Full-saturation learning type motion control method, system and equipment based on disturbance separation

Abstract

The invention provides a full-saturation learning type motion control method, system and equipment based on disturbance separation, and relates to the field of electromechanical system servo control. The method comprises the following steps: constructing a control model for a direct-drive motion unit, wherein the control model has periodic and non-periodic uncertainty characteristics; determining at least one of a compensation adaptive control law, an expected compensation control law and a rapid compensation term of the control model; determining one or both of a feedback control law and a robust control law of a control system of the direct-drive motion unit; determining a motion controller of the direct-drive motion unit according to the control model and the control system; the direct-drive motion unit is controlled based on the motion controller to realize electromechanical servo precision motion control, so that the problems that the existing precision tracking control method for repeated operation of the direct-drive unit cannot fully analyze periodic and non-periodic dynamic characteristics of the system and cannot effectively compensate and suppress disturbance are solved.

Description

Technical Field

[0001] The present application relates to the field of servo control of electromechanical systems, and in particular to a fully saturated learning motion control method, system, and device based on disturbance separation. Background Art

[0002] With the development of new-generation information technology, modern electronic manufacturing equipment is gradually moving towards refinement and intelligence. Direct-drive units are often used as core moving components of electromechanical systems due to their simple structure, high positioning accuracy, and strong scalability. In order to improve efficiency and ensure product quality consistency, actuators are usually required to repeatedly perform certain tasks in actual production processes, such as robotic assembly, CNC machining, and automated placement. However, the dynamics of servo systems often exhibit periodic characteristics, and non-periodic unknown disturbances also pose serious challenges to the motion performance of closed-loop systems. Therefore, research on the precise tracking control of repetitive operations of direct-drive units is in urgent need of improvement. Summary of the Invention

[0003] This application aims to address the problem that existing precision tracking control methods for repetitive operations of direct-drive units fail to fully analyze the system's periodic and non-periodic dynamic characteristics, and are unable to effectively compensate for and suppress disturbances. In response to the demand for refined repetitive operations of motion units in electromechanical systems, a fully saturated learning motion control method, system, and device based on disturbance separation are provided. These methods are used to achieve finite-time asymptotic convergence of tracking errors under the adverse influence of periodic and non-periodic parameters, while ensuring the stability of the closed-loop servo system and guaranteeing processing quality and production efficiency in the field of high-end manufacturing.

[0004] The first aspect of the present application provides a fully saturated learning motion control method based on disturbance separation, comprising:

[0005] Constructing a control model for a direct-drive motion unit, wherein the control model has periodic and aperiodic uncertainty characteristics;

[0006] Determining at least one of a compensation adaptive control law, an expected compensation control law, and a fast compensation term of the control model;

[0007] Determining one or both of a feedback control law and a robust control law of a control system of the direct-drive motion unit;

[0008] Determining a motion controller for the direct-drive motion unit according to the control model and the control system;

[0009] The direct-drive motion unit is controlled based on the motion controller to realize electromechanical servo precision motion control.

[0010] In one possible design, determining the motion controller of the direct-drive motion unit based on the control model and the control system includes:

[0011] The motion controller is formed by adding the compensation adaptive control law, the expected compensation control law and the fast compensation term of the control model and the feedback control law and the robust control law of the control system.

[0012] In one possible design, the control model for the direct-drive motion unit is constructed as follows:

[0013] Constructing a second-order dynamics model of the direct-drive motion unit;

[0014] The lumped error in the second-order dynamics model is decomposed into periodic and non-periodic parts to obtain the control model.

[0015] In a possible implementation, the expression of the above control model is:

[0016]

[0017] Among them, x1, x2 and are the position, velocity and acceleration of the direct drive motion unit, respectively,

[0018] S f (·) represents a nonlinear smooth function,

[0019] D p is a periodic function representing uncertainty,

[0020] D n is a non-periodic function representing uncertainty,

[0021] K is the electromagnetic torque coefficient, B and A are the velocity-related viscous and Coulomb friction coefficients, respectively, and M is the inertia information of the direct-drive motion unit.

