Rope parallel robot vibration suppression control method based on mechanism model and data driving

Abstract

The invention discloses a cable parallel robot vibration suppression control method based on a mechanism model and data driving, and the method comprises the steps: carrying out the dynamic modeling of a cable-driven parallel robot, and solving the inherent frequency of each point in a working space; randomly selecting points in the working space, moving the movable platform to a selected position, exciting by using a counter-acting force device, and recording pose changes by using a vibration sensor; according to the actual inherent frequency of the vibration data identification system, taking the spatial position-inherent frequency residual data set as input to train a deep learning model describing the relationship between the position coordinates and the inherent frequency residual; and a controller is designed with the minimum inherent frequency vibration signal as a design target to perform vibration suppression control, and the inherent frequency of each position is obtained by summation of residual errors given by a dynamic model predicted value and a deep learning model. Through the method, different cable parallel robots can be rapidly inhibited, and the operation precision and the working stability of the tail ends of the cable parallel robots are effectively improved.

Description

Technical Field

[0001] The invention relates to the technical field of cable parallel robots, and in particular to a vibration suppression control method of a cable parallel robot based on a mechanism model and data drive. Background Art

[0002] The cable-driven parallel robot consists of multiple cables and a suspended rigid mobile platform. It has great application potential in application scenarios that require high-speed operation or large workspaces due to the characteristics of lightweight, low-cost, easy replacement and reconfiguration of the cables. However, compared with traditional rigid-link robots, the cable-driven parallel robot uses lightweight cables to connect the base platform and the mobile platform. Due to the flexibility of the cable structure and the unidirectional force limitation, it is susceptible to external interference, resulting in unexpected vibrations, affecting the operation accuracy and operation stability.

[0003] Model-based controllers are often used in existing research on vibration suppression control of cable-connected parallel robots. High accuracy is required for the dynamic modeling of cable-connected parallel robots. Too much neglect of factors such as the hysteresis, creep, and stiffness of the rope can easily lead to inaccurate system models, which in turn leads to poor vibration suppression effects. An overly complex modeling process will greatly increase the difficulty of calculation and lack feasibility in actual operations. Summary of the invention

[0004] In order to overcome the problems existing in the above-mentioned prior art, the purpose of the present invention is to provide a vibration suppression control method for a cable parallel robot based on a mechanism model and data-driven, use a reaction force device to excite for modal analysis, obtain the natural frequency residual to train a deep learning model, compensate the dynamic modeling results, and design a state-space controller such as a sliding mode controller with the goal of minimizing a certain order of natural frequency vibration signals, thereby avoiding overly complex dynamic modeling, and at the same time achieving rapid vibration suppression of the cable-driven robot, thereby improving the operating accuracy and operation stability of the cable-driven robot.

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

[0006] A vibration suppression control method for a cable parallel robot based on a mechanism model and data-driven includes the following steps: treating the rope as a massless spring with constant stiffness, ignoring minor factors to perform dynamic modeling on the cable parallel robot, and solving the natural frequency of each point in the workspace of the cable parallel robot; selecting a certain number of points in the workspace, moving the terminal moving platform of the cable parallel robot until the center of mass coincides with the selected position, using a reaction force device to excite, and using an inertial sensor unit to record the position and posture changes of the cable parallel robot during vibration; identifying the actual natural frequency of the cable parallel robot at the position according to the data recorded by the inertial sensor unit after excitation, and using the selected points to obtain the natural frequency of the cable parallel robot at the selected position. The position coordinates of a point and the natural frequency residual of the cable parallel robot at that point generate a data set; the spatial position-natural frequency residual data set is used as input to train a deep learning model that describes the relationship between the workspace position coordinates and the corresponding natural frequency residual, and the natural frequency solved by the dynamic model is corrected; according to the dynamic model of the cable parallel robot, a state space controller is designed with the goal of minimizing the natural frequency vibration signal of a certain order, and the reaction force device is driven to output a vibration suppression force to suppress the vibration of the cable parallel robot, where the natural frequency at each position is obtained by summing the predicted value of the dynamic model and the residual given by the deep learning model.

