A magnetic drive robot adaptive control method and system considering hysteresis compensation

Abstract

The application discloses a kind of magnetic drive robot adaptive control method and system considering hysteresis compensation, belong to the nonlinear control technical field of magnetic drive robot system with hysteresis loop. Including: the dynamics model of magnetic drive robot is constructed, and then the kinematics prediction state space matrix of magnetic drive robot is obtained;Unknown time delay is estimated based on kinematics prediction state space matrix and unknown disturbance is estimated and compensated;Magnetic drive robot error dynamics model is established based on unknown time delay;And in the prediction time domain step, the state prediction model after time delay compensation is obtained;Design objective function based on state prediction model, and obtain control sequence based on objective function;Actual control signal is obtained through hysteresis compensation based on control sequence, and magnetic drive robot is controlled based on actual control signal.The application can realize high-precision and high-stability control in magnetic drive robot system by combining preset performance function.

Description

Technical Field

[0001] This invention belongs to the field of nonlinear control technology for magnetically driven robot systems with hysteresis, and specifically relates to an adaptive control method and system for magnetically driven robots that considers hysteresis compensation. Background Technology

[0002] Magnetic-driven robots, controlled by external magnetic fields, can achieve sub-millimeter precision directional movement and targeted manipulation in complex biological fluids (such as blood and lymph), demonstrating unique advantages in recent years in scenarios such as thrombus removal and tumor drug delivery. For example, magnetic-driven robots hold promise for applications in kidney stone fragmentation or ophthalmic surgery, as well as for transporting and sorting microorganisms, assembling micromechanical components, and minimally invasive medical applications. Due to their small size, magnetic-driven robots are difficult to power themselves using their own energy sources. Various magnetic fields for driving these robots have been studied, such as rotational, oscillating, and gradient magnetic fields. Different magnetic field driving schemes are suitable for different scenarios. For microrobots, high-precision tracking control is essential for task completion. However, magnetic-driven robots are driven by external magnetic fields, unlike traditional macroscopic robots which are self-driven. This makes their control more difficult. Furthermore, the unknown resistance of tissue fluids and the uncertain dynamic parameters of magnetic-driven robots further complicate achieving satisfactory tracking performance. Therefore, designing a controller for magnetic-driven robots is one of the main challenges.

[0003] Early research primarily employed open-loop control strategies, allowing magnetically driven robots to move along curved paths using predefined magnetic fields and to manually control them to complete turns in the vertical plane. While classic PID control is widely used due to its simple structure, its linear adjustment mechanism struggles to adapt to the non-minimum phase characteristics of magnetically driven systems, and parameter tuning heavily relies on operator experience. With the increasing demand for closed-loop control, researchers have proposed various solutions: planar path tracking methods for magnetically driven robots based on three-dimensional steering control, and image-based visual servoing arbitrary path tracking control methods.

[0004] Another solution is a control method with less stringent model requirements, such as model predictive control (MPC). However, existing MPC methods typically incorporate constraints into the objective function in the form of penalty terms, but static weights are difficult to balance dynamic performance and robustness under different operating conditions. In this case, pre-defined performance control (PPC) is introduced. PPC is highly effective at compensating for uncertainties and disturbances in nonlinear systems, and it is widely used in many control schemes. Combining these two methods ensures that the trajectory tracking error and the changes in each variable remain within the set ideal range, thus quickly achieving the control objective and ensuring system stability during operation. For example, for a nonlinear system, a combination of PPC and robust control is proposed, guaranteeing both transient and steady-state performance while maintaining robust disturbance suppression. An adaptive fuzzy control method based on pre-defined performance is proposed, which can ensure that the tracking error is constrained within a specified range without requiring known initial errors. Therefore, this paper integrates these relatively mature control methods to address the error tracking problem of magnetically driven robots, filling a research gap in this area.