[0022] In one possible design, determining the compensation adaptive control law of the control model includes:

[0023] An estimate of a periodic function that defines uncertainty;

[0024] The estimated value of the periodic function of the uncertainty is negated to obtain the compensated adaptive control law.

[0025] In one example, the estimated value of the periodic function of the uncertainty is defined above, and the estimated value of the periodic function of the uncertainty is taken as a negative value to obtain the compensated adaptive control law, including:

[0026] definition is the periodic function of uncertainty D p The estimated value of , designing the fully saturated repeated learning law:

[0027]

[0028] Wherein, μ(t) is a monotonically increasing function, T is the task period, for t∈[0,T], μ(0)=0 and μ(t)=1 are satisfied, l is the learning law gain coefficient, D p The saturation upper bound of , σ is an exponential switching function,

[0029] is a standard saturation function, and has the expression:

[0030]

[0031] The compensation adaptive control law υ of the control model a3 for:

[0032]

[0033] In one possible design, determining the desired compensation control law of the control model includes:

[0034] Define the nominal parameter estimates of the system;

[0035] The desired compensation control law is determined based on the nominal parameter estimates.

[0036] In one example, the above-mentioned definition of the estimated nominal parameters of the system; and determining the expected compensation control law according to the estimated nominal parameters include:

[0037] The nominal parameter estimates of the system are defined as Design of saturation adaptive law

[0038]

[0039] Where Γ is the learning update rate, Θ is the adaptive function; Proj(*) represents the standard projection operator;

[0040] Design the expected compensation control law υ of the control model a1 for:

[0041]

[0042] in,

[0043] and x dThe first and second derivatives of x d is the expected trajectory.

[0044] In one possible design, determining the fast compensation term of the control model includes:

[0045] defining a compensation residual of the control model;

[0046] The fast compensation term is determined according to the compensation residual.

[0047] In one example, defining the compensation residual of the control model and determining the fast compensation term of the control model according to the compensation residual includes:

[0048] The compensation residual of the control model is defined as:

[0049]

[0050] Among them, d and d t are the unknown constant parameter term and the time-varying parameter term, D p estimated value of;

[0051] Determine the fast compensation term υ of the control model a2 :

[0052]

[0053] in, is an estimated value of d, and satisfies:

[0054]

[0055] ψ is the parameter update rate, and are the upper and lower bounds of the update rate ψ, and σ is an exponential switching function.

[0056] In one possible design, the above-mentioned determination of the feedback control law of the control system of the direct-drive motion unit includes:

[0057] Acquiring system information of the control system;

[0058] The feedback control law is determined based on the system information.

[0059] In one example, determining the feedback control law based on the system information includes:

[0060] The feedback control law υ r1 The expression is:

[0061]

[0062] Among them, k1, k2, k3, and k4 are all positive feedback gain coefficients;

[0063] σ is an exponential switching function;

[0064] Both m and n are positive odd numbers and satisfy

[0065] e is the error of the direct drive motion unit control system, and e=x1-x d , x d is the expected trajectory.

[0066] In one possible design, the above-mentioned determination of the robust control law of the control system of the direct-drive motion unit includes:

[0067] determining aperiodic disturbance rejection of the control system;

[0068] The robust control law is determined for the non-periodic disturbance rejection.

[0069] In one example, determining the robust control law for the non-periodic disturbance suppression includes:

[0070] The robust control law υ r2 The expression is:

[0071]

[0072] in, And satisfy:

[0073] υ r2 σ≤0 and

[0074] and x d The first and second derivatives of x d is the expected trajectory;

[0075] σ is an exponential switching function, θ max and θ min are the maximum and minimum values of θ, is the upper bound of the update rate ψ, d t is the time-varying parameter term for model compensation residual, is the upper bound of the lumped error, η is a positive constant and d is an unknown constant parameter, is the estimated value of d.

[0076] Furthermore, the above exponential switching function σ is expressed as:

[0077]

[0078] Among them, e and Both are errors of the direct drive motion unit control system, and e=x1-x d , x d is the expected trajectory, is x d The first-order derivative of , x1 and x2 are the position and velocity of the direct-drive motion unit respectively;

[0079] Both λ1 and λ2 are positive constants;

[0080] Both m and n are positive odd numbers and satisfy

[0081] A second aspect of the present application provides a fully saturated learning motion control system based on disturbance separation, comprising:

[0082] A model building unit is used to: build a control model for a direct-drive motion unit, wherein the control model has periodic and non-periodic uncertainty characteristics;

[0083] a controller design unit, configured to: determine at least one of a compensation adaptive control law, a desired compensation control law, and a fast compensation term of the control model; determine one or both of a feedback control law and a robust control law of a control system of the direct-drive motion unit; and determine a motion controller of the direct-drive motion unit based on the control model and the control system;

[0084] A control unit is used to: control the direct-drive motion unit based on the motion controller to realize electromechanical servo precision motion control.

[0085] Specifically, the electromechanical servo precision motion control system may further include functional components that can implement the functional components corresponding to the various steps in the method described in the first aspect of the present application, which will not be elaborated here.

[0086] The third aspect of the present application provides a fully saturated learning motion control device based on disturbance separation, wherein the electromechanical servo precision motion control device includes a processor and a memory, wherein the memory stores at least one instruction, and the at least one instruction is loaded and executed by the processor to implement the fully saturated learning motion control method based on disturbance separation as described in the first aspect of the present application.

[0087] The fourth aspect of the present application provides a computer storage medium, in which at least one instruction is stored. The at least one instruction is loaded and executed by a processor to implement the full saturation learning motion control method based on disturbance separation as described in the first aspect of the present application.

[0088] Beneficial effects of this application:

[0089] On the one hand, this application proposes a fully saturated learning motion control method based on disturbance separation. This method is a fully saturated learning control method based on disturbance separation, which solves the control problems of low tracking accuracy and poor stability under periodic and non-periodic uncertainty disturbances. Compared with the existing technology, it has significant effects:

[0090] (1) A feedforward compensation term that does not depend on model information is constructed based on the fully saturated learning algorithm, reducing the complexity of system design;

[0091] (2) Constructing a hybrid sliding surface with exponential convergence characteristics to accelerate the stable convergence of the system;

[0092] (3) The combination of the fully saturated learning mechanism and the projection-type adaptive law solves the divergence risk in traditional repetitive control and has certain practical application value.

[0093] On the other hand, the fully saturated learning motion control method based on disturbance separation described in this application is different from the existing design process that considers disturbance suppression as a whole and is based on model information. This application uses a feedforward control law designed by a repetitive strategy and a fully saturated learning algorithm to achieve bounded estimation and compensation of periodic uncertainty. This process does not rely on model information and is easy to implement. It is suitable for repetitive operation scenarios with periodic parameter uncertainty and non-periodic unknown disturbances. In addition, the model parameter disturbance problem is solved by developing an adaptive law based on expected information. The remaining non-periodic uncertainties, residuals and disturbances are weakened and stabilized by the designed robust and feedback control laws. In particular, a sliding mode switching surface with an exponential term is constructed to accelerate the convergence time of the tracking error to the equilibrium point. Finally, different control strategies are used to conduct comparative experiments on a direct-drive motion platform, and the results are given to prove the advantages of the proposed method. BRIEF DESCRIPTION OF THE DRAWINGS

[0094] Figure 1 A flowchart of a fully saturated learning motion control method based on disturbance separation provided in an embodiment of the present application;

[0095] Figure 2 A principle block diagram of a fully saturated learning electromechanical servo precision motion control method based on disturbance separation provided in an embodiment of the present application;

[0096] Figure 3 A control principle diagram of a motion controller provided in an embodiment of the present application;

[0097] Figure 4 A motion curve diagram of cosine tracking provided in an embodiment of the present application;

[0098] Figure 5 This is a graph showing the error of cosine tracking provided in an embodiment of the present application;

[0099] Figure 6 A control curve diagram for cosine tracking provided in an embodiment of the present application;

[0100] Figure 7 A schematic diagram of a fully saturated learning motion control system based on disturbance separation provided in an embodiment of the present application;

[0101] Figure 8 A schematic diagram of a fully saturated learning motion control device based on disturbance separation provided in an embodiment of the present application;

[0102] Figure 9 A schematic diagram of a computer storage medium provided in an embodiment of the present application. DETAILED DESCRIPTION

[0103] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other in the absence of conflict.