[0007] Alternatively, the rope is regarded as a massless spring with constant stiffness, and minor factors are ignored. The Newton-Euler method or Lagrangian method is used to model the dynamics of the cable parallel robot. The dynamic model is as follows:

[0008]

[0009] Where M is the mass matrix of the cable parallel robot, C is the damping matrix of the cable parallel robot, K is the stiffness matrix of the cable parallel robot, and F is the external load that changes with time. is the node acceleration vector, is the node velocity vector, q is the node displacement vector;

[0010] Without considering the damping effect and external force, assuming that the free vibration is a harmonic response, the following equation is obtained:

[0011] (KM -1 -ω 2 I)φ=0

[0012] KM -1 The generalized stiffness matrix, ω is the natural frequency of the cable parallel robot, φ is the vibration mode of the cable parallel robot, and I is the unit matrix; the dynamic model predicts the natural frequency of all points in the workspace of the cable parallel robot.

[0013] Optionally, the secondary factors include the weight of the rope itself, hysteresis effects, creep effects and pulley size.

[0014] Optionally, the reaction force device adopts an unbalanced rotary inertia actuator, a jet nozzle or a rotor-type pneumatic reaction force device;

[0015] According to needs, a reaction force device is used to cause vibration of the moving platform of the cable parallel robot, and an inertial sensor unit is used to record the posture changes of the moving platform in different directions.

[0016] Optionally, the step of identifying the actual natural frequency of the cable parallel robot vibrating at the position according to the data recorded by the inertial sensor unit after the excitation, and generating a data set by the position coordinates of the selected point and the residual of the natural frequency of the cable parallel robot at the point, comprises:

[0017] Preprocess the data collected by the inertial sensor unit, including filtering, denoising, and normalization operations;

[0018] Use Fourier transform to convert the collected time series data into frequency domain data, and use modal analysis technology to obtain the actual natural frequency of the cable parallel robot;

[0019] The natural frequency residual Δω of the cable parallel robot at that position is obtained by the actual natural frequency ω of the cable parallel robot and the predicted natural frequency ω' calculated based on the dynamic model, and a spatial position-natural frequency residual data set is generated.

[0020] Optionally, the modal analysis technique adopts a peak picking method, a curve fitting method or an autoregressive model.

[0021] Optionally, the spatial position-natural frequency residual data set is used as input to train a deep learning model describing the relationship between the workspace position coordinates and the corresponding natural frequency residuals to obtain

[0022]

[0023] Where q is the node displacement vector, are the deep learning model parameters.

[0024] The natural frequency residual of the cable parallel robot at each point in the workspace is obtained by the deep learning model, and the natural frequency solved by the dynamic model is corrected.

[0025] Optionally, the deep learning model adopts a multi-layer perceptron model, which specifically includes an input layer, several hidden layers, and an output layer; the input layer includes three neurons, representing the spatial position coordinates of the selected points, and the output layer includes one neuron, representing the natural frequency residual; the ReLU function is used as the activation function, the MSE function is used as the loss function, and Adam is used as the optimizer.

[0026] Optionally, according to the dynamic model of the cable parallel robot, a state space controller is designed with the goal of minimizing the vibration signal of a certain order natural frequency, and the reaction force device is driven to output a vibration suppression force to perform vibration suppression control on the cable parallel robot.

[0027] The terminal trajectory of the cable parallel robot is divided into several segments. For each segment of the trajectory, the average value of the natural frequencies of all path points is taken as the natural frequency of the segment. The goal is to minimize the natural frequency vibration signal of a certain order. According to the characteristics of the cable parallel robot, a controller is designed. The controller adopts pole configuration, linear quadratic regulator, linear quadratic Gaussian regulator, sliding mode controller, model predictive control, adaptive control or H∞ robust control.

[0028] Compared with the prior art, the present invention has the following advantages:

[0029] The cable parallel robot vibration suppression control method based on the mechanism model and data-driven of the present invention compensates and corrects the deviation between the dynamic model and the actual situation by means of deep learning, thus avoiding the complicated dynamic modeling and analysis process, and realizing the rapid vibration suppression of the end of the cable parallel robot, thereby improving the positioning accuracy and stability of the cable parallel robot.

