Tea bud recognition and sorting control method based on deep learning and robust control algorithm

Through deep learning recognition model and robust control algorithm, combined with DELTA parallel robot arm, the problems of high cost and unstable quality caused by relying on labor in tea bud recognition and sorting are solved, and high-precision and efficient tea bud recognition and sorting are achieved.

CN119017381APending Publication Date: 2024-11-26VALLEY OF SCI & TECH OF CHINA
View PDF 7 Cites 0 Cited by

Patent Information

Application Number
CN202411160032.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-22
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

The existing tea bud identification and sorting mainly relies on labor, resulting in high production costs and difficult to guarantee the sorting quality, which may be possible for identification errors.

Method used

Using deep learning-based recognition model and robust control algorithm, the tea bud position is identified through the YOLOv3 detector, and the DELTA parallel robot arm is used for sorting, combined with the PD robust control algorithm to reduce the impact of uncertainty, improve sorting accuracy and Fast.

Benefits of technology

It realizes high-precision identification and sorting of tea buds, improves the accuracy and speed of sorting, reduces manual intervention, and reduces production costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119017381A_ABST
    Figure CN119017381A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of tea production, in particular to a tea bud recognition and sorting control method based on deep learning and a robust control algorithm. Comprising the steps of training a deep learning recognition model by a tea bud data set; performing DELTA on a kinematics inverse solution equation of the parallel mechanical arm; establishing a kinetic equation of the DELTA parallel mechanical arm; and controlling the DELTA parallel mechanical arm to sort tea buds by using a PD robust control algorithm. According to the invention, deep learning is applied to tea bud identification, and high-precision identification of tea buds is realized. A robust control algorithm is designed and is used for the DELTA parallel mechanical arm, so that the influence of uncertainty on the tracking performance of the DELTA parallel mechanical arm for tea bud sorting is eliminated, and the accuracy and rapidity of tea bud sorting are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of tea production, and in particular to a tea bud identification and sorting control method based on deep learning and robust control algorithms. Background Art

[0002] Tea buds are the youngest nutrient buds on the branches of tea trees. They are the main objects of tea picking. The proportion of active nutrient buds in the raw materials for tea making is one of the important indicators for measuring the quality of tea. Tea bud classification requires identification and sorting, which is a huge workload during the production and processing process.

[0003] At present, tea bud identification and sorting are mainly done manually, and its shortcomings are:

[0004] 1. It consumes a lot of manpower and has high production costs;

[0005] 2. Manual identification may result in errors and the sorting quality cannot be guaranteed. Summary of the invention

[0006] The invention discloses a tea bud identification and sorting control method based on deep learning and robust control algorithm.

[0007] It is achieved through such a technical solution, the specific method is as follows:

[0008] Use tea bud dataset to train deep learning recognition model;

[0009] Through the deep learning recognition model, the YOLOv3 detector is used to identify and locate the tea buds in the sorting area;

[0010] According to the tea bud location information, the DELTA parallel robotic arm is controlled to sort the tea buds.

[0011] Furthermore, the DELTA parallel robot is controlled to sort tea buds. The specific method is as follows:

[0012] Simplified DELTA parallel robot arm;

[0013] Establish the inverse kinematics equation of the DELTA parallel manipulator and solve the joint angles of the DELTA parallel manipulator;

[0014] Establish the dynamic equations of the DELTA parallel manipulator;

[0015] The PD robust control algorithm is used to control the DELTA parallel robot arm to sort tea buds.

[0016] Furthermore, before training the deep learning recognition model, the images of the tea bud dataset were rotated and mirrored.

[0017] Furthermore, the YOLOv3 detector is used to identify and locate the tea buds in the sorting area. The specific method is as follows:

[0018] Divide the input image into multiple grids as bounding boxes;

[0019] Predict the vertex position information and center point coordinates of each bounding box, as well as the width and height of the bounding box;

[0020] Decode the prediction result and get the position of the final prediction bounding box on the image;

[0021] Get the recognition results and location of the tea buds within the bounding box.

[0022] Furthermore, the inverse kinematics equation of the DELTA parallel manipulator is established as follows:

[0023] After simplifying the DELTA parallel robot, the control variables include the angles α1, α2, and α3 between each active arm and the X-axis;

[0024] The inverse kinematic equation of the DELTA parallel manipulator is established as:

[0025]

[0026]

[0027]

[0028] Where, L a is the active arm length, L b is the length of the follower arm, θ i is the angle between the active arm and the x-axis, R is the horizontal distance between the upper end of the active arm and the center point, r is the horizontal distance between the lower end of the driven arm and the center point, x, y, z are the spatial coordinates of the target, and z is a constant representing the height of the tea bud conveyor belt.

[0029] Furthermore, since the DELTA parallel manipulator has only three degrees of freedom θ1, θ2, and θ3, the dynamic analysis can be completed by listing three Lagrange equations based on three generalized coordinates. To simplify the solution process, three generalized coordinates P are added. x , P v , P z and the corresponding Lagrange multiplier λ i (i=1,2,3);

[0030] Three generalized coordinates P x , P v , P z The relationship with θ1, θ2, θ3 is:

[0031]

[0032] In the formula

[0033]

[0034] Where m1 and m2 are the equivalent masses of the active arm and the passive arm, a i1 ,a i2 ,a i3 ,m i The expressions for (i=1,2,3) are:

[0035]

[0036] The overall Jacobian matrix J of the mechanism is

[0037]

[0038] Furthermore, the dynamic equation of the DELTA parallel manipulator is established. The specific method is as follows:

[0039] Construct the dynamic equation of the DELTA parallel robot:

[0040]

[0041] In the formula

[0042]

[0043]

[0044]

[0045]

[0046] where M(θ) is a symmetric and positive definite inertia matrix, is the centrifugal force and Coriolis force matrix, G(θ) is the gravity matrix, θ, They represent the angle, angular velocity and angular acceleration between the active arm and the fixed platform respectively, and τ represents the driving torque of the main manipulator.

[0047] Furthermore, the dynamic equation is divided into a certain part without robust terms and an uncertain part with robust terms, and the uncertain part is controlled by the PD robust control algorithm.

[0048] Furthermore, the PD robust control algorithm is used to control the DELTA parallel robot arm to sort tea buds. The specific method is as follows:

[0049] Considering the influence of external interference and friction on the actual control object, the dynamic equation is rewritten as:

[0050]

[0051] The joint torque without considering the external force, is the torque composed of Coulomb friction and viscous friction:

[0052]

[0053] in is the viscous friction coefficient, represents the Coulomb friction coefficient

[0054] Coulomb friction is expressed by the tangent hyperbolic function approximating the sign function sign,

[0055]

[0056] Assume that the angular displacement of the desired trajectory of the active arm of the DELTA parallel manipulator is θ d (t), the first-order derivative angular velocity with respect to time is The second derivative of angular acceleration with respect to time is And it is uniformly bounded, let:

[0057] e(t)=θ(t)-θ d (t)

[0058] Then there is e(t) is the error between the actual angular displacement and the expected angular displacement, is the first-order derivative of e(t), is the second-order derivative of e(t);

[0059] make

[0060]

[0061] The kinetic equation is expressed as:

[0062]

[0063] Split the dynamic equation into a nominal part and an uncertain part:

[0064]

[0065] in and The nominal terms of the ideal situation are the nominal terms, while ΔM(·), ΔC(·), ΔF(·), and ΔG(·) are uncertain terms depending on σ, where σ is an uncertain parameter in the mechanical system;

[0066] choose and Nominal terms are used to estimate M(·), C(·), F(·) and G(·). For a given S = diag{[s i ] n×n},s i > 0, assuming that the boundary of model uncertainty is a scalar ρ, and putting all uncertainties into the function of φ, we have the following relationship:

[0067]

[0068] In the formula

[0069]

[0070] Ignoring uncertainty,

[0071] For any given ε>0 and a positive definite diagonal matrix K p and K v , the control algorithm designed for the nonlinear dynamics model is as follows:

[0072]

[0073] In the formula

[0074]

[0075]

[0076] K p =diag{[k pi ] n×n}

[0077] K v =diag{[k vi ] n×n}

[0078] The τ with p term is regarded as the PD robust control algorithm, and the τ without p term is regarded as the PD control algorithm without robust term.

[0079] Due to the adoption of the above technical solution, the present invention has the following beneficial effects:

[0080] 1. Apply deep learning to tea bud recognition to achieve high-precision recognition of tea buds.

[0081] 2. A robust control algorithm was designed and used in the DELTA parallel robot arm, eliminating the impact of uncertainty on the tracking performance of the DELTA parallel robot arm for tea bud sorting, and improving the accuracy and speed of tea bud sorting.