[0104] Due to the nonlinear characteristics and unknown dynamic characteristics of actual physical mechanisms (friction, dead zone, cogging forces), the mathematical control models of direct-drive motion units inevitably contain modeling errors and unmodeled dynamics. This greatly reduces the performance of traditional control strategies designed based on model accuracy, making it difficult to meet the needs of refined production. In addition, unknown disturbances constantly affect the motion accuracy of direct-drive units during production operations and may even cause system instability or damage. Many existing classic control algorithms fail to fully utilize the system's historical periodic data, making it difficult to achieve periodic bounded compensation from a mechanistic perspective. At the same time, the adverse effects of non-periodic uncertainty disturbances greatly increase the difficulty of system analysis and refined control.

[0105] Based on the above analysis, existing technologies still face the following challenges:

[0106] 1) Traditional model-dependent control strategies, such as proportional-integral-derivative (PID) controllers and model predictive control, are unable to effectively handle periodic modeling errors caused by nonlinear system dynamics (friction dead zone, cogging, etc.);

[0107] 2) There is a lack of a unified stability analysis framework under complex disturbance scenarios, and parameter tuning relies on empirical trial and error.

[0108] In view of this, the present invention provides a method, system, device and medium for full saturation learning motion control based on disturbance separation, in order to solve the above problems. Figures 1 to 6 , the solution of the embodiment of this application is described in detail.

[0109] refer to Figure 1 An embodiment of the present application provides a fully saturated learning motion control method based on disturbance separation, including: constructing a control model for a direct-drive motion unit, which control model has periodic and non-periodic uncertainty characteristics; determining at least one of the compensation adaptive control law, the expected compensation control law and the fast compensation term of the control model; determining one or all of the feedback control law and the robust control law of the control system of the direct-drive motion unit; determining a motion controller of the direct-drive motion unit based on the control model and the control system; and controlling the direct-drive motion unit based on the motion controller to realize electromechanical servo precision motion control.

[0110] Specifically, determining a motion controller for a direct-drive motion unit based on a control model and a control system may include: adding the compensation adaptive control law, the desired compensation control law, and the fast compensation term of the control model, and the feedback control law and robust control law of the control system to form a motion controller. It should be noted that this is only one example of a motion controller configuration method, and the motion controller may also have other configuration methods. For example, one or two of the compensation adaptive control law, the desired compensation control law, and the fast compensation term of the control model may be added to the feedback control law and robust control law of the control system to form a motion controller; another example, one or two of the compensation adaptive control law, the desired compensation control law, and the fast compensation term of the control model may be added to one of the feedback control law and robust control law of the control system to form a motion controller; another example, one or two of the compensation adaptive control law, the desired compensation control law, and the fast compensation term of the control model may be added to one of the feedback control law and robust control law of the control system to form a motion controller. It is understood that there may be other ways to configure a motion controller, and this application is not limited to this.

[0111] In one embodiment, constructing a control model for a direct-drive motion unit may include: constructing a second-order dynamic model of the direct-drive motion unit; and performing periodic and non-periodic decomposition of a lumped error in the second-order dynamic model to obtain a control model.

[0112] In one embodiment, determining the compensation adaptive control law of the control model may include: defining an estimated value of a periodic function of uncertainty; and taking a negative value of the estimated value of the periodic function of uncertainty to obtain the compensation adaptive control law.

[0113] In one embodiment, determining the expected compensation control law of the control model may include: defining estimated values of nominal parameters of the system; and determining the expected compensation control law according to the estimated values of the nominal parameters.

[0114] In one embodiment, determining the fast compensation term of the control model may include: defining a compensation residual of the control model; and determining the fast compensation term according to the compensation residual.