[0005] Hysteresis is a physical phenomenon related to both the past state and current input of a system. It can be understood as a delayed response effect and is one of the main causes of system instability and oscillations. In recent years, research on hysteresis compensation methods has been booming. For example, adaptive fuzzy control schemes under a backpropagation framework have been established for stochastic pure feedback nonlinear systems with unknown directional hysteresis; and an implicit inverse compensator has been proposed for a butterfly PI model used to predict hysteresis effects. These control methods can mitigate the impact of hysteresis on control accuracy, but they all rely heavily on establishing a complete and accurate hysteresis mathematical model and mostly depend on constructing a state observer for real-time monitoring. Research on how to compensate for hysteresis in magnetically driven robot systems remains insufficient and requires further experimentation.

[0006] Research on adaptive pseudo-inverse compensation control for magnetically driven robot systems with hysteresis is very limited. Summary of the Invention

[0007] The purpose of this invention is to address the shortcomings of existing technologies by providing a novel adaptive control structure that suppresses nonlinear hysteresis, achieves more accurate and stable trajectory tracking and error control, and overcomes the limitations of existing methods, thereby improving the performance of magnetically driven robots in practical applications. By designing an adaptive control method for magnetically driven robots that considers hysteresis compensation and combining it with a preset performance function, high-precision and high-stability control can be achieved in magnetically driven robot systems.

[0008] To achieve the above objectives, the present invention provides the following solution: an adaptive control method for a magnetically driven robot considering hysteresis compensation, comprising the following steps: S1. Construct a dynamic model of the magnetically driven robot, and then obtain the kinematic prediction state space matrix of the magnetically driven robot; S2. Estimate the unknown time delay and the unknown disturbance based on the kinematic prediction state space matrix; S3. Establish an error dynamics model for the magnetically driven robot based on the unknown time delay; and obtain a state prediction model after time delay compensation within the predicted time domain step. S4. Design an objective function based on the state prediction model, and obtain a control sequence based on the objective function; S5. Based on the control sequence, obtain the actual control signal through hysteresis compensation, and control the magnetically driven robot based on the actual control signal.

[0009] More preferably, in S1, the dynamic model includes:

[0010] In the formula, Forward speed; Angular velocity; and These represent the drag coefficients perpendicular to and parallel to the helical axis, respectively. N , r , These represent the number of turns, radius, and pitch angle of the magnetically driven robot.

[0011] More preferably, in S1, the kinematic prediction state space matrix includes: ; In the formula, For the system matrix, For inputs that include hysteresis, For translation terms, The time delay step size; For unknown disturbances, This is the interference gain matrix; B This represents the input matrix.

[0012] More preferably, in k At time t, methods for estimating unknown time delays include: .

[0013] More preferably, in S2, the method for estimating the unknown disturbance includes: ; In the formula, superscript Indicates an estimated value; L This represents the gain matrix.

[0014] More preferably, in S3, the error dynamics model includes: ; In the formula, , The tracking error is for the two position components; , This is the preset change in the position of the magnetically driven robot; B x ( ) B y ( ) represents the x and y components of the input matrix; The state prediction model includes:

[0015] In the formula, m Indicates the current step index; n This represents the prediction time-domain step offset; w future ( ) represents the future estimated input; N p Indicates the predicted horizon.

[0016] More preferably, in S4, the objective function includes: ; In the formula, To control the field of vision; Q Represents the state error weight matrix; R This represents the control input weight matrix; These are the perturbation suppression weights.

[0017] More preferably, the method for obtaining the actual control signal based on the control sequence includes: ; ; In the formula, O , W They represent S and Z The upper bound; S and Z For threshold; and These are the system's input and output signals, respectively. , For weighted functions and relay operators.

