A method for dynamic parameter identification of hybrid robot based on sensitivity analysis

Abstract

The present invention discloses a robot dynamic parameter identification method based on sensitivity analysis. The method sequentially performs position inverse solution, velocity inverse solution, and acceleration inverse solution on the robot. The robot dynamic model is established using the principle of virtual work. The inertial force and inertial moment of each component of the robot are equivalent to the corresponding key points. The dynamic model of the five-degree-of-freedom robot is linearized. The parameters to be identified are separated. The initial or final values ​​of position, velocity, and acceleration are all set to 0. During the motion process, the parameter constraints of the workspace, position, velocity, and acceleration are satisfied. The identification trajectory of each joint of the robot is planned according to a seventh-order polynomial. During the operation of the robot, the position information of each joint and the driving current of each servo motor are collected. The Sobol global sensitivity analysis method is used to perform parameter sensitivity analysis on the linearized dynamic model and identify the parameters. The present invention reduces the number of parameters to be identified and improves the identification efficiency and accuracy.

Description

Technical Field

[0001] The present invention relates to the field of industrial robot dynamics, and in particular to a method for identifying dynamic parameters of a hybrid robot based on sensitivity analysis. Background Art

[0002] With the development of the aerospace industry, the demand for large structural components continues to increase. These components are characterized by thin walls, large dimensions, easy deformation, high precision requirements, and large material removal volumes. Hybrid robotic machining is becoming increasingly popular for these components. However, high-precision machining using hybrid robots requires complex control strategies and accurate dynamic models. To obtain a more accurate dynamic model for the hybrid robot, dynamic parameter identification is a crucial step.

[0003] Dynamic parameters are primarily divided into the inertia parameters of the robot body and the friction parameters of the joints. While the inertia parameters can be automatically derived from the robot's 3D model using 3D modeling software, due to part machining errors and unavoidable assembly errors, the inertia parameters calculated using this method can have significant deviations, and friction parameters cannot be derived. A more feasible approach involves designing an experiment in which the robot moves along a predetermined trajectory and then using the collected and calculated joint position, velocity, acceleration, and torque information for identification. The identified inertia and friction parameters can truly reflect the robot's operating state.

[0004] While this experimental-based identification method can identify true kinetic parameters, the resulting results often couple multiple parameters together, preventing the ability to determine the value of a single kinetic parameter. Furthermore, due to the large number of parameters to be identified and the presence of noise in the collected data, low identification accuracy is often a problem. Summary of the Invention

[0005] The present invention provides a method for identifying dynamic parameters of a hybrid robot based on sensitivity analysis in order to solve the technical problems existing in the known technology.

[0006] The technical solution adopted by the present invention to solve the technical problems existing in the known technology is: a method for identifying the dynamic parameters of a hybrid robot based on sensitivity analysis, the hybrid robot is a five-degree-of-freedom hybrid robot, which includes a three-degree-of-freedom parallel mechanism and a series mechanism with two rotational degrees of freedom connected in series therewith, the three-degree-of-freedom parallel mechanism includes a moving platform, a first active arm, a second active arm, a third active arm, a driven support arm, a first rotating bracket and a second rotating bracket; wherein: the moving platform is hinged on the circumference of three active arms that are driven to extend and retract by a servo motor, namely the first active arm, the second active arm, and the third active arm; the rear end of the moving platform is fixed with a driven support arm; the series mechanism includes an A / C axis double swing head driven by a servo motor, wherein the C axis of the double swing head is The shaft is rotatably connected to the moving platform, and the A axis of the double swing head is connected to the tool; the first active arm is rotatably connected to the first rotating bracket through the first rotating shaft; the first rotating bracket is rotatably connected to the fixed bearing seat through the second rotating shaft; the second active arm, the third active arm, and the driven support arm are rotatably connected to the second rotating bracket through the third rotating shaft, the fourth rotating shaft, and the fifth rotating shaft respectively; the second rotating bracket is rotatably connected to the fixed bearing seat through the sixth rotating shaft; the C axis of the double swing head is rotatably connected to the A axis through the seventh rotating shaft; the axis of the first rotating shaft is perpendicular to the axis of the second rotating shaft; the axes of the third rotating shaft, the fourth rotating shaft, and the fifth rotating shaft are parallel to each other and perpendicular to the axis of the sixth rotating shaft; the center of the second rotating bracket is located at the intersection of the axis of the fifth rotating shaft and the axis of the sixth rotating shaft; the following steps are included:

[0007] Step 1: Perform position inverse solution, velocity inverse solution, and acceleration inverse solution for the hybrid robot in sequence. Then, the dynamic model of the hybrid robot is established using the principle of virtual work. The dynamic equation of the dynamic model is as follows:

[0008] F=Mξ a +Cξ v +G+τ f ;

[0009] in:

[0010] F=[F1 F2 F3 τ4 τ5] T ;

[0011]

[0012] Where:

[0013] F1 is the driving force of the first active arm;

[0014] F2 is the driving force of the second active arm;

[0015] F3 is the driving force of the third active arm;

[0016] τ4 is the driving torque of the C-axis;

[0017] τ5 is the driving torque of axis A;

[0018] f1 is the friction force of the moving pair of the first active arm;

[0019] f2 is the friction force of the moving pair of the second active arm;

[0020] f3 is the friction force of the moving pair of the third active arm;

[0021] τ f4 is the friction torque of the rotating pair of the C axis;

[0022] τ f5 is the friction torque of the revolute pair of axis A;

[0023] is the inertia matrix of the hybrid robot;

[0024] is the centrifugal force and Coriolis force matrix of the hybrid robot;

[0025] is the gravity vector of the hybrid robot;

[0026] is the generalized velocity of the tool tip of the rigid body at the end of the hybrid robot;

[0027] is the generalized acceleration of the tool tip of the rigid body at the end of the hybrid robot;

[0028] Step 2: Equivalent the inertial force and inertial moment of each component of the hybrid robot to the corresponding key points, linearize the dynamic model of the five-degree-of-freedom hybrid robot, separate the parameters to be identified, and obtain the following formula:

[0029] F = ΦP;

[0030] Where:

[0031] Φ is the system observation matrix of the hybrid robot;

[0032] P is the dynamic parameter to be identified of the hybrid robot;

[0033] Step 3: Assume that the initial or final values ​​of position, velocity, and acceleration are all 0, and that the parameter constraints of workspace, position, velocity, and acceleration are satisfied during the motion process. Plan the identification trajectory of each joint of the hybrid robot according to a seventh-order polynomial.

[0034] Step 4: Run the hybrid robot, collect the position information of each joint of the hybrid robot and the driving current of each servo motor during the operation, and obtain ξ v ,ξ a ,Φ,F;

[0035] Step 5: Use the Sobol global sensitivity analysis method to perform parameter sensitivity analysis on the linearized dynamic model, and divide the inertial parameters to be identified into three categories according to their sensitivity: high sensitivity parameters P high , low sensitivity parameter P low and zero sensitivity parameter P zero ; For zero sensitivity parameter P zero No processing is done, for low sensitivity parameter P low The three-dimensional software is used to automatically derive the values, and for the high sensitivity parameter P high Genetic algorithm is used for identification.

[0036] Furthermore, step 1 includes the following sub-steps:

[0037] Step 1-1, assume that: the intersection of the first active arm axis and the first rotating shaft axis is B1; the intersection of the second active arm axis and the third rotating shaft axis is B2; the intersection of the third active arm axis and the fourth rotating shaft axis is B3; the intersection of the driven support arm axis and the fifth rotating shaft axis is B4; the hinge point of the first active arm and the moving platform is A1; the hinge point of the second active arm and the moving platform is A2; the hinge point of the third active arm and the moving platform is A3; the intersection of the line connecting A2 and A3 and the C-axis axis of the double swing head is A4; the base coordinate system coordinate axes include the x0 axis, y0 axis, and z0 axis that are perpendicular to each other; B4 is the origin of the base coordinate system, B2 and B3 are located on the x0 axis; the displacement vector of A1 relative to B1 is q 3,1 ; The displacement vector of A2 relative to B2 is q 3,2 ; The displacement vector of A3 relative to B3 is q 3,3 ; The displacement vector of A4 relative to B4 is q 3,4 The angle of rotation of the C axis of the double swing head around the axis of the platform is θ 4,4 The angle of rotation of the A axis of the double swing head around the seventh axis is θ 5,4 The intersection of the seventh axis and the axis of the moving platform is P; the tool tip of the hybrid robot end rigid body is point G, and its position vector corresponding to the origin of the base coordinate system is r G The rotation angle of the tool axis around the base coordinate system x0 axis is α; the rotation angle around the base coordinate system y0 axis is β;

[0038] Perform inverse position solution on the hybrid robot, and use r G , α, β, and get q 3,1 ,q 3,2 ,q 3,3 ,q 3,4 ,θ 4,4 ,θ 5,4 ;

[0039] Step 1-2, perform inverse velocity and acceleration solutions for the hybrid robot, and use ξ vObtain: the extension and retraction speeds and corresponding accelerations of the first to third active arms, and the rotation speeds and corresponding accelerations of the C-axis and A-axis of the double swing head;

[0040] Step 1-3: Disassemble the first to third active arms and the driven support arms into several components; obtain the velocity and acceleration of the center of mass of each component from the extension and retraction speeds and corresponding accelerations of the first to third active arms, and the rotation speeds and corresponding accelerations of the C-axis and A-axis of the double swing head;

[0041] In steps 1-4, the inertial force and inertial moment of each component are calculated respectively, and then the dynamic model of the hybrid robot is established using the principle of virtual work.

[0042] Furthermore, step 1-2 includes the following sub-steps:

[0043] Step 1-2-1, by taking the derivative of the following formula, we can get the velocity v of point P: P ;

[0044] r G =r P +r PG ;

[0045] Where: r P is the position vector of point P; r PG is the position vector from point P to point G;

[0046] Step 1-2-2, take point P as the node, use the following closed-loop constraint equations to derive

[0047] r P =(q 3,4 +e)s 3,4 ,r P =b k +q 3,k s 3,k -a k +es 3,4 ,k=1~3;

[0048] Where: e is the distance from point A4 to point P; s 3,k , k = 1 to 3, corresponding to the unit vectors of the first, second, and third active arm axes; b k The origin O of the base coordinate system points to point B k vector of a k Point A4 points to point A k vector;

[0049] Step 1-2-3, according to the principle of angular velocity superposition, we get the following formula, and further get and

[0050]

[0051] Where: w 5,4 is the angular velocity of the tool tip of the rigid body at the end of the hybrid robot; w 3,4 is the angular velocity of the moving platform; s 4,4 is the unit vector of the C-axis rotation axis of the double swing head; s 5,4 is the unit vector of the A-axis rotation axis of the double swing head.

[0052] Furthermore, step 2 includes the following sub-steps:

[0053] Step 2-1, select the key nodes of each component;

[0054] Step 2-2: Calculate the generalized velocity of the key nodes corresponding to each component, and equate the inertia force and inertia moment at the center of mass of each component to the corresponding key nodes, thereby separating the inertia parameters of each component;

[0055] In step 2-3, the linearized dynamic model of the hybrid robot is established based on F and the generalized velocity, inertia force, and inertia moment of the key nodes using the principle of virtual work.

[0056] Furthermore, step 3 includes the following sub-steps:

[0057] Step 3-1, set the constraints for the identified trajectory as shown below:

[0058]

[0059]

[0060] Where:

[0061] t is the robot movement time;

[0062] q 3,k (t) is the displacement of the kth active arm moving pair at time t;

[0063] q 3,k_0 is the initial displacement of the kth active arm moving pair;

[0064] θ u,4 (t) is the rotation angle of the u-th active rotational pair at time t; when u=4, the u-th active rotational pair is the C-axis rotational pair of the double swing head;

[0065] θ u,4_0 is the initial rotation angle of the u-th active rotation pair; when u=5, the u-th active rotation pair is the A-axis rotation pair of the double swing head;

[0066] In step 3-2, the identification trajectory of each joint of the hybrid robot is planned according to the seventh-order polynomial model; wherein the seventh-order polynomial model is shown as follows:

[0067] x(t)=a0+a1t+a2t 2 +a3t 3 +a4t 4 +a5t 5 +a6t 6 +a7t 7 ;

[0068] x(t) is the displacement or rotation of the joint at time t;

[0069] a0~a7 are the coefficients of the terms from degree 0 to degree 7 of the seventh-order polynomial;

[0070] According to the constraints in step 3-1, the polynomial coefficients of each joint identification trajectory are calculated to obtain each joint identification trajectory.

[0071] Furthermore, the first to third active arms each include a sleeve, a telescopic rod, and a screw pair driven by a servo motor; the telescopic rod is driven by the screw pair to extend and retract relative to the sleeve; the sleeve of the first active arm is fixedly connected to the first rotating shaft; the sleeve of the second active arm is fixedly connected to the third rotating shaft; and the sleeve of the third active arm is fixedly connected to the fourth rotating shaft; step 4 includes the following sub-steps:

[0072] Step 4-1: Write the obtained joint identification trajectory into G code, and then input it into the controller of the hybrid robot to drive the movement of each joint of the hybrid robot;

[0073] Step 4-2: While the hybrid robot is running along the identified trajectory, the position data of the hinge points between the hybrid robot's telescopic rod and the moving platform and the corresponding servo motor current data are synchronously collected to generate position curves and current curves for each joint;

[0074] Step 4-3: Use a low-pass filter to filter the position curve of each joint, and then differentiate it in turn to obtain the velocity and acceleration of each joint;

[0075] Step 4-4, using the forward kinematics of the hybrid robot, calculate the position information of the end rigid body and the generalized velocity ξ of the end tool tip according to the position, velocity and acceleration of each joint v and generalized acceleration ξ a ;

[0076] Step 4-5: Based on the sampled servo motor currents for driving the C-axis and A-axis of the double swing head, the corresponding driving torque τ for driving the C-axis and A-axis is obtained according to the following formula: d :

[0077] τd =r e k a i m ;

[0078] Where: k a is the torque constant of the servo motor; r e is the reduction ratio of the reducer connected to the servo motor; i m is the servo motor current;

[0079] Then, based on the sampled servo motor currents that drive the first to third active arms, the corresponding driving forces F that drive the first to third active arms are obtained according to the following formula: d :

[0080]

[0081] Where: b is the lead of the screw;

[0082] The calculated driving force and driving torque are filtered using a low-pass filter;

[0083] In steps 4-6, the position information of the end rigid body and the generalized velocity and generalized acceleration of the end tool tip are substituted into the expression of the measurement matrix Φ to obtain the result of the measurement matrix Φ.

[0084] Furthermore, step 5 includes the following sub-steps:

[0085] Step 5-1: Map the inertia parameters to be identified to the unit interval according to their respective value ranges, and then decompose the linearized dynamic model into the following function form with respect to the inertia parameters to be identified:

[0086]

[0087] P=[x1 x2 ... x n ] T ;

[0088] Where:

[0089] f0 is a constant;

[0090] n is the number of parameters;

[0091] x i is the i-th parameter to be identified;

[0092] x j is the jth parameter to be identified;

[0093] f i (x i ) contains only the parameters to be identified x i Function subitem of ;

[0094] f i,j (x i ,x j ) contains the parameters to be identified x i and x j Function subitem of ;

[0095] f 1,2,…,n (x1,x2,…,x n ) contains the parameters to be identified x1, x2,…, x n Function subitem of ;

[0096] All sub-items decomposed from the dynamic model are orthogonal to each other, so:

[0097]

[0098] I n =(P|0≤x i ≤1; i=1,2,…,n);

[0099] Where:

[0100] I n The n-dimensional unit body mapped out by the parameters to be identified;

[0101] s is the number of parameters to be identified contained in the function sub-term to be orthogonalized;

[0102] e1,e2…,e s Corresponding to the 1st, 2nd,…, sth parameters in the parameter set e to be identified;

[0103] h1, h2…, h s Corresponding to the 1st, 2nd,…, sth parameters in the parameter set h to be identified;

[0104] is a function sub-item containing the 1st, 2nd, …, sth parameters in the parameter set e to be identified;

[0105] is a function sub-item containing the 1st, 2nd, …, sth parameters in the parameter set h to be identified;

[0106] In step 5-2, multiple integrals are taken on the expression after the linearized dynamic model decomposition to obtain the following sub-terms:

[0107]

[0108] Where:

[0109] P -i Divide by x i Other parameters except

[0110] P -(ij) Divide by x i and x j Other parameters except

[0111] f j (x j ) contains the parameter x j Function subitem of ;

[0112] And so on, to obtain other sub-items;

[0113] Step 5-3, according to the sub-items solved in step 5-3, the total variance and each partial variance of the linearized dynamic model function are obtained:

[0114]

[0115]

[0116] Among them, the variance and total variance have the following characteristics:

[0117]

[0118] x s is the sth parameter;

[0119] D is the total variance of the linearized kinetic model function;

[0120] D i is the parameter x i The partial variance of

[0121] D ij is the parameter x i and x j The partial variance of

[0122] D 1,2,…,s For parameters x1, x2, …, x s The partial variance of

[0123] D 1,2,…,n For parameters x1, x2, …, x n The partial variance of

[0124] Step 5-4: Based on the total variance and each variance obtained in step 3, calculate the sensitivity of each order of all parameters according to the following formula:

[0125]

[0126] The parameter x is calculated by the following formula: i The total sensitivity coefficient is:

[0127]

[0128]

[0129] Where:

[0130] S Ti is the parameter x i The total sensitivity coefficient of

[0131] D -i Except for the parameter x i In addition, the sum of the partial variances of other parameters;

[0132] Step 5-5, perform two independent samplings on the parameter variables to obtain two matrices A N×n and B N×n , where N is the number of samples and n is the number of parameters; then use the Monte Carlo method to calculate according to the following formula and

[0133]

[0134] Where:

[0135] is the new sample matrix obtained by replacing the vth column in sample A with the vth column element in B;

[0136] is the new sample matrix obtained by replacing the vth column in sample B with the vth column element in A;

[0137] f(A) w is the dynamic model function value corresponding to sample A;

[0138] For samples The corresponding dynamic model function value;

[0139] For samples The corresponding dynamic model function value;

[0140] is the estimated value of f0 using the Monte Carlo method;

[0141] is the estimated value of the total variance of the kinetic model function using the Monte Carlo method;

[0142] To use Monte Carlo method to calculate the parameter x v Estimates of partial variance;

[0143] To use Monte Carlo method to calculate the parameter x vIn addition, the estimated value of the sum of partial variances of other parameters;

[0144] Step 5-6: The sensitivity values ​​of the inertial parameters to be identified are divided into three categories; the first 30% parameters in the sensitivity value range are selected and defined as high sensitivity parameters P high , the last 20% parameters in the sensitivity value range are defined as zero sensitivity parameters P zero , the remaining parameters are defined as low sensitivity parameters P low ;

[0145] Step 5-7, set the inertia parameter P to be identified high and friction parameters are the population of the genetic algorithm, and the objective function is as follows:

[0146]

[0147] Where: Q is the number of samples;

[0148] q is the sampling number;

[0149] τ qa is the qth actual sampling moment;

[0150] τ qt is the theoretical calculated torque corresponding to the qth actual sampling torque;

[0151] In steps 5-8, based on the fitness value, the population is continuously updated iteratively using crossover, inheritance, and mutation characteristics to obtain the population with the optimal fitness, which is the parameter identified by the genetic algorithm.

[0152] Furthermore, the method further includes step 6, verifying and evaluating the dynamic model through experiments based on the identified parameters, and correcting the dynamic model of the hybrid robot.

[0153] Furthermore, step 6 includes the following sub-steps:

[0154] Step 6-1: Calculate the driving force and driving torque of each joint using the identified dynamic parameters, draw a theoretical prediction curve, and compare it with the collected driving force and driving torque data;

[0155] In step 6-2, the obtained dynamic model is evaluated using the following root mean square error:

[0156]

[0157] Where:

[0158] Q is the number of samples;

[0159] q is the sampling number;

[0160] τqa is the qth actual sampling moment;

[0161] Φ is the observation matrix;

[0162] are the identified kinetic parameters.

[0163] Further, make These include 70 robot body inertia parameters and 10 joint friction parameters.

[0164] The advantages and positive effects of the present invention are as follows: the present invention uses the Sobol global sensitivity analysis method to perform parameter sensitivity analysis on the linearized dynamic model, and divides the inertial parameters to be identified into high sensitivity parameters P high , low sensitivity parameter P low And the zero sensitivity parameter P zero , the zero-sensitivity parameters are not processed, the low-sensitivity parameters use the values ​​derived from the three-dimensional software, and the genetic algorithm is used to identify the high-sensitivity parameters and friction parameters, which reduces the parameters to be identified and improves the identification efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0165] Figure 1 It shows the front view of the three-dimensional structure of the hybrid robot in the present invention.

[0166] Figure 2 A rear view of the three-dimensional structure of the hybrid robot according to the present invention is shown.

[0167] Figure 3 A schematic diagram showing the mechanism of the hybrid robot of the present invention.

[0168] Figure 4 It represents the identified trajectory curves of each joint of the hybrid robot parallel mechanism in the present invention.

[0169] Figure 5 It represents the identified trajectory curves of each joint of the serial mechanism of the hybrid robot in the present invention.

[0170] Figure 6 Represents the fitness value change curve.

[0171] Figure 7 Represents the driving force collected by the first active arm of the hybrid robot and the calculated driving force prediction curve.

[0172] Figure 8 The driving force collected by the second active arm of the hybrid robot and the calculated driving force prediction curve are shown.

[0173] Figure 9 The driving force collected by the third active arm of the hybrid robot and the calculated driving force prediction curve are shown.

[0174] Figure 10 The curve represents the driving torque collected from the C-axis of the dual swing head of the hybrid robot and the calculated driving torque prediction curve.

[0175] Figure 11 The curve represents the collected driving torque of the A-axis of the double swing head of the hybrid robot and the calculated driving torque prediction curve.

[0176] Figure 1 、 Figure 2 middle:

[0177] 1: Second fixed shaft seat; 2: First fixed shaft seat; 3: First servo motor; 4: Second servo motor; 5: Third servo motor; 6: First active arm; 7: Second active arm; 8: Third active arm; 9: First ball joint; 10: Second ball joint; 11: Third ball joint; 12: Second rotating bracket; 13: Driven support arm; 14: Moving platform; 15: C-axis of double swing head; 16: A-axis of double swing head; 17: Spindle; 8: Tool; 19: First rotating bracket.

[0178] Figure 3 middle:

[0179] O(B4): The intersection of the axis of the driven support arm and the axis of the rotating bracket;

[0180] A i (i=1,2,3): corresponds to the center points of the first, second and third spherical joints;

[0181] A4: The midpoint between points A2 and A3 on the moving platform;

[0182] B i (i=1,2,3): corresponds to the intersection of the first, second and third active arm axes and the rotation bracket axis;

[0183] P: The intersection of the seventh rotating shaft axis and the moving platform axis;

[0184] G: The tip of the tool;

[0185] x0, y0, z0: correspond to the x-axis, y-axis and z-axis of the base coordinate system established with point O as the origin;

[0186] s 3,i (i=1-4): corresponds to the unit vectors of the first, second, and third active arm axes and the driven support arm axis;

[0187] s 4,4 : is the unit vector of the C-axis rotation axis of the double swing head;

[0188] s 5,4: is the unit vector of the A-axis rotation axis of the double swing head;

[0189] a i (i=1,3): corresponds to the vector from point A4 to points A1 and A3;

[0190] b i (i=1,3): corresponds to the vector from point O to points B1 and B3;

[0191] q 3,i (i=1~4): corresponds to the displacement of the first, second and third active arm axes and the moving pair of the driven support arm (B i Point to A i distance between points);

[0192] θ 4,4 : is the rotation angle of the C axis of the double swing head around the axis of the platform;

[0193] θ 5,4 : is the rotation angle of the A-axis of the double swing head around the seventh axis. DETAILED DESCRIPTION

[0194] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention and are not used to limit the present invention.

[0195] In the description of the present invention, the terms "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", etc., indicating directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and do not require that the present invention must be constructed and operated in a specific direction. Therefore, they should not be understood as limitations on the present invention. The terms "connected" and "connected" used in the present invention should be understood in a broad sense. For example, they can be fixed connections or detachable connections; they can be directly connected or indirectly connected through intermediate components. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0196] See Figures 1 to 11 A method for identifying dynamic parameters of a hybrid robot based on sensitivity analysis is disclosed. The hybrid robot is a five-degree-of-freedom hybrid robot, which includes a three-degree-of-freedom parallel mechanism and a series mechanism with two rotational degrees of freedom connected in series. The three-degree-of-freedom parallel mechanism includes a moving platform 14, a first active arm 6, a second active arm 7, a third active arm 8, a driven support arm 13, a first rotating bracket 19, and a second rotating bracket 12; wherein:

[0197] The movable platform 14 is hinged on its circumference with three active arms that are driven to extend and retract by servo motors, namely the first active arm 6, the second active arm 7 and the third active arm 8; the first active arm 6 is connected to the movable platform 14 through the first ball joint 9; the second active arm 7 is connected to the movable platform 14 through the second ball joint 10; the third active arm 8 is connected to the movable platform 14 through the third ball joint 11; the first active arm 6 is driven to extend and retract by the first servo motor 3; the second active arm 7 is driven to extend and retract by the second servo motor 4; the third active arm 8 is driven to extend and retract by the third servo motor 5; a driven support arm 13 is fixed to the rear end of the movable platform 14; the series mechanism includes an A / C-axis double swing head driven by a servo motor, wherein the C-axis 15 of the double swing head is rotatably connected to the movable platform 14, the first rotary axis of the A / C-axis double swing head is the C-axis 15 of the double swing head, the second rotary axis is the A-axis 16 of the double swing head, and the A-axis 16 of the double swing head is connected to the tool 18 or to the main shaft 17 for installing the tool 18 Connection; the A / C-axis double swing head is connected in series at the end of the moving platform 14; the first active arm 6 is rotatably connected to the first rotating bracket 19 through the first rotating shaft; the first rotating bracket 19 is rotatably connected to the fixed bearing seat through the second rotating shaft; the two side ends of the first rotating bracket 19 are respectively rotatably connected to the first fixed shaft seat 2 through the second rotating shaft, and the second active arm 7, the third active arm 8, and the driven support arm 13 are correspondingly rotatably connected to the second rotating bracket 12 through the third rotating shaft, the fourth rotating shaft, and the fifth rotating shaft; that is: the second active arm 7 is rotatably connected to the second rotating bracket 12 through the third rotating shaft; the third active arm 8 is rotatably connected to the second rotating bracket 12 through the fourth rotating shaft; the driven support arm 13 is rotatably connected to the second rotating bracket 12 through the fifth rotating shaft; the second rotating bracket 12 is rotatably connected to the fixed bearing seat through the sixth rotating shaft; the two side ends of the second rotating bracket 12 are respectively rotatably connected to the second fixed shaft seat 1 through the sixth rotating shaft. The C-axis 15 of the double swing head is rotatably connected to the A-axis 16 of the double swing head through the seventh rotating shaft; the axis of the first rotating shaft is perpendicular to the axis of the second rotating shaft; the axes of the third rotating shaft, the fourth rotating shaft, and the fifth rotating shaft are parallel to each other and perpendicular to the axis of the sixth rotating shaft; the center of the second rotating bracket 12 is located at the intersection of the axis of the fifth rotating shaft and the axis of the sixth rotating shaft.

[0198] The first rotating bracket 19, the first active arm 6 and the first ball joint 9 are denoted as branch chain 1, and q 3,1 The displacement of the moving pair in the first active arm 6 is represented by F1. The second rotating bracket 12, the second active arm 7 and the second ball joint 10 are denoted as branch chain 2, and q 3,2 The displacement of the moving pair in the second active arm 7 is represented by F2. The second rotating bracket 12, the third active arm 8 and the third ball joint 11 are denoted as branch chain 3, and q 3,3 The displacement of the moving pair in the third active arm 8 is represented by F3. The second rotating bracket 12, the driven support arm 13, the moving platform 14, and the A / C axis double swing head are denoted as branch chain 4, and q3,4 The displacement of the moving pair in the driven support arm 13 is expressed by θ 4,4 The angle of rotation of the C-axis 15 of the double swing head around the axis of the platform 14 is represented by τ4, and the driving torque of the rotation pair is represented by θ 5,4 The angle of rotation of the revolving pair of the A-axis 16 of the double swing head around the C-axis 15 of the double swing head is represented by τ5, and the driving torque of the revolving pair is represented by τ5.

[0199] The method comprises the following steps:

[0200] Step 1: Perform position inverse solution, velocity inverse solution, and acceleration inverse solution for the hybrid robot in sequence. Then, the dynamic model of the hybrid robot is established using the principle of virtual work. The dynamic equation of the dynamic model is as follows:

[0201] F=Mξ a +Cξ v +G+τ f ;

[0202] in:

[0203] F=[F1 F2 F3 τ4 τ5] T ;

[0204]

[0205] Where:

[0206] F1 is the driving force of the first active arm 6;

[0207] F2 is the driving force of the second active arm 7;

[0208] F3 is the driving force of the third active arm 8;

[0209] τ4 is the driving torque of the C-axis;

[0210] τ5 is the driving torque of axis A;

[0211] f1 is the friction force of the moving pair of the first active arm 6;

[0212] f2 is the friction force of the moving pair of the second active arm 7;

[0213] f3 is the friction force of the moving pair of the third active arm 8;

[0214] τ f4 is the friction torque of the rotating pair of the C axis;

[0215] τ f5 is the friction torque of the revolute pair of axis A;

[0216]

[0217] is the inertia matrix of the hybrid robot;

[0218]

[0219] is the centrifugal force and Coriolis force matrix of the hybrid robot;

[0220] is the gravity vector of the hybrid robot;

[0221] is the generalized velocity of the tool tip of the rigid body at the end of the hybrid robot;

[0222] is the generalized acceleration of the tool tip of the rigid body at the end of the hybrid robot;

[0223] f 1～ f3 is related to the moving speed of the corresponding active arm's moving pair; It is related to the rotation speed of the corresponding rotating pair.

[0224] Among them, the friction model adopts the Coulomb-viscosity model, and the specific formula is as follows:

[0225]

[0226] Where, f c represents the Coulomb friction coefficient; f v represents the viscous friction coefficient; Represents the relative velocity of the joint. represents the friction function.

[0227] Step 2: Equivalent the inertial force and inertial moment of each component of the hybrid robot to the corresponding key points, linearize the dynamic model of the five-degree-of-freedom hybrid robot, separate the parameters to be identified, and obtain the following formula:

[0228] F = ΦP;

[0229] Where:

[0230] Φ is the system observation matrix of the hybrid robot;

[0231] P is the dynamic parameter to be identified of the hybrid robot;

[0232] Can be used These include 70 robot body inertia parameters and 10 joint friction parameters.

[0233] Step 3: Assume that the initial or final values ​​of position, velocity, and acceleration are all 0, and that the parameter constraints of workspace, position, velocity, and acceleration are satisfied during the motion process. Plan the identification trajectory of each joint of the hybrid robot according to a seventh-order polynomial.

[0234] Step 4: Run the hybrid robot, collect the position information of each joint of the hybrid robot and the driving current of each servo motor during the operation, and obtain ξ v ,ξ a ,Φ,F;

[0235] Step 5: Use the Sobol global sensitivity analysis method to perform parameter sensitivity analysis on the linearized dynamic model, and divide the inertial parameters to be identified into three categories according to their sensitivity: high sensitivity parameters P high , low sensitivity parameter P low and zero sensitivity parameter P zero ; For zero sensitivity parameter P zero No processing is done, for low sensitivity parameter P low The three-dimensional software is used to automatically derive the values, and for the high sensitivity parameter P high Genetic algorithm is used for identification.

[0236] Preferably, step 1 may include the following sub-steps:

[0237] In step 1-1, it can be assumed that: the intersection of the axis of the first active arm 6 and the axis of the first rotating shaft is B1; the intersection of the axis of the second active arm 7 and the axis of the third rotating shaft is B2; the intersection of the axis of the third active arm 8 and the axis of the fourth rotating shaft is B3; the intersection of the axis of the driven support arm 13 and the axis of the fifth rotating shaft is B4; the hinge point of the first active arm 6 and the moving platform 14 is A1; the hinge point of the second active arm 7 and the moving platform 14 is A2; the hinge point of the third active arm 8 and the moving platform 14 is A3; the intersection of the line connecting A2 and A3 and the axis of the C axis 15 of the double swing head is A4; the coordinate axes of the base coordinate system include the x0 axis, y0 axis, and z0 axis that are perpendicular to each other; B4 is the origin of the base coordinate system, B2 and B3 are located on the x0 axis; the displacement vector of A1 relative to B1 is q 3,1 ; The displacement vector of A2 relative to B2 is q 3,2 ; The displacement vector of A3 relative to B3 is q 3,3 ; The displacement vector of A4 relative to B4 is q 3,4 The C axis 15 of the double swing head rotates around the axis of the platform 14 at an angle of θ 4,4 The angle of rotation of the double swing head A axis 16 around the seventh axis is θ 5,4 The intersection of the seventh axis and the axis of the moving platform 14 is P; the end rigid body tip point of the hybrid robot is point G, and its position vector corresponding to the origin of the base coordinate system is r G The rotation angle of the tool axis around the base coordinate system x0 axis is α; the rotation angle around the base coordinate system y0 axis is β;

[0238] Perform inverse position solution on the hybrid robot, and use r G , α, β, and get q 3,1 ,q 3,2 ,q3,3 ,q 3,4 ,θ 4,4 ,θ 5,4 ;

[0239] Step 1-2, perform inverse velocity and acceleration solutions for the hybrid robot, and use ξ v Obtain: the extension and retraction speeds and corresponding accelerations of the first to third active arms 8, and the rotation speeds and corresponding accelerations of the C-axis and A-axis of the double swing head;

[0240] Step 1-3, the first active arm 6, the second active arm 7, the third active arm 8 and the driven support arm 13 are separated into several components; the first active arm 6, the second active arm 7, the third active arm 8 can be separated into P 1,i 、P 2,i (i=1~3) two components, i represents the first to third active arm sequence number, where component P 1,i Contains servo motor and sleeve, component P 2,i It includes a telescopic rod and a screw pair corresponding to the active arm; the driven support arm 13 can be decomposed into P 1,4 、P 2,4 Two components; the first rotating bracket 19, the second rotating bracket 12, the C swing head and the A swing head can be used as independent components, the first rotating bracket 19 is recorded as component R1; the second rotating bracket 12 is recorded as component R2; wherein component P 1,4 It can be the inner ring part of the driven support arm 13, component P 2,4 It may include a guide sleeve of a driven support arm 13 and a moving platform 14 .

[0241] The velocity and acceleration of the center of mass of each component are obtained from the extension and retraction speeds and corresponding accelerations of the first to third active arms, and the rotation speeds and corresponding accelerations of the C-axis and A-axis of the double swing head;

[0242] In steps 1-4, the inertial force and inertial moment of each component are calculated respectively, and then the dynamic model of the hybrid robot is established using the principle of virtual work.

[0243] Preferably, steps 1-2 may include the following sub-steps:

[0244] Step 1-2-1, by taking the derivative of the following formula, we can get the velocity v of point P: P ;

[0245] r G =r P +r PG ;

[0246] Where: r P is the position vector of point P; r PG is the position vector from point P to point G;

[0247] Step 1-2-2, take point P as the node, use the following closed-loop constraint equations to derive

[0248] r P =(q 3,4 +e)s 3,4 ,r P =b k +q 3,k s 3,k -a k +es 3,4 ,k=1~3;

[0249] Where: e is the distance from point A4 to point P; s 3,k , k = 1 to 3, corresponding to the unit vectors of the first, second, and third active arm 8 axes, namely: s 3,k For the corresponding q 3,k Unit displacement vector of b k The origin O of the base coordinate system points to point B k vector of a k Point A4 points to point A k vector;

[0250] Step 1-2-3, according to the principle of angular velocity superposition, we get the following formula, and further get and

[0251]

[0252] Where: w 5,4 is the angular velocity of the tool tip of the rigid body at the end of the hybrid robot; w 3,4 is the angular velocity of the moving platform 14; s 4,4 is the unit vector of the C-axis 15 rotation axis of the double swing head; s 5,4 is the unit vector of the rotation axis of the A-axis 16 of the double swing head.

[0253] Preferably, step 2 may include the following sub-steps:

[0254] Step 2-1, select the key nodes of each component; Figure 3 As shown, points A1~A4 and B1~B4 can be selected as key points, point O is the key point of the integrated hinge R, point P is the key point of the swing head of the series part C, and point C is the key point of the swing head of the series part A.

[0255] Step 2-2: Calculate the generalized velocity of the key nodes corresponding to each component, and equate the inertia force and inertia moment at the center of mass of each component to the corresponding key nodes, thereby separating the inertia parameters of each component;

[0256] Step 2-3: Based on F and the generalized velocity, inertia force, and inertia moment of key nodes, the linearized dynamic model of the hybrid robot is established using the principle of virtual work:

[0257] F=ΦP

[0258] Where: Φ is the system observation matrix of the hybrid robot; P is the dynamic parameter to be identified of the hybrid robot. The inertia parameters of the first active arm 6, the second active arm 7, and the third active arm 8 of the parallel mechanism can be considered the same. These include 70 robot body inertia parameters and 10 joint friction parameters.

[0259] Preferably, step 3 may include the following sub-steps:

[0260] Step 3-1, set the constraints for the identified trajectory as shown below:

[0261]

[0262] The specific settings can be as follows:

[0263]

[0264]

[0265] The specific settings can be as follows:

[0266]

[0267] Where:

[0268] t is the robot movement time;

[0269] q 3,k (t) is the displacement of the kth active arm moving pair at time t;

[0270] q 3,k_0 is the initial displacement of the kth active arm moving pair;

[0271] θ u,4 (t) is the rotation angle of the u-th active rotational pair at time t; when u=4, the u-th active rotational pair is the C-axis 15 rotational pair of the double swing head;

[0272] θ u,4_0 is the initial rotation angle of the u-th active rotation pair; when u=5, the u-th active rotation pair is the A-axis 16 rotation pair of the double swing head;

[0273] In step 3-2, the identification trajectory of each joint of the hybrid robot is planned according to the seventh-order polynomial model; wherein the seventh-order polynomial model is shown as follows:

[0274] x(t)=a0+a1t+a2t 2 +a3t 3 +a4t 4 +a5t 5 +a6t 6 +a7t 7 ;

[0275] x(t) is the displacement or rotation of the joint at time t;

[0276] a0~a7 are the coefficients of the terms from degree 0 to degree 7 of the seventh-order polynomial;

[0277] According to the constraints in step 3-1, the polynomial coefficients of each joint identification trajectory are calculated to obtain each joint identification trajectory. The obtained identification trajectory is as follows Figure 4 、 Figure 5 to shown.

[0278] Preferably, the first to third active arms 8 may each include a sleeve, a telescopic rod, and a screw pair driven by a servo motor; the telescopic rod is driven by the screw pair to extend and retract relative to the sleeve; the sleeve of the first active arm 6 is fixedly connected to the first rotating shaft; the sleeve of the second active arm 7 is fixedly connected to the third rotating shaft; and the sleeve of the third active arm 8 is fixedly connected to the fourth rotating shaft. Step 4 may include the following sub-steps:

[0279] Step 4-1: Write the obtained joint identification trajectory into G code, and then input it into the controller of the hybrid robot to drive the movement of each joint of the hybrid robot;

[0280] Step 4-2: While the hybrid robot is running along the identified trajectory, the position data of the hinge points between the hybrid robot's telescopic rod and the moving platform 14 and the corresponding servo motor current data are synchronously collected to generate position curves and current curves for each joint;

[0281] Step 4-3: Use a low-pass filter to filter the position curve of each joint, and then differentiate it in turn to obtain the velocity and acceleration of each joint;

[0282] Step 4-4, using the forward kinematics of the hybrid robot, calculate the position information of the end rigid body and the generalized velocity ξ of the end tool tip according to the position, velocity and acceleration of each joint v and generalized acceleration ξ a ;

[0283] Step 4-5: Based on the sampled servo motor currents for driving the C-axis 15 and the A-axis 16 of the double swing head, the corresponding driving torque τ for the C-axis and the A-axis is obtained according to the following formula: d :

[0284] τ d =r e ka i m ;

[0285] Where: k a is the torque constant of the servo motor; r e is the reduction ratio of the reducer connected to the servo motor; i m is the servo motor current;

[0286] Then, according to the sampled servo motor currents for driving the first to third active arms 8, the corresponding driving forces F for driving the first to third active arms are obtained according to the following formula: d :

[0287]

[0288] Where: b is the lead of the screw;

[0289] The calculated driving force and driving torque are filtered using a low-pass filter.

[0290] In steps 4-6, the position information of the end rigid body and the generalized velocity and generalized acceleration of the end tool tip are substituted into the expression of the measurement matrix Φ to obtain the result of the measurement matrix Φ.

[0291] Preferably, step 5 may include the following sub-steps:

[0292] Step 5-1: Map the inertia parameters to be identified to the unit interval according to their respective value ranges, and then decompose the linearized dynamic model into the following function form with respect to the inertia parameters to be identified:

[0293]

[0294] P=[x1 x2 ... x n ] T ;

[0295] Where:

[0296] f0 is a constant;

[0297] n is the number of parameters;

[0298] x i is the i-th parameter to be identified;

[0299] x j is the jth parameter to be identified;

[0300] f i (x i ) contains only the parameters to be identified x i Function subitem of ;

[0301] fi,j (x i ,x j ) contains the parameters to be identified x i and x j Function subitem of ;

[0302] f 1,2,,n (x1,x2,…,x n ) contains the parameters to be identified x1, x2,…, x n Function subitem of ;

[0303] All sub-items decomposed from the dynamic model are orthogonal to each other, so:

[0304]

[0305] I n =(P|0≤x i ≤1; i=1,2,…,n);

[0306] Where:

[0307] I n The n-dimensional unit body mapped out by the parameters to be identified;

[0308] s is the number of parameters to be identified contained in the function sub-term to be orthogonalized;

[0309] e1,e2…,e s Corresponding to the 1st, 2nd,…, sth parameters in the parameter set e to be identified;

[0310] h1, h2…, h s Corresponding to the 1st, 2nd,…, sth parameters in the parameter set h to be identified;

[0311] is a function sub-item containing the 1st, 2nd, …, sth parameters in the parameter set e to be identified;

[0312] is a function sub-item containing the 1st, 2nd, …, sth parameters in the parameter set h to be identified;

[0313] In step 5-2, multiple integrals are taken on the expression after the linearized dynamic model decomposition to obtain the following sub-terms:

[0314]

[0315] Where:

[0316] P -i Divide by x i Other parameters except

[0317] P -(ij)Divide by x i and x j Other parameters except

[0318] f j (x j ) contains the parameter x j Function subitem of ;

[0319] And so on, to obtain other sub-items;

[0320] Step 5-3, according to the sub-items solved in step 5-3, the total variance and each partial variance of the linearized dynamic model function are obtained:

[0321]

[0322]

[0323] Among them, the variance and total variance have the following characteristics:

[0324]

[0325] x s is the sth parameter;

[0326] D is the total variance of the linearized kinetic model function;

[0327] D i is the parameter x i The partial variance of

[0328] D ij is the parameter x i and x j The partial variance of

[0329] D 1,2,…,s For parameters x1, x2, …, x s The partial variance of

[0330] D 1,2,...,n For parameters x1, x2, …, x n The partial variance of

[0331] Step 5-4: Based on the total variance and each variance obtained in step 3, calculate the sensitivity of each order of all parameters according to the following formula:

[0332]

[0333] The parameter x is calculated by the following formula: i The total sensitivity coefficient is:

[0334]

[0335]

[0336] Where:

[0337] S Ti is the parameter x i The total sensitivity coefficient of

[0338] D -i Except for the parameter x i In addition, the sum of the partial variances of other parameters;

[0339] Step 5-5, perform two independent samplings on the parameter variables to obtain two matrices A N×n and B N×n , where N is the number of samples and n is the number of parameters; then use the Monte Carlo method to calculate according to the following formula and

[0340]

[0341] Where:

[0342] is the new sample matrix obtained by replacing the vth column in sample A with the vth column element in B;

[0343] is the new sample matrix obtained by replacing the vth column in sample B with the vth column element in A;

[0344] f(A) w is the dynamic model function value corresponding to sample A;

[0345] For samples The corresponding dynamic model function value;

[0346] For samples The corresponding dynamic model function value;

[0347] is the estimated value of f0 using the Monte Carlo method;

[0348] is the estimated value of the total variance of the kinetic model function using the Monte Carlo method;

[0349] To use Monte Carlo method to calculate the parameter x v Estimates of partial variance;

[0350] To use Monte Carlo method to calculate the parameter x v In addition, the estimated value of the sum of partial variances of other parameters;

[0351] Step 5-6: The sensitivity values ​​of the inertial parameters to be identified are divided into three categories; the first 30% parameters in the sensitivity value range are selected and defined as high sensitivity parameters P high , the last 20% parameters in the sensitivity value range are defined as zero sensitivity parameters P zero , the remaining parameters are defined as low sensitivity parameters P low ;

[0352] The sensitivity values ​​of the 70 inertial parameters to be identified can be divided into three categories. The first 20 parameters with the largest sensitivity values ​​are selected as high sensitivity parameters P high , the 14 parameters with a sensitivity value of 0 are zero sensitivity parameters P zero The remaining 36 parameters are low-sensitivity parameters P low .

[0353] Step 5-7, set the inertia parameter P to be identified high and friction parameters are the population of the genetic algorithm, and the objective function is as follows:

[0354]

[0355] Where: Q is the number of samples;

[0356] q is the sampling number;

[0357] τ qa is the qth actual sampling moment;

[0358] τ qt is the theoretical calculated torque corresponding to the qth actual sampling torque;

[0359] In steps 5-8, based on the fitness value, the population is continuously updated iteratively using crossover, inheritance, and mutation characteristics to obtain the population with the optimal fitness, which is the parameter identified by the genetic algorithm.

[0360] Preferably, the method may further include step 6, verifying and evaluating the dynamic model through experiments based on the identified parameters, and correcting the dynamic model of the hybrid robot.

[0361] Preferably, step 6 may include the following sub-steps:

[0362] Step 6-1: Calculate the driving force and driving torque of each joint using the identified dynamic parameters, draw a theoretical prediction curve, and compare it with the collected driving force and driving torque data;

[0363] In step 6-2, the obtained dynamic model is evaluated using the following root mean square error:

[0364]

[0365] Where:

[0366] Q is the number of samples;

[0367] q is the sampling number;

[0368] τ qa is the qth actual sampling moment;

[0369] Φ is the observation matrix;

[0370] are the identified kinetic parameters.

[0371] Preferably, These include 70 robot body inertia parameters and 10 joint friction parameters. The inertia parameters of the first, second, and third active arms 6, 7, and 8 of the parallel mechanism can be considered identical. Therefore, the total number of separated inertia parameters for the hybrid robot is 70, along with 10 joint friction parameters.

[0372] The embodiments described above are only used to illustrate the technical ideas and features of the present invention. Their purpose is to enable those skilled in the art to understand the contents of the present invention and implement them accordingly. The scope of the patent of the present invention cannot be limited by these embodiments alone. That is, any equivalent changes or modifications made to the spirit disclosed by the present invention still fall within the scope of the patent of the present invention.

Claims

1. A method for identifying dynamic parameters of a hybrid robot based on sensitivity analysis. The hybrid robot is a five-degree-of-freedom hybrid robot comprising a three-degree-of-freedom parallel mechanism and a series mechanism having two rotational degrees of freedom connected thereto. The three-degree-of-freedom parallel mechanism comprises a moving platform, a first active arm, a second active arm, a third active arm, a driven support arm, a first rotating bracket, and a second rotating bracket; wherein: The movable platform is hinged on the circumference of three active arms that are driven to extend and retract by a servo motor, namely the first active arm, the second active arm, and the third active arm; a driven support arm is fixed to the rear end of the movable platform; the serial mechanism includes an A / C-axis double swing head driven by a servo motor, wherein the C-axis of the double swing head is rotatably connected to the movable platform, and the A-axis of the double swing head is connected to the tool; the first active arm is rotatably connected to the first rotating bracket through the first rotating shaft; the first rotating bracket is rotatably connected to the fixed bearing seat through the second rotating shaft; the second active arm, the third active arm, and the driven support arm are rotatably connected to the second rotating bracket through the third rotating shaft, the fourth rotating shaft, and the fifth rotating shaft respectively; the second rotating bracket is rotatably connected to the fixed bearing seat through the sixth rotating shaft; the C-axis of the double swing head is rotatably connected to the A-axis through the seventh rotating shaft; the axis of the first rotating shaft is perpendicular to the axis of the second rotating shaft; the axes of the third rotating shaft, the fourth rotating shaft, and the fifth rotating shaft are parallel to each other and perpendicular to the axis of the sixth rotating shaft; the center of the second rotating bracket is located at the intersection of the axis of the fifth rotating shaft and the axis of the sixth rotating shaft; it is characterized in that it includes the following steps: Step 1: Perform position inverse solution, velocity inverse solution, and acceleration inverse solution for the hybrid robot in sequence. Then, the dynamic model of the hybrid robot is established using the principle of virtual work. The dynamic equation of the dynamic model is as follows: F=Mξ a +Cξ v +G+t f ; in: F=[F1 F2 F3 τ4 τ5] T ; Where: F1 is the driving force of the first active arm; F2 is the driving force of the second active arm; F3 is the driving force of the third active arm; τ4 is the driving torque of the C-axis; τ5 is the driving torque of axis A; f1 is the friction force of the moving pair of the first active arm; f2 is the friction force of the moving pair of the second active arm; f3 is the friction force of the moving pair of the third active arm; τ f4 is the friction torque of the rotating pair of the C axis; τ f5 is the friction torque of the revolute pair of axis A; is the inertia matrix of the hybrid robot; is the centrifugal force and Coriolis force matrix of the hybrid robot; is the gravity vector of the hybrid robot; is the generalized velocity of the tool tip of the rigid body at the end of the hybrid robot; is the generalized acceleration of the tool tip of the rigid body at the end of the hybrid robot; Step 2: Equivalent the inertial force and inertial moment of each component of the hybrid robot to the corresponding key points, linearize the dynamic model of the five-degree-of-freedom hybrid robot, separate the parameters to be identified, and obtain the following formula: F = ΦP; Where: Φ is the system observation matrix of the hybrid robot; P is the dynamic parameter to be identified of the hybrid robot; Step 3: Assume that the initial or final values ​​of position, velocity, and acceleration are all 0, and that the parameter constraints of workspace, position, velocity, and acceleration are satisfied during the motion process. Plan the identification trajectory of each joint of the hybrid robot according to a seventh-order polynomial. Step 4: Run the hybrid robot, collect the position information of each joint of the hybrid robot and the driving current of each servo motor during the operation, and obtain ξ v ,ξ a ,Φ,F; Step 5: Use the Sobol global sensitivity analysis method to perform parameter sensitivity analysis on the linearized dynamic model, and divide the inertial parameters to be identified into three categories according to their sensitivity: high sensitivity parameters P high , low sensitivity parameter P low and zero sensitivity parameter P zero ; For zero sensitivity parameter P zero No processing is done, for low sensitivity parameter P low The three-dimensional software is used to automatically derive the values, and for the high sensitivity parameter P high Genetic algorithm is used for identification.

2. The method for identifying dynamic parameters of a hybrid robot based on sensitivity analysis according to claim 1, characterized in that: Step 1 includes the following sub-steps: Step 1-1, assume that: the intