact - format universal model is:

[0078] Δy(k + 1) = φ c (k)u(k)

[0079] where φ c (k) is the model parameter called the pseudo - partial derivative. The position - type compact - format universal model only attributes the change in heading Δy(k + 1) to the rudder angle command u(k) at the previous moment. However, when the marine robot itself has a large heading angular velocity, even if the rudder is turned in the opposite direction, due to the inertia effect, the direction of the change in heading often does not immediately change with the direction of the rudder angle command. Since the position - type compact - format universal model does not consider the inertia effect, it is less suitable for large - scale marine robots. In contrast, what the present invention adopts attempts to construct the relationship between Δy(k + 1) and Δy(k), Δy(k - 1), …, Δy(k - L y+1), u(k), u(k - 1), …, u(k - L u +1), that is, it is considered that the change in the heading at the next moment is related to the change in the heading at the previous L y moments and the rudder angle command at the previous L u moments, fully considering the influence of historical control inputs on the current state and the turning inertia effect, which is more suitable for the heading control system.

[0080] Figure 3 This is a comparison chart of the effect improvement of a position-type full-format generalized model-driven model-free adaptive control method (P-FFDL-MFAC) in the present invention. As Figure 3 shown, the solid line is the desired heading of the marine robot, the dotted line is the control effect of the standard full-format dynamic linearization model-free adaptive control method (FFDL-MFAC), the equal-length dashed line is the control effect of a position-type compact format dynamic linearization model-free adaptive control method (P-CFDL-MFAC), and the dash-dotted line is the control effect of a position-type full-format generalized model-driven model-free adaptive control method (P-FFDL-MFAC) described in the embodiment of the present application. It can be clearly seen from Figure 3 this that the present invention can effectively solve the oscillation problem and has a faster response speed compared with a position-type compact format model-free adaptive control method (P-CFDL-MFAC).

[0081] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A model-free adaptive motion control method for an ocean robot driven by a position-type full-format universal model, characterized in that: Obtain the desired motion control system output and the initial motion control system output of the ocean robot; Preset an initial pseudo-gradient estimation vector, and calculate the control input according to the position-type full-format model-free adaptive control law; transmit the control input to the motion control system to obtain the output and output change of the motion control system; Construct a sliding time window for the control input and the system output; form a sliding time window vector by all the control inputs within the control input sliding time window and all the output changes within the system output sliding time window; Update the pseudo-gradient estimation vector according to the sliding time window vector and the output change of the motion control system; after updating the pseudo-gradient estimation vector, determine whether the pseudo-gradient reset condition is satisfied; if satisfied, reset the updated pseudo-gradient estimation vector to the initial pseudo-gradient estimation vector; Repeat the above steps until the deviation between the output of the motion control system and the desired motion control system output meets the requirements.

2. A model-free adaptive motion control method for an ocean robot driven by a position-type full-format general model as claimed in claim 1, characterized in that: Obtain the expected motion control system output y of the marine robot * and the output y0 of the initial motion control system; preset the initial pseudo-gradient estimation vector wherein, is the i-th element of; i = 1, 2,..., L y +L u ; L y is the output linearization length constant of the motion control system; L u is the control input linearization length constant; L u and L y are both positive integers, and L u ≥1, L y ≥1; If the control input is positive and the output of the motion control system increases, then If the control input is negative and the output of the motion control system increases, then 3. A model-free adaptive motion control method for an ocean robot driven by a position-based full-format general model according to claim 2, characterized in that: Calculate the control input u(k) according to the position-type full-format model-free adaptive control law; wherein, is the i-th element of; ρ i is the step factor, ρ i ∈(0,1]; λ is the weight coefficient, λ > 0; Input the control input u(k) into the motion control system to obtain the output y(k + 1) of the motion control system and the output change Δy(k + 1) = y(k + 1) - y(k); If the deviation between y(k + 1) and y * is less than the preset threshold, the motion control is terminated.

4. A model-free adaptive motion control method for an ocean robot driven by a position-type full-format general model according to claim 3, characterized in that: The control input sliding time window [k - L u + 1, k] includes all control inputs within it, and the system output sliding time window [k - L y + 1, k] includes all output changes within it. These form a sliding time window vector 5. A model-free adaptive motion control method for an ocean robot driven by a position-based full-format general model according to claim 4, characterized in that: During the iterative calculation process, if k < j, then Δy(k - j) = 0 and u(k - j) = 0; when k = 1, Δy(1) = 0 and y(1) = y0, 6. A model-free adaptive motion control method for an ocean robot driven by a position-based full-format general model according to claim 4, characterized in that: According to the sliding time window vector and the output change Δy(k + 1) of the motion control system, update the pseudo-gradient estimation vector Wherein, η is the step size factor, η ∈ (0, 2]; μ is the weight coefficient, μ > 0.

7. A model-free adaptive motion control method for a marine robot driven by a position-based full-format general model according to claim 6, characterized in that: The pseudo-gradient reset condition is: If or or then reset the pseudo-gradient estimation vector, and let 8. A computer device / apparatus / system, comprising a memory, a processor, and a computer program stored on the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having computer programs / instructions stored thereon, characterized in that: When the computer program / instructions are executed by the processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer program product, comprising a computer program / instructions, characterized in that: When the computer program / instructions are executed by the processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images