[0115] In one embodiment, determining the feedback control law of the control system of the direct-drive motion unit may include: acquiring system information of the control system; and determining the feedback control law based on the system information.

[0116] In one embodiment, determining a robust control law for a control system of a direct-drive motion unit may include: determining a non-periodic disturbance rejection of the control system; and determining a robust control law for the non-periodic disturbance rejection.

[0117] The solution of the embodiment of the present application is to construct a motion controller based on a fully saturated learning method with disturbance separation, which can solve the problem that the existing precision tracking control method for repetitive operations of direct-drive units fails to fully analyze the periodic and non-periodic dynamic characteristics of the system and cannot effectively compensate and suppress disturbances.

[0118] To further introduce the solution of the embodiment of this application, Figure 2 The principle block diagram of a fully saturated learning electromechanical servo precision motion control method based on disturbance separation is provided. It can be understood that Figure 2 In the scheme shown, Figure 1 For the same or similar contents as shown in the scheme, please refer to the above Figure 1 The detailed description of the scheme shown will not be repeated in the following. Figure 2 The fully saturated learning electromechanical servo precision motion control method based on disturbance separation provided in the embodiment of the present application includes steps (1) to (5), wherein the numbering of each step does not necessarily limit the order in which they are executed. Each step is described in detail below:

[0119] Step (1) constructs a control model for a direct drive motion unit, wherein the control model has periodic and non-periodic uncertainty characteristics, and the specific operations include:

[0120] Considering the dynamic characteristics of the direct-drive motion unit under conventional tasks, a second-order dynamic model of the direct-drive motion unit is constructed:

[0121]

[0122] Among them, the parameters x1, x2, and M represent the position, velocity, acceleration and inertia information of the direct-drive motion unit respectively; K is the electromagnetic torque coefficient; υ is the control voltage input; the lumped error of the direct-drive motion unit control system is recorded as D; K f =Bx2+AS f (x2) is the friction damping of the moving unit, and B and A are the velocity-dependent viscous and Coulomb friction coefficients, respectively. represents a nonlinear smooth function.

[0123] Consider a direct drive motion unit performing repetitive work, with the task cycle being a constant T. Then we have:

[0124]

[0125] Among them, x d is a continuous function, representing the expected trajectory; without loss of generality, we have: d (0) = x d (T).

[0126] Taking into account the uncertainty of the direct drive motion unit control system (lumped error D), the lumped error in the second-order dynamic model is decomposed into periodic and non-periodic components under repeated tasks:

[0127] D(t)=D p (t)+D n (t)(1.3),

[0128] Among them, D p (t) is a periodic function representing uncertainty, and D p (t) = D p (tT); D n (t) represents a non-periodic function of uncertainty; in practical systems, uncertainty is bounded and and are all bounded constants.

[0129] Based on formula (1.1), formula (1.3) is simplified to obtain the control model of the direct-drive motion unit control system with periodic and non-periodic uncertainty characteristics:

[0130]

[0131] in, A parameterized vector representing model information.

[0132] Define the system tracking error e = x1-x d and Construct an exponential switching function σ to accelerate the convergence of the system:

[0133]

[0134] Among them, λ1 and λ2 are both positive numbers; m and n are both positive odd numbers, and satisfy

[0135] It can be seen that the purpose of this step is to design a high-performance controller to ultimately achieve convergence of the tracking error e and asymptotic stability of the system.

[0136] Step (2), defining an estimated value of a periodic function of uncertainty, and taking a negative value of the estimated value of the periodic function of uncertainty to obtain the compensated adaptive control law, specifically includes:

[0137] definition D p The estimated value of , and the estimation error is

[0138] To ensure that the learning law is bounded and the estimated value Within the preset range, a fully saturated repetitive learning law is designed based on historical control information:

[0139]

[0140] Among them, μ(t) is a monotonically increasing function, such as And for t∈[0,T], μ(0)=0 and μ(t)=1; l is the learning law gain coefficient; D p The saturation upper bound of is a standard saturation function, the specific form is:

[0141]

[0142] Under the action of the saturation function, the periodic uncertainty satisfies:

[0143]

[0144] It is worth noting that the fully saturated repetitive learning law designed above only requires a memory for recording the estimated value of the previous cycle, and the control structure does not increase the computational burden of the system.