[0030] The cable parallel robot vibration suppression control method based on the mechanism model and data drive of the present invention uses the same set of reaction force devices to achieve vibration excitation and vibration suppression functions, improves the integration of the cable parallel robot, and reduces the requirements on the load performance of the cable parallel robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is a flow chart of a vibration suppression control method of a cable parallel robot based on a mechanism model and data driven provided in an embodiment of the present application. DETAILED DESCRIPTION

[0032] The technical solution of the present invention is described clearly and completely below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present application, but cannot be understood as limiting the present invention.

[0033] Specifically, Figure 1 A schematic flow chart of a vibration suppression control method for a cable parallel robot based on a mechanism model and data driven provided in an embodiment of the present application.

[0034] like Figure 1 As shown, the vibration suppression control method of the cable parallel robot based on the mechanism model and data-driven includes the following steps:

[0035] In step S101, the rope is regarded as a massless spring with constant stiffness, and minor factors such as the size of the pulley are ignored. The cable parallel robot is dynamically modeled using the Newton-Euler method or the Lagrangian method, and the dynamic model of all points in the workspace is solved to predict the natural frequency.

[0036] In the actual implementation process,

[0037] The dynamic model of the cable-driven parallel robot is written as follows:

[0038]

[0039] Where M is the mass matrix of the cable parallel robot, C is the damping matrix of the cable parallel robot, K is the stiffness matrix of the cable parallel robot, and F is the external load that changes with time. is the node acceleration vector, is the node velocity vector, q is the node displacement vector;

[0040] Without considering the damping effect and external force, assuming that the free vibration is a harmonic response, the following equation can be obtained:

[0041] (KM -1 -ω 2 I)φ=0

[0042] KM -1 Generalized stiffness matrix, ω is the natural frequency of the cable parallel robot, φ is the vibration mode of the cable parallel robot, and I is the unit matrix.

[0043] Rewrite the above formula into the following form:

[0044] ω'=f(q,KM -1 )

[0045] That is, the natural frequency ω' predicted by the dynamic model is a function of the spatial position coordinates and the generalized stiffness matrix. Since the weight of the rope itself, the hysteresis effect, the creep effect and the effect of the pulley size on the cable exit point are not considered in the dynamic modeling process, there is a residual Δω between the natural frequency ω' predicted by the dynamic model and the actual natural frequency ω.

[0046] In step S102, a certain number of points are selected in the workspace, and the terminal moving platform of the cable parallel robot is moved until the center of mass coincides with the selected position. A reaction force device is used for excitation, and the posture changes of the cable parallel robot during vibration are recorded by an inertial sensor unit.

[0047] In the actual implementation process, considering the influence of the workspace boundary on the rope stiffness, the natural frequency of the end of the cable parallel robot changes more rapidly near the workspace boundary, and more points should be selected near the workspace boundary for excitation. Let G(p) be the probability that the point corresponding to the displacement vector p is selected, and p0 is the displacement vector corresponding to the geometric center of the workspace. It is required that G(p) obeys the Cauchy distribution relative to 1 / ||p-p0||.

[0048] In step S103, the actual natural frequency of the cable parallel robot at the position is identified based on the data recorded by the inertial sensor unit after the excitation, and a data set is generated by the position coordinates of the selected point and the residual of the natural frequency of the cable parallel robot at the point.

[0049] In the actual implementation process, it is necessary to first preprocess the inertial sensor unit data collected under certain excitation conditions, including filtering, denoising, normalization and other operations, to obtain the data of vibration direction displacement x, velocity v, acceleration a or angular displacement θ, angular velocity ω and angular acceleration α changing with time.