[0018] The present invention also provides an adaptive control system for a magnetically driven robot considering hysteresis compensation, comprising: The dynamic model building module is used to construct the dynamic model of the magnetically driven robot, and then obtain the kinematic prediction state space matrix of the magnetically driven robot. The estimation module is used to estimate unknown time delays and estimate and compensate for unknown disturbances based on the kinematic prediction state space matrix. The state prediction module is used to establish an error dynamics model of the magnetically driven robot based on the unknown time delay; and to obtain the state prediction model after time delay compensation within the prediction time domain step. The control sequence acquisition module is used to design an objective function based on the state prediction model and obtain a control sequence based on the objective function. The control signal acquisition module is used to obtain the actual control signal based on the control sequence through hysteresis compensation, and to control the magnetically driven robot based on the actual control signal.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. A pseudo-inverse control algorithm is proposed to address the input hysteresis nonlinearity in the motion control system of magnetically driven robots. This method avoids constructing an explicit hysteresis inverse model, instead achieving the control objective by directly searching for the actual control signal within the control law. This method eliminates the path dependence bias of the control signal caused by hysteresis, broadening the research scope of magnetically driven robot systems exhibiting hysteresis.

[0020] 2. To address the challenge of establishing accurate dynamic models for magnetically driven robots, a Model Predictive Control (MPC) approach is adopted to directly design an objective function matrix incorporating disturbance suppression terms, combined with a pre-defined Performance Control (PPC) strategy. This strategy not only achieves excellent path tracking performance but also demonstrates good generalization ability, enabling joint optimization of tracking errors in both transient and steady-state states.

[0021] 3. The method of the present invention introduces a preset response parameter as a weight adjustment matrix, which enables the system error to converge in a timely manner and is independent of the initial state of the system, thereby enhancing the flexibility and adaptability of the control strategy and making it suitable for a variety of practical engineering application scenarios.

[0022] 4. To verify the trajectory tracking effect of the PPMPC strategy, a micro magnetic drive robot control experimental platform was built. This platform mainly consists of a magnetic drive robot, a three-dimensional Helmholtz magnetic drive system, and a data acquisition and host computer. The proposed control scheme was experimentally verified against a control scheme without hysteresis inverse compensation and a traditional PID control scheme. Experimental results show that the proposed control scheme has the advantages of small tracking error and fast convergence speed. Attached Figure Description

[0023] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly introduced below. Obviously, the 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.

[0024] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 The figures show the experimental results of the PPMPC algorithm of the present invention on a magnetic drive robot system; where (a) is the experimental result of the PPMPC control method on a curved path; and (b) is the experimental result of the PMPC control method on a circular path. Figure 3 The figures show the comparative experimental results of the control algorithm of the present invention; where (a) shows the comparative results of the three control methods under curved paths; and (b) shows the comparative results of the three control methods under circular paths. Figure 4 This is a graph showing the error comparison results of the control algorithm of the present invention. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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.

[0026] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0027] Example 1: like Figure 1 As shown, this embodiment provides an adaptive control method for a magnetically driven robot considering hysteresis compensation, including the following steps: S1. Construct a dynamic model of the magnetically driven robot, and then obtain the kinematic prediction state space matrix of the magnetically driven robot.

[0028] The direction of motion of a magnetically driven robot can be represented by two characteristic angles: pitch angle and yaw angle. That is, the direction of motion relative to the horizontal plane The included angle; azimuth angle That is, the projection of the direction of motion onto the horizontal plane and the axis OL-Y The angle between them. Driven by a uniform rotating magnetic field with a strength of... The magnetic torque applied to a magnetically driven robot under a magnetic field The following is represented as: (1) In the formula, The volume of the magnetized object. Let T be the magnetic flux density of the applied field. The magnetization of the object is expressed in A / m.

[0029] The magnetic field of a three-dimensional orthogonal Helmholtz coil can be considered uniform, so the magnetic force is zero. Therefore, the magnetically driven robot is only driven by magnetic torque. and thrust With the forward speed of the magnetically driven robot and angular velocity The relationship between them is described by the following advancement matrix: (2) in, (3) (4) (5) In the formula, and These represent the drag coefficients perpendicular to and parallel to the helical axis, respectively. N , r , These represent the number of turns, radius, and pitch angle of the magnetically driven robot.