[0145] Therefore, the design model compensation adaptive control law υ a3 The expression is:

[0146]

[0147] The purpose of this step is to design a fully saturated repetitive learning law for the control model (1.4) of the direct-drive motion unit with periodic and non-periodic uncertainty characteristics, so as to achieve effective compensation of periodic uncertainty.

[0148] Step (3) is to design a parameter estimation and projection type adaptive algorithm to perform dynamic compensation of model parameters, specifically including:

[0149] First, the nominal parameter estimates of the system are defined, and the expected compensation control law is determined according to the nominal parameter estimates, including:

[0150] The nominal parameter estimates of the system are defined as And the estimated error is

[0151] In order to realize the dynamic compensation of the model, a saturation adaptive law is designed

[0152]

[0153] Where Γ>0 is the learning update rate, Θ is the adaptive function; Proj(*) represents the standard projection operator, and the subscript This means that the standard projection operator has Participate in calculations.

[0154] In order to accelerate the convergence of model parameters, a parameter estimation algorithm based on covariance reset is constructed:

[0155]

[0156] Among them, ζ、 and ξ M are constants to be designed, representing the forgetting factor, normalization factor and upper bound of learning update rate respectively; tr(*) and δ max (*) respectively represent the trace and eigenvalue of the parameter *; the estimated error after the system linearization is and express The filtered regressor matrix,

[0157] The expected compensation control law υ of the control model in the design step 1 a1 for:

[0158]

[0159] in, is the expected regressor matrix.

[0160] Then, defining the compensation residual of the control model and determining the fast compensation term according to the compensation residual includes:

[0161] Define the compensation residual of the control model in step 1 as:

[0162]

[0163] And d and d t are unknown constant parameter terms and time-varying parameter terms respectively. Therefore, the fast compensation term υ of the control model in the design step 1 is a2 :

[0164]

[0165] Where, ψ is the parameter update rate; are the upper and lower bounds of the update rate ψ, respectively.

[0166] The purpose of this step is to design an adaptive control law for the system dynamics (1.4) in step 1 to achieve effective parameter estimation and model compensation.

[0167] Step (4) is to develop a robust and feedback controller to suppress non-periodic disturbances and stabilize the system, specifically including:

[0168] First, system information of the control system is obtained, and the feedback control law is determined based on the system information, including:

[0169] Based on the system information, design the feedback control law υ r1 :

[0170]

[0171] Among them, k1, k2, k3, and k4 are all positive feedback gain coefficients; by selecting appropriate parameters, the coefficients satisfy:

[0172]

[0173] in, and Nonlinear parameter ω e By the formula Obtained by calculation.

[0174] At the same time, determining the non-periodic disturbance suppression of the control system, and determining the robust control law for the non-periodic disturbance suppression, including:

[0175] Design a robust control law υ for non-periodic disturbance suppression r2 :

[0176]

[0177] in, Obviously, this inequality satisfies:

[0178] 1)υ r2 σ≤0;

[0179] 2) η is a positive constant and

[0180] θ max and θ min are the maximum and minimum values of θ, is the upper bound of the update rate ψ, d t is the time-varying parameter term for model compensation residual, is the upper bound of the lumped error, η is a positive constant and d is an unknown constant parameter, is the estimated value of d.

[0181] This step aims to design feedback controllers and robust controllers to minimize the impact of non-periodic disturbances and thus achieve system stabilization.

[0182] Step (5) integrates the control modules, proves the system stability and performs precise repeatable control, specifically including:

[0183] The motion controller is constructed by adding the compensation adaptive control law, the expected compensation control law and the fast compensation term of the control model and the feedback control law and the robust control law of the control system:

[0184] υ=υ a1 +υ a2 +υ a3 +υ r1 +υ r2 (5.1),

[0185] Among them, a1 and υ a2 are the model compensation terms based on parameter estimation and projection-type adaptive algorithms respectively; a3 is a fully saturated repetitive learning law to achieve effective compensation for periodic disturbances; r1 and υ r2 They are feedback and robust controllers based on system information, respectively, which are used to weaken and suppress the impact of non-periodic disturbances while ensuring the stability of the system.