[0050] According to the Fourier transform formula:

[0051]

[0052] The collected displacement x, velocity v, acceleration a or angular displacement θ, angular velocity ω and angular acceleration α time series data are converted into frequency domain data using Fourier transform. The actual natural frequency ω of the cable parallel robot is extracted using modal analysis techniques such as peak picking method, curve fitting method and autoregressive model. The selected point position p is taken as the sample, and the corresponding natural frequency residual Δω is used as the label and added to the dataset in the form of (p, Δω).

[0053] In step S104, the spatial position-natural frequency residual data set is used as input to train a deep learning model that describes the relationship between the workspace position coordinates and the corresponding natural frequency residuals, and the natural frequency solved by the dynamic model is corrected.

[0054] Optionally, a multi-layer perceptron model is used for learning, which specifically includes an input layer, several hidden layers, and an output layer. The input layer includes three neurons, representing the spatial position coordinates of the selected points, and the output layer includes one neuron, representing the natural frequency residual. The number of hidden layers and the number of neurons in each layer can be set according to the actual working scenario, and no specific restrictions are made here.

[0055] Use the ReLU function as the activation function, the MSE function as the loss function, and Adam as the optimizer. During the execution, the technician sets the initial learning rate according to the actual working scenario. Divide the spatial position-natural frequency residual data set into a training set and a test set in proportion, such as 80% as a training set and 20% as a test set. Train the model several times, and evaluate the model performance on the validation set after each round of training. Adjust the model structure and hyperparameters based on the results on the test set until satisfactory prediction accuracy is obtained.

[0056] In step S105, according to the dynamic model of the cable parallel robot, a state space controller such as a sliding mode controller is designed to minimize the vibration signal of a certain order natural frequency, and the reaction force device is driven to output a vibration suppression force to perform vibration suppression control on the cable parallel robot.

[0057] In the actual implementation process, the dynamic equation of the cable parallel robot needs to be converted into a state space expression form, which is as follows:

[0058]

[0059] Where x(t) is the state vector, is the first-order derivative of the state vector with respect to time, u(t) is the controlled variable, y(t) is the output variable, A is the system matrix, B is the input matrix, C is the output matrix, and D is the direct matrix.

[0060] The controller is designed according to the state-space equation. It should be pointed out that the embodiment is illustrated by taking sliding mode control as an example. In the actual operation process, various controllers such as pole configuration, linear quadratic regulator, linear quadratic Gaussian regulator, model predictive control, adaptive control, H∞ robust control, etc. can be used, and there is no limitation here.

[0061] Specifically, the sliding surface S is designed to satisfy the following form:

[0062] S=c1x1+x2

[0063] Among them, x1 and x2 are the corresponding components of the state vector, and c1 is the design parameter of the sliding surface, which is determined according to the dynamic equation of the actual system. The reaching law is designed according to the needs, and the stability of the cable parallel robot is judged by the Lyapunov function, so as to obtain the specific parameter range of the designed controller.

[0064] It is worth noting that although the technical solutions and embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings, the present invention is not limited to the above-mentioned specific implementation modes. The above-mentioned implementation modes are merely illustrative. Relevant technical personnel in the field can be inspired by the present invention and make many forms without departing from the purpose of the present invention and the scope of protection of the claims. These all belong to the scope of protection of the present invention.

Claims

1. A vibration suppression control method for a cable parallel robot based on a mechanism model and data driven, characterized in that: The following steps are involved: The cable parallel robot is modeled dynamically by treating the rope as a massless spring with constant stiffness and ignoring minor factors, and the natural frequency of each point in the workspace of the cable parallel robot is solved; a certain number of points are selected in the workspace, and the end moving platform of the cable parallel robot is moved to coincide with the selected position with the center of mass, and the reaction force device is used for excitation, and the inertial sensor unit is used to record the posture changes of the cable parallel robot during the vibration process; the actual natural frequency of the cable parallel robot at this position is identified according to the data recorded by the inertial sensor unit after excitation, and a data set is generated by the position coordinates of the selected point and the natural frequency residual of the cable parallel robot at this point; the spatial position-natural frequency residual data set is used as input to train a deep learning model that describes the relationship between the workspace position coordinates and the corresponding natural frequency residual, and the natural frequency solved by the dynamic model is corrected; according to the dynamic model of the cable parallel robot, a state space controller is designed with the goal of minimizing the natural frequency vibration signal of a certain order, and the reaction force device is driven to output a vibration suppression force to suppress the vibration of the cable parallel robot, where the natural frequency at each position is obtained by summing the predicted value of the dynamic model and the residual given by the deep learning model.