[0030] When only a uniform rotating magnetic field is applied and the motion state of the magnetically driven robot reaches stability, the dynamic model of the magnetically driven robot can be calculated according to the following formula: (6) The target parameters of the magnetically driven robot converge within a set time using a preset performance function (PPC). The implementation of the preset performance function involves two key techniques: the preset performance function itself and error transformation. The tracking error of the two position components of the magnetically driven helical robot is defined as: (7) In the formula, This refers to the actual robot position. , This indicates the desired robot position.

[0031] The most commonly used default performance function is a function that converges in infinite time: (8) In the formula, This is the initial error tolerance; This represents the upper limit of steady-state error. To control the convergence rate.

[0032] To achieve the preset performance requirements, a preset performance function is used to constrain the system's error variables, limiting the error to a certain range. Specifically, this involves mapping the constrained error to unconstrained variables using an error transformation function, which is constructed as follows: (9) Three-dimensional orthogonal Helmholtz coils carrying triaxial current A uniform magnetic field is generated, the direction of which is determined by the azimuth angle. and pitch angle describe: (10) In the formula, , , The electromagnetic conversion coefficient of the coil. For current input, The magnetic field frequency is denoted as .

[0033] Helmholtz coils exhibit typical hysteresis under alternating current, meaning the change in the direction of the rotating magnetic field lags behind the input command. This causes a nonlinearity problem in the control input. To address this issue, this invention uses the Preisach model to describe it.

[0034] The predictive model for a magnetically driven robot system with hysteresis nonlinearity is described below: (11) In the formula, In spatial coordinate state; For unknown disturbances; This is the interference gain matrix.

[0035] Define angle increment State vector X =[ x , y Then, the angle increment at each step can be expressed as: (12) In the formula, T s Indicates the system sampling period; The time delay step size; B Represents the input matrix; This is a translation term; For inputs that include hysteresis.

[0036] The kinematic prediction state space matrix is ​​then expressed as: (13) In the formula, A This is the system matrix.

[0037] S2. Estimate unknown time delays and estimate and compensate for unknown disturbances based on the kinematic prediction state space matrix.

[0038] exist k At time t, methods for estimating unknown time delays include: (14) In the sliding window N Minimize the prediction residual within the inner range: (15) In the formula, These are the minimum and maximum values ​​of the time delay step size; To predict the window size.

[0039] Introducing Exponentially Weighted Moving Average (EWMA) optimizes the results in terms of time-lag smoothing: (16) In the formula, β Denotes the smoothing factor, where This is used to control the proportion of historical information retained. This represents the estimated time delay at the current moment.

[0040] Design a disturbance observer to estimate unknown disturbances: (17) In the formula, superscript Indicates an estimated value; L This represents the gain matrix.

[0041] The gain matrix is: (18) In the formula, It is determined by the parameters of the experimental setup.

[0042] Define the disturbance estimation error as: (19) The state estimation error equation is: (20) In the formula, This represents the deviation coefficient.

[0043] S3. Establish an error dynamic model for the magnetically driven robot based on the unknown time delay; and obtain the state prediction model after time delay compensation within the predicted time domain step.

[0044] According to formula (7), the error dynamics equation of the magnetically driven robot can be obtained: ;(twenty one) In the formula, and This represents the preset change in the position of the magnetically driven robot; where, ; B x ( k ), B y ( k ) represents the input matrix x , y Quantity.

[0045] Within the prediction time step, the predicted state after time delay compensation is as follows: (twenty two) In the formula, m Indicates the current step index; n This represents the prediction time-domain step offset; w future ( ) represents the future estimated input; N p Indicates the predicted horizon.