[0082] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be achieved and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] The accompanying drawings of the present invention are as follows.

[0084] Figure 1 Schematic diagram of the three-dimensional structure of the DELTA parallel robotic arm.

[0085] Figure 2 This is a simplified schematic diagram of the DELTA parallel robot arm.

[0086] Figure 3 for Figure 2 Top view of the .

[0087] Figure 4 for Figure 2 Front view of .

[0088] Figure 5 Schematic diagram of the PD robust control algorithm flow chart.

[0089] Figure 6 Schematic diagram of simulation comparison of the embodiment.

[0090] Figure 7 A three-dimensional schematic diagram of the mechanical gripper of the sorting robot.

[0091] Figure 8 This is the overall three-dimensional schematic diagram of the sorting robot.

[0092] Fig. 9 Schematic diagram of simulation results of the embodiment.

[0093] Fig.10 It is a schematic diagram of the overall process of the embodiment. DETAILED DESCRIPTION

[0094] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0095] A tea bud recognition and sorting control method based on deep learning and robust control algorithm, such as Fig.10 As shown, the specific steps are as follows:

[0096] S1. Train the deep learning recognition model using the tea bud dataset.

[0097] In step S1, the existing tea bud dataset is expanded, and the data enhancement method includes rotating the originally collected tea bud images by 90° and 180° clockwise, mirroring them horizontally and mirroring them vertically, thereby establishing a tea bud variety dataset.

[0098] S2. Through the deep learning recognition model, the YOLOv3 detector is used to identify and locate the tea buds in the sorting area.

[0099] In step S2, the YOLOv3 detector divides the input image into multiple grids. After convolution and upsampling, it can directly predict the category and bounding box of the object. The YOLOv3 model predicts four position information of each bounding box, namely the coordinates of the center point of the bounding box (t x , t y ), width t w and height t h , however, the prediction result does not correspond to the position of the final prediction box on the image, and decoding is required. The decoding process is shown in formula (1):

[0100]

[0101] In the formula,

[0102] (c x , c y ): The number of grids between the upper left corner of the grid where the point is located and the upper left corner.

[0103] (p w , p h ): The side length of the prior box.

[0104] (t x , t y ): The offset of the target center point relative to the upper left corner of the grid where the point is located.

[0105] (t w , t h ): predict the width and height of the bounding box.

[0106] (b x , b y ): The target center point coordinates of the final predicted bounding box.

[0107] (b w , b h ): The width and height of the final predicted bounding box.

[0108] σ: sigmoid activation function.

[0109] σ(t x ) and σ(t y ) is the offset based on the coordinates of the upper left corner of the rectangular center.

[0110] S3. Simplify the DELTA parallel robot arm.

[0111] DELTA parallel robot arm Figure 1 As shown, it is a typical parallel mechanism of spatial three-degree-of-freedom X, Y, and Z translational motion. Its characteristics are that its own weight load is relatively small compared with the robotic arm, it has good dynamic performance, strong bearing capacity, and fast running speed. It consists of three motors, three pairs of main robotic arms, and three pairs of auxiliary robotic arms.

[0112] The operating platform at the bottom of the Delta robot arm, i.e. the moving platform, has only translational motion but no rotation. It is fixed relative to the fixed platform, i.e. the static platform. The parallelogram frame composed of all the link arms in the mechanism is always a plane quadrilateral. Therefore, when performing motion analysis, the parallelogram structure can be replaced by a line connecting the midpoints of its upper and lower sides. The simplified diagram of the Delta robot arm mechanism is shown below: Figure 2 , Figure 3 and Figure 4 shown.

[0113] The fixed platform is centered on O, and the operating platform is centered on O'. The active arm is P i A i , length L a , the slave arm is A i B i , length L b , the coordinate system O-XYZ is established on the fixed platform, and the coordinate system O-X'Y'Z' is established on the moving platform, OX is parallel to O-X'. The clamp is located on the Y' axis, and the angular position relationship between the structural dimensions is shown in the figure below, θ i is the angle between the active arm and the static platform, and α1, α2, α3 are the angles between the active arm and the X-axis.

[0114] S4. According to the positioning of tea buds, the inverse kinematic equation of the DELTA parallel robot arm is established to solve the joint angles of the DELTA parallel robot arm.