[0186] Define the Lyapunov function V3(t)=V1(t)+V2(t), and we have:

[0187]

[0188] Taking the time derivative of V1 with respect to time t, and simplifying formulas (1.4), (1.5) and (5.1), we can obtain: in,

[0189] Taking the time derivative of V2 with respect to time t, we can simplify and get

[0190] Differentiate V3 and simplify to obtain in, and

[0191] From the comparison lemma we can see that the system is bounded.

[0192] Assume that after a period of time nT, the periodic disturbance and the non-periodic disturbance are eliminated ( and D n are all zero), redesign the Lyapunov function:

[0193]

[0194] Differentiate the above equation and simplify it by combining equation (5.2) to get By Barbalat's lemma, we can see that the system is asymptotically stable.

[0195] The control principle diagram of the motion controller constructed above can be referred to Figure 3 shown.

[0196] A specific embodiment of applying the above motion controller is given below:

[0197] Given the expected input as a cosine Tracking curve. Based on the identification experiment, the initial values of the model parameters are:

[0198] θ0=[0.23,0.16,0.08].

[0199] The remaining parameters of this controller are designed as follows: Sliding mode switching function Full saturation repeated learning law parameter setting Parameter estimation algorithm design ζ=0.02, ξ M =5000, and Γ(0)=diag[1000,1000,1000,1000]; projection operator parameter ψ=500, The feedback control law is The robust controller parameters are ρ = 0.15 and η = 0.02. The traditional PID control strategy parameters are designed as: k p =3750,k i =2800 and k d =55.

[0200] Figure 4Figure 2 shows the dynamic tracking curves for this application and the traditional PID control strategy. As can be seen from the curves, the tracking curve for this application is closer to the desired trajectory. This is because the proposed control strategy effectively compensates for and dynamically suppresses periodic and non-periodic disturbances, while also enabling rapid adaptive adjustment of system parameters, resulting in higher tracking accuracy.

[0201] Figure 5 Figure 2 shows the error curves of the two control methods during the tracking process. Compared with traditional PID control, the control strategy proposed in this application achieves smaller error tracking while ensuring system stability.

[0202] Figure 6 The display in the middle describes the control output curve.

[0203] The present application also provides an electromechanical servo precision motion control system, referring to Figure 7 As shown, the system includes:

[0204] Model building unit: used to build a control model of a direct-drive motion unit with periodic and non-periodic uncertainty characteristics.

[0205] A controller design unit: used to respectively design the compensation adaptive control law, the expected compensation control law and the fast compensation term of the control model; used to respectively design the feedback control law and the robust control law of the direct-drive motion unit control system; and also used to add the compensation adaptive control law, the expected compensation control law and the fast compensation term of the control model and the feedback control law and the robust control law of the direct-drive motion unit control system to form a motion controller of the direct-drive motion unit, and use the motion controller to realize electromechanical servo precision motion control.

[0206] It is understandable that the electromechanical servo precision motion control system may also include a device capable of achieving Figure 1-6 The functional components corresponding to each step in the scheme shown in the figure can be referred to the above related Figure 1-6 The detailed description is omitted here.

[0207] The present application also provides an electromechanical servo precision motion control device, referring to Figure 8 As shown, the electromechanical servo precision motion control device includes a processor and a memory, wherein the memory stores at least one instruction, and the at least one instruction is loaded and executed by the processor to implement the full saturation learning motion control method based on disturbance separation as provided in the embodiment of the present application above. Figure 1-6 The same or corresponding contents in the scheme shown can be referred to the above Figure 1-6 The detailed description is omitted here.