[0046] S4. Design an objective function based on the state prediction model, and obtain the control sequence based on the objective function.

[0047] S5. Based on the control sequence, the actual control signal is obtained through hysteresis compensation, and the magnetically driven robot is controlled based on the actual control signal.

[0048] The MPC method optimizes the prediction error on the future control input sequence and then applies the first element of the output to the control. The optimization objective is a quadratic function of the state and control inputs. The objective function is designed as follows: ;(twenty three) In the formula, To control the field of vision, Q Represents the state error weight matrix; R This represents the control input weight matrix; These are the perturbation suppression weights.

[0049] The weight matrix generally takes the following form:

[0050] In the formula,K Represents the dimension of the state quantity.

[0051] Traditional model predictive control uses a fixed weight matrix. and In the large error region, fixed weights may lead to insufficient control input and limited convergence speed; in the small error region, excessively high error weights may cause high-frequency oscillations in the control input and reduce smoothness.

[0052] The weight matrix is ​​dynamically adjusted by tracking error feedback in real time. This problem can be improved by defining a time-varying weight matrix. For diagonal matrices: ;(twenty four) Among them, weight The adjustment factor determines the sensitivity of the error to the weights; substituting the above equation into the objective function, we can expand and rearrange it into the standard form of a quadratic programming problem: (25) in: (26) Among them, superscript T Indicates transpose; H It is a Hessian matrix, and it is positive definite. It represents the quadratic part of the objective function. For the prediction model matrix, vector F Indicates the linear part; , This represents the error dynamic weight matrix and the input dynamic weight matrix.

[0053] Clearly, in the real world, control inputs are physically constrained. Therefore, the optimization problem must be solved by ensuring that the control satisfies the following constraint: Input rate of change constraint: .in, This indicates the maximum change value of the control input.

[0054] Therefore, the optimization problem is to find the objective function that makes the objective function... J Minimized w Control sequence.

[0055] Helmholtz coils exhibit typical hysteresis under alternating current, meaning the change in the direction of the rotating magnetic field lags behind the input command. This causes a nonlinearity problem in the control input. To address this issue, the Preisach model is used, as shown in the following expression: (27) in, O , W They representS and Z The upper bound; and These are the system's input and output signals, respectively. , The weighting function and relay operator are expressed as follows: (28) (29) In the formula, S and Z The threshold is used. The hysteresis loop formed by the Preisach model is monotonic, which provides a theoretical basis for the subsequently proposed hysteresis pseudo-inverse compensation scheme.

[0056] It is obvious that the actual control signal It is coupled in the output signal of the MPC. Therefore, a method is designed from Extract The algorithm is crucial. Let satisfy: (30) definition The actual input range is And assume exist It is monotonic within the range. For any have: (31) For all ,definition and ,in For variable parameters, ,therefore: (32) (33) Using the pseudo-inverse algorithm in Table 1, Equivalent to step size. In Time, through continuous accumulation The value is determined and .Will With control signals Compare until satisfied. Finally determined To achieve control signals containing hysteresis Extracting actual control signals The actual control signal is selected as follows: (34) Table 1

[0057] Example 2: To better understand the present invention, this embodiment provides a detailed description of the proposed method in conjunction with specific experimental verification.

[0058] To verify the effectiveness of the control algorithm proposed in this invention, a control experiment was conducted on a physical robot on a magnetic drive robot control experimental platform. The entire experimental process will be described in three parts below.

[0059] Part 1: Experimental Equipment.

[0060] The Magnetic Drive Robot Control Experimental Platform (MDREP) mainly consists of the following components: Magnetic field generating device: It adopts a structure of three sets of orthogonally arranged Helmholtz coils, with each set of coils corresponding to the three coordinate axes in space. It can generate a uniform rotating magnetic field in a working space of 100mm×100mm×100mm to provide driving force for the micro-robot.