2. The method according to claim 1, characterized in that The rope is regarded as a massless spring with constant stiffness. Ignoring minor factors, the Newton-Euler method or Lagrange method is used to model the dynamics of the cable parallel robot. The dynamic model is as follows: Where M is the mass matrix of the cable parallel robot, C is the damping matrix of the cable parallel robot, K is the stiffness matrix of the cable parallel robot, and F is the external load that changes with time. is the node acceleration vector, is the node velocity vector, q is the node displacement vector; Without considering the damping effect and external force, assuming that the free vibration is a harmonic response, the following equation is obtained: (KM -1 -ω 2 I)φ=0 KM -1 The generalized stiffness matrix, ω is the natural frequency of the cable parallel robot, φ is the vibration mode of the cable parallel robot, and I is the unit matrix; the dynamic model predicts the natural frequency of all points in the workspace of the cable parallel robot.

3. The method according to claim 2, characterized in that The secondary factors include the weight of the rope itself, hysteresis effects, creep effects and pulley size.

4. The method according to claim 1, characterized in that: The reaction force device adopts an unbalanced rotary inertia actuator, a jet nozzle or a rotor-type pneumatic reaction force device; According to needs, a reaction force device is used to cause vibration of the moving platform of the cable parallel robot, and an inertial sensor unit is used to record the posture changes of the moving platform in different directions.

5. The method according to claim 1, characterized in that According to the data recorded by the inertial sensor unit after excitation, the actual natural frequency of the cable parallel robot at the position is identified, and a data set is generated by the position coordinates of the selected point and the residual of the natural frequency of the cable parallel robot at the point, including: Preprocess the data collected by the inertial sensor unit, including filtering, denoising, and normalization operations; Use Fourier transform to convert the collected time series data into frequency domain data, and use modal analysis technology to obtain the actual natural frequency of the cable parallel robot; The natural frequency residual Δω of the cable parallel robot at that position is obtained by the actual natural frequency ω of the cable parallel robot and the predicted natural frequency ω' calculated based on the dynamic model, and a spatial position-natural frequency residual data set is generated.

6. The method according to claim 5, characterized in that The modal analysis technique adopts a peak picking method, a curve fitting method or an autoregressive model.

7. The method according to claim 1, characterized in that The spatial position-natural frequency residual dataset is used as input to train a deep learning model that describes the relationship between the workspace position coordinates and the corresponding natural frequency residuals. Where q is the node displacement vector, are the deep learning model parameters. The natural frequency residual of the cable parallel robot at each point in the workspace is obtained by the deep learning model, and the natural frequency solved by the dynamic model is corrected.

8. The method according to claim 1, characterized in that The deep learning model adopts a multi-layer perceptron model, which specifically includes an input layer, several hidden layers, and an output layer; the input layer contains three neurons, representing the spatial position coordinates of the selected points, and the output layer contains one neuron, representing the natural frequency residual; the ReLU function is used as the activation function, the MSE function is used as the loss function, and Adam is used as the optimizer.

9. The method according to claim 1, characterized in that: According to the dynamic model of the cable parallel robot, a state space controller is designed with the goal of minimizing the vibration signal of a certain order natural frequency. The reaction force device is driven to output vibration suppression force to control the cable parallel robot. The terminal trajectory of the cable parallel robot is divided into several segments. For each segment of the trajectory, the average value of the natural frequencies of all path points is taken as the natural frequency of the segment. The goal is to minimize the vibration signal of a certain order of natural frequency. According to the characteristics of the cable parallel robot, a controller is designed. The controller adopts pole configuration, linear quadratic regulator, linear quadratic Gaussian regulator, sliding mode controller, model predictive control, adaptive control or H∞ robust control.

Images