[0208] The present application also provides a computer storage medium, referring to Figure 9 As shown, the computer storage medium stores at least one instruction, which is loaded and executed by the processor to implement the full saturation learning motion control method based on disturbance separation as provided in the embodiment of the present application above. Figure 1-6 The same or corresponding contents in the scheme shown can be referred to the above Figure 1-6 The detailed description is omitted here.

[0209] Although the present application is described herein with reference to specific embodiments, it should be understood that these embodiments are merely illustrative of the principles and applications of the present application. It should be understood that many modifications may be made to the illustrative embodiments, and that other arrangements may be devised, without departing from the spirit and scope of the present application as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in ways other than those described in the original claims. It should also be understood that features described in conjunction with individual embodiments may be used in conjunction with other described embodiments.

Claims

1. A fully saturated learning motion control method based on disturbance separation, characterized in that: include: Constructing a control model for a direct-drive motion unit, wherein the control model has periodic and aperiodic uncertainty characteristics; Determining at least one of a compensation adaptive control law, an expected compensation control law, and a fast compensation term of the control model; Determining one or both of a feedback control law and a robust control law of a control system of the direct-drive motion unit; Determining a motion controller for the direct-drive motion unit according to the control model and the control system; The direct-drive motion unit is controlled based on the motion controller to realize electromechanical servo precision motion control.

2. The fully saturated learning motion control method based on disturbance separation according to claim 1 is characterized in that: Determining the motion controller of the direct-drive motion unit according to the control model and the control system includes: The motion controller is formed by adding the compensation adaptive control law, the expected compensation control law and the fast compensation term of the control model and the feedback control law and the robust control law of the control system.

3. The full saturation learning motion control method based on disturbance separation according to claim 1 or 2, characterized in that: The control model for the direct-drive motion unit is constructed, comprising: Constructing a second-order dynamics model of the direct-drive motion unit; The lumped error in the second-order dynamics model is decomposed into periodic and non-periodic parts to obtain the control model.

4. The full saturation learning motion control method based on disturbance separation according to claim 1 or 2, characterized in that: Determining the compensation adaptive control law of the control model includes: An estimate of a periodic function that defines uncertainty; The estimated value of the periodic function of the uncertainty is negated to obtain the compensated adaptive control law.

5. The full saturation learning motion control method based on disturbance separation according to claim 1 or 2, characterized in that: Determining the desired compensation control law of the control model includes: Define the nominal parameter estimates of the system; The desired compensation control law is determined based on the nominal parameter estimates.

6. The full saturation learning motion control method based on disturbance separation according to claim 1 or 2, characterized in that: Determining the fast compensation term of the control model includes: defining a compensation residual of the control model; The fast compensation term is determined according to the compensation residual.

7. The full saturation learning motion control method based on disturbance separation according to claim 1 or 2, characterized in that: Determining the feedback control law of the control system of the direct-drive motion unit includes: Acquiring system information of the control system; The feedback control law is determined based on the system information.

8. The full saturation learning motion control method based on disturbance separation according to claim 1 or 2, characterized in that: Determining the robust control law of the control system of the direct-drive motion unit includes: determining aperiodic disturbance rejection of the control system; The robust control law is determined for the non-periodic disturbance rejection.

9. A fully saturated learning motion control system based on disturbance separation, characterized in that: include: A model building unit is used to: build a control model for a direct-drive motion unit, wherein the control model has periodic and non-periodic uncertainty characteristics; a controller design unit, configured to: determine at least one of a compensation adaptive control law, a desired compensation control law, and a fast compensation term of the control model; determine one or both of a feedback control law and a robust control law of a control system of the direct-drive motion unit; and determine a motion controller of the direct-drive motion unit based on the control model and the control system; A control unit is used to: control the direct-drive motion unit based on the motion controller to realize electromechanical servo precision motion control.

10. A fully saturated learning motion control device based on disturbance separation, characterized in that: The electromechanical servo precision motion control device includes a processor and a memory, wherein the memory stores at least one instruction, and the at least one instruction is loaded and executed by the processor to implement a fully saturated learning motion control method based on disturbance separation as described in one of claims 1 to 8.

Images