[0061] Controller System: The Speedgoat real-time simulation platform was used as the main controller in the experiment. The platform is equipped with a 2.0 GHz Intel quad-core CPU and a high-performance FPGA module integrating Xilinx Artix-7 and Kintex-7 series CPUs. In the control strategy deployment, the Speedgoat platform loads the control algorithm through the Simulink Real-Time environment to achieve real-time closed-loop control of the entire magnetic drive system.

[0062] Current amplification module: Three Aigtek ATA-214 high-bandwidth servo amplifiers are used to amplify the analog signal output by the controller and drive the Helmholtz coil system.

[0063] Position Awareness System: To achieve real-time perception of the robot's position information, the experimental platform is equipped with a vision system consisting of a Keyence CV-X450F industrial camera, which captures the target robot's motion trajectory in the workspace at a frame rate of 30 fps.

[0064] Part 2: Preparations before the experiment.

[0065] The basic setup data for MDREP required for the experiment are shown in Table 2.

[0066] Table 2

[0067] The experiment compared three control schemes: 1) the proposed PPC cascaded MPC pseudo-inverse control method considering hysteresis compensation (PPMPC), 2) the control method without considering hysteresis effect (PPMPC without PI), and 3) the PID control method. Figure 2 The experimental results of the proposed PPMPC control method are shown on predetermined circular and curved paths. Real-time results of motion along the curve are recorded in the top camera. The black line represents the reference path. The red line represents the real-time path. 3) Figure 3 The experimental results of the three control methods under a given path are compared. In order to verify the universality of the proposed control method, a simulated blood vessel curve path was set up, and multiple experiments were carried out for different starting points. Figure 4 The error comparison of the three control methods and the error distribution comparison chart show that the PPMPC control method has the smallest median error and the smallest error spread in the specific control process.

[0068] To highlight the advantages of this control scheme, root mean square error (RMSE) and mean absolute error (MAE) are defined. PPMPC, PPMPC without PI, and PID control methods are shown in detail. Detailed data are shown in Table 3.

[0069] Table 3

[0070] This invention focuses on a magnetically driven robot system exhibiting hysteresis and unknown dynamic characteristics. By integrating Preset Performance Control (PPC), Model Predictive Control (MPC), and a hysteresis pseudo-inverse compensation algorithm, the tracking accuracy of the magnetically driven robot is effectively improved, and the response of the control algorithm is accelerated. The hysteresis nonlinearity present in the system is successfully resolved, thereby enhancing the overall system stability—a challenge often overlooked in previous research. Furthermore, the control signal is represented as a double integral function, and the pseudo-inverse control algorithm is incorporated to further optimize the state constraints and error limits of the magnetically driven robot system.

[0071] To validate the proposed control algorithm, an experimental platform for controlling a magnetically driven robot was built, and extensive comparative experiments were conducted using different control strategies, taking into account as many variables as possible. The experimental results and error analysis confirmed the effectiveness and superiority of the proposed algorithm, significantly enhancing the controller's performance and better fulfilling the control tasks of the magnetically driven robot system.

[0072] Example 3: This embodiment also provides an adaptive control system for a magnetically driven robot considering hysteresis compensation, including: a dynamic model construction module for constructing a dynamic model of the magnetically driven robot, thereby obtaining the kinematic prediction state space matrix of the magnetically driven robot; an estimation module for estimating unknown time delays and estimating and compensating for unknown disturbances based on the kinematic prediction state space matrix; a state prediction module for establishing an error dynamic model of the magnetically driven robot based on unknown time delays, and obtaining a time-delay-compensated state prediction model within the prediction time domain step; a control sequence acquisition module for designing an objective function based on the state prediction model, and obtaining a control sequence based on the objective function; and a control signal acquisition module for obtaining actual control signals based on the control sequence through hysteresis compensation, and controlling the magnetically driven robot based on the actual control signals.

[0073] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. An adaptive control method for a magnetically driven robot considering hysteresis compensation, characterized in that, Includes the following steps: S1. Construct a dynamic model of the magnetically driven robot, and use a preset performance function to make the target parameters of the magnetically driven robot converge within a set time, thereby obtaining the kinematic prediction state space matrix of the magnetically driven robot. S2. Estimate the unknown time delay and estimate and compensate for the unknown disturbance based on the kinematic prediction state space matrix; S3. Establish an error dynamics model for the magnetically driven robot based on the unknown time delay; and obtain a state prediction model after time delay compensation within the predicted time domain step. S4. Design an objective function based on the state prediction model, and obtain a control sequence based on the objective function; S5. Based on the control sequence, obtain the actual control signal through hysteresis compensation, and control the magnetically driven robot based on the actual control signal.

2. The adaptive control method for a magnetically driven robot considering hysteresis compensation according to claim 1, characterized in that, In S1, the dynamic model includes: In the formula, Forward speed; Angular velocity; and These represent the drag coefficients perpendicular to and parallel to the helical axis, respectively. N , r , These represent the number of turns, radius, and pitch angle of the magnetically driven robot.

3. The adaptive control method for a magnetically driven robot considering hysteresis compensation according to claim 1, characterized in that, In S1, the kinematic prediction state space matrix includes: ； In the formula, For the system matrix, For inputs that include hysteresis, For translation terms, The time delay step size; For unknown disturbances, This is the interference gain matrix; B This represents the input matrix.

4. The adaptive control method for a magnetically driven robot considering hysteresis compensation according to claim 3, characterized in that, exist k At time t, methods for estimating unknown time delays include: 。 5. The adaptive control method for a magnetically driven robot considering hysteresis compensation according to claim 4, characterized in that, In S2, the methods for estimating unknown disturbances include: ； In the formula, superscript Indicates an estimated value; L This represents the gain matrix.

6. The adaptive control method for a magnetically driven robot considering hysteresis compensation according to claim 5, characterized in that, In S3, the error dynamics model includes: ； In the formula, , The tracking error is for the two position components; , This is the preset change in the position of the magnetically driven robot; B x ( ) B y ( ) represents the x and y components of the input matrix; The state prediction model includes: In the formula, m Indicates the current step index; n This represents the prediction time-domain step offset; w future ( ) represents the future estimated input; N p Indicates the predicted horizon.

7. The adaptive control method for a magnetically driven robot considering hysteresis compensation according to claim 6, characterized in that, In S4, the objective function includes: ； In the formula, To control the field of vision; Q Represents the state error weight matrix; R This represents the control input weight matrix; These are the perturbation suppression weights.

8. The adaptive control method for a magnetically driven robot considering hysteresis compensation according to claim 1, characterized in that, The method for obtaining the actual control signal based on the control sequence includes: ； ； In the formula, O , W They represent S and Z The upper bound; S and Z For the threshold; and These are the system's input and output signals, respectively. , For weighted functions and relay operators.

9. An adaptive control system for a magnetically driven robot considering hysteresis compensation, the control system being used to implement the control method according to any one of claims 1-8, characterized in that, include: The dynamics model building module is used to build the dynamics model of the magnetically driven robot. By using a preset performance function, the target parameters of the magnetically driven robot converge within a set time, thereby obtaining the kinematic prediction state space matrix of the magnetically driven robot. The estimation module is used to estimate unknown time delays and estimate and compensate for unknown disturbances based on the kinematic prediction state space matrix. The state prediction module is used to establish an error dynamics model of the magnetically driven robot based on the unknown time delay; and to obtain the state prediction model after time delay compensation within the prediction time domain step. The control sequence acquisition module is used to design an objective function based on the state prediction model and obtain a control sequence based on the objective function. The control signal acquisition module is used to obtain the actual control signal based on the control sequence through hysteresis compensation, and to control the magnetically driven robot based on the actual control signal.

Images