[0115] The kinematic inverse equations of the DELTA parallel manipulator are shown in equations (2), (3) and (4):

[0116]

[0117] The coordinate system O-XYZ is established with the fixed platform center O, where x, y, and z represent the position coordinates of the center point of the tea bud, l is the offset of the gripper relative to the Y' axis, R represents the radius of the fixed platform, r represents the radius of the moving platform, and L a is the arm length, L b is the length of the main boom, and θ1, θ2, and θ3 represent the angles between the three main booms and the fixed platform, respectively.

[0118] Wherein, x and y can be obtained according to step 2 described in claim 1, z is a constant, which is the height of the tea bud conveyor belt. Substituting x, y, z into equations (2), (3) and (4) can obtain the angles θ1, θ2, θ3 between the three active arms and the fixed platform.

[0119] Since the Delta parallel robot has only three degrees of freedom θ1, θ2, θ3, the dynamic analysis can be completed by listing three Lagrange equations based on three generalized coordinates, but the solution analysis process will be quite complicated. Therefore, the system adds three generalized coordinates P x , P v , P z , the same number of Lagrange multipliers λ as those corresponding to the generalized coordinates i (i=1,2,3) is used to simplify the dynamic model. At this time, the system has 6 degrees of freedom.

[0120] P x , P v , P z It is not independent, and its relationship with θ1, θ2, and θ3 is

[0121]

[0122] In the formula

[0123]

[0124] Where m1 and m2 are the equivalent masses of the active arm and the passive arm, a i1 , a i2 , a i3 , m i The expressions for (i=1,2,3) are

[0125]

[0126] The overall Jacobian matrix J of the mechanism is

[0127]

[0128] S5. Establish the dynamic equation of the DELTA parallel robot.

[0129] In step S5, the dynamic equation of the DELTA parallel manipulator can be obtained according to the Lagrangian modeling method, which is generally written in the following standard form:

[0130]

[0131] Where:

[0132]

[0133]

[0134]

[0135]

[0136] where M(θ) is a symmetric and positive definite inertia matrix, is the centrifugal force and Coriolis force matrix, G(θ) is the gravity matrix, θ, They represent the angle, angular velocity and angular acceleration between the active arm and the fixed platform respectively, and τ represents the driving torque of the main manipulator.

[0137] Formula (7) can be considered as the driving torque of the main robot arm under ideal conditions, while the actual control object will be affected by external interference and friction. When the angular velocity of each robot arm is not zero, the torque of the main robot arm can be approximately considered to be affected by Coulomb friction and viscous friction. The dynamic equation of the robot arm is generally written as follows:

[0138]

[0139] Formula (8) does not consider the joint torque caused by external force. It is the torque composed of Coulomb friction and viscous friction.

[0140]

[0141] in is the viscous friction coefficient, represents the Coulomb friction coefficient.

[0142] In general, Coulomb friction is represented by a symbolic function sign. In order to avoid non-smooth functions in the dynamic model, the tangent hyperbolic function is used to approximate the symbolic function, as shown in formula (10):

[0143]

[0144] where λ is a number large enough so that the slope of the tangent hyperbolic function is steep around zero.

[0145] The robot arm dynamics equation of formula (6) has the following two basic properties:

[0146] (1) M(θ) is a positive definite symmetric matrix. For all generalized variables θ, the matrix M(θ) has upper and lower bounds:

[0147] χ min I≤M(θ)≤χ max I (11)

[0148] where χ min and χ max is a scalar that can be determined, and I is the identity matrix.

[0149] (2) Matrix is a skew-symmetric matrix, is the time derivative of M(θ) for x∈R n Both have:

[0150] x T Sx=0 (12)

[0151] S6. Use PD robust control algorithm to control the DELTA parallel robot arm to sort tea buds.

[0152] PD robust control algorithm such as Figure 5 As shown in the figure, considering the trajectory tracking control of the mechanical system, it is assumed that the angular displacement of the desired trajectory of the active arm of the DELTA parallel manipulator is θ d (t), the first-order derivative angular velocity with respect to time is The second derivative of angular acceleration with respect to time is And they are uniformly bounded.

[0153] e(t)=θ(t)-θ d (t) (13)

[0154] Then there is e(t) is the error between the actual angular displacement and the expected angular displacement, is the first-order derivative of e(t), is the second-order derivative of e(t).

[0155] make

[0156]

[0157] Formula (8) can be written as follows:

[0158]

[0159] In fact, due to the variation of model parameters and the unmodeled part of the dynamics and the uncertainty in the system, it is impossible to obtain an accurate model characteristic matrix. Therefore, the dynamic model characteristic matrix can be divided into a nominal part and an uncertain part.

[0160]

[0161] in and The nominal terms are ideal terms, while ΔM(·), ΔC(·), ΔF(·), and ΔG(·) are uncertain terms that depend on σ, which is an uncertain parameter in the mechanical system that may change over time.

[0162] To reduce the impact of parameter uncertainty on control performance, first select and Nominal terms are used to estimate M(·), C(·), F(·) and G(·). For a given S = diag{[s i ]n ×n},s i > 0, assuming that the boundary of model uncertainty is a scalar ρ, and putting all uncertainties into the function of φ, we have the following relationship:

[0163]

[0164] In the formula

[0165]

[0166] If all uncertainties are ignored, then

[0167] For any given ε>0 (usually a small value is chosen) and a positive definite diagonal matrix K p and K v , the control algorithm designed for the nonlinear dynamics model is as follows:

[0168]

[0169] In the formula

[0170]

[0171]

[0172] K p =diag{[k pi ] n×n}

[0173] K v =diag{[k vi ] n×n}

[0174] As shown in the figure, τ with p term is called model-based PD robust control algorithm MPDRC, and τ without p term is called PD control algorithm without robust term MPD. Note that from the structure of PD robust control algorithm, it can be seen that this control algorithm is not only based on the model, but also inherits the characteristics of PID control algorithm, and the robust term of p is used to estimate external disturbances and uncertainties in the system, and due to the form of piecewise function, when μ is very small and close to zero, it can prevent the robust term p from tending to infinity. These characteristics make this control algorithm theoretically have better dynamic performance than PID control algorithm and PD control algorithm without robust term.

[0175] Simulation 1:

[0176] In order to verify the effectiveness of the control algorithm, the PID control algorithm and the PD control algorithm without robust terms are compared, and the three algorithms are numerically simulated to track the step signal. The simulation results are shown in Figure 6 :

[0177] From the simulation results, it can be seen that the model-based PD robust control algorithm MPDRC has better tracking performance for the desired angular displacement than the PID control algorithm and the PD control algorithm MPD without robust terms.

[0178] Tea bud recognition and sorting robot based on deep learning and robust control algorithm Figure 7 and Figure 8 As shown in the figure, the simulation diagram of the step signal tracking by the model-based PD robust control algorithm, PID control algorithm and PD control algorithm without robust terms is shown in Fig. 9 shown.

[0179] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A tea bud identification and sorting control method based on deep learning and robust control algorithm, characterized in that: The specific method is as follows: Use tea bud dataset to train deep learning recognition model; Through the deep learning recognition model, the YOLOv3 detector is used to identify and locate the tea buds in the sorting area; According to the tea bud location information, the DELTA parallel robotic arm is controlled to sort the tea buds.

2. The tea bud identification and sorting control method based on deep learning and robust control algorithm as claimed in claim 1, characterized in that: Control the DELTA parallel robot to sort tea buds. The specific method is as follows: Simplified DELTA parallel robot arm; Establish the inverse kinematics equation of the DELTA parallel manipulator and solve the joint angles of the DELTA parallel manipulator; Establish the dynamic equations of the DELTA parallel manipulator; The PD robust control algorithm is used to control the DELTA parallel robot arm to sort tea buds.

3. The tea bud identification and sorting control method based on deep learning and robust control algorithm as claimed in claim 1, characterized in that: Before training the deep learning recognition model, the images of the tea bud dataset were rotated and mirrored.

4. The tea bud identification and sorting control method based on deep learning and robust control algorithm as claimed in claim 1, characterized in that: The YOLOv3 detector is used to identify and locate the tea buds in the sorting area. The specific method is as follows: Divide the input image into multiple grids as bounding boxes; Predict the vertex position information and center point coordinates of each bounding box, as well as the width and height of the bounding box; Decode the prediction result and get the position of the final prediction bounding box on the image; Get the recognition results and location of the tea buds within the bounding box.

5. The tea bud identification and sorting control method based on deep learning and robust control algorithm as claimed in claim 2, characterized in that: The method to establish the inverse kinematics equation of the DELTA parallel manipulator is as follows: After simplifying the DELTA parallel robot, the control variables include the angles α1, α2, and α3 between each active arm and the X-axis; The inverse kinematic equation of the DELTA parallel manipulator is established as: Where, L a is the active arm length, L b is the length of the follower arm, θ i is the angle between the active arm and the x-axis, R is the horizontal distance between the upper end of the active arm and the center point, r is the horizontal distance between the lower end of the driven arm and the center point, x, y, z are the spatial coordinates of the target, and z is a constant representing the height of the tea bud conveyor belt.

6. The tea bud identification and sorting control method based on deep learning and robust control algorithm as claimed in claim 5, characterized in that: Since the DELTA parallel manipulator has only three degrees of freedom θ1, θ2, and θ3, the dynamic analysis can be completed by listing three Lagrange equations based on three generalized coordinates. To simplify the solution process, three generalized coordinates P are added. x , P v , P z and the corresponding Lagrange multiplier λ i (i=1,2,3); Three generalized coordinates P x , P v , P z The relationship with θ1, θ2, θ3 is: In the formula Where m1 and m2 are the equivalent masses of the active arm and the passive arm, a i1 , a i2 , a i3 , m i The expressions for (i=1,2,3) are: The overall Jacobian matrix J of the mechanism is 。 7. The tea bud identification and sorting control method based on deep learning and robust control algorithm as claimed in claim 2, characterized in that: The dynamic equation of the DELTA parallel manipulator is established as follows: Construct the dynamic equation of the DELTA parallel robot: In the formula where M(θ) is a symmetric and positive definite inertia matrix, is the centrifugal force and Coriolis force matrix, G(θ) is the gravity matrix, θ, They represent the angle, angular velocity and angular acceleration between the active arm and the fixed platform respectively, and τ represents the driving torque of the main manipulator.

8. The tea bud identification and sorting control method based on deep learning and robust control algorithm as claimed in claim 7, characterized in that: The dynamic equation is divided into a certain part without robust terms and an uncertain part with robust terms. The uncertain part is controlled by the PD robust control algorithm.

9. The tea bud identification and sorting control method based on deep learning and robust control algorithm as claimed in claim 8, characterized in that: The PD robust control algorithm is used to control the DELTA parallel robot arm to sort tea buds. The specific method is as follows: Considering the influence of external interference and friction on the actual control object, the dynamic equation is rewritten as: The joint torque without considering the external force, is the torque composed of Coulomb friction and viscous friction: in is the viscous friction coefficient, represents the Coulomb friction coefficient Coulomb friction is expressed by the tangent hyperbolic function approximating the sign function sign, Assume that the angular displacement of the desired trajectory of the active arm of the DELTA parallel manipulator is θ d (t), the first-order derivative angular velocity with respect to time is The second derivative of angular acceleration with respect to time is And it is uniformly bounded, let: e(t)=θ(t)-θ d (t) Then there is e(t) is the error between the actual angular displacement and the expected angular displacement, is the first-order derivative of e(t), is the second-order derivative of e(t); make The kinetic equation is expressed as: Split the dynamic equation into a nominal part and an uncertain part: in and The nominal terms of the ideal situation are the nominal terms, while ΔM(·), ΔC(·), ΔF(·), and ΔG(·) are uncertain terms depending on σ, where σ is an uncertain parameter in the mechanical system; choose and Nominal terms are used to estimate M(·), C(·), F(·) and G(·). For a given S = diag{[s i ] n×n },s i > 0, assuming that the boundary of model uncertainty is a scalar ρ, and putting all uncertainties into the function of φ, we have the following relationship: In the formula Ignoring uncertainty, For any given ε>0 and a positive definite diagonal matrix K p and K v , the control algorithm designed for the nonlinear dynamics model is as follows: In the formula K p =diag{[k pi ] n×n } K v =diag{[k vi ] n×n } The τ with p term is regarded as the PD robust control algorithm, and the τ without p term is regarded as the PD control algorithm without robust term.

Citation Information

Patent Citations

  • Robust control method for three-degree of freedom Delta parallel robot

    CN108693776A

  • Adaptive robust control method for three-degree-of-freedom Delta parallel robot

    CN109062039A

  • Delta robot control method based on fuzzy set theory

    CN110977971A

  • High-performance control method for cantilever type Stewart parallel mechanism of sand blasting derusting robot

    CN116237946A

  • Tea leaf impurity variable-diameter negative pressure removing suction nozzle, removing robot and removing method

    CN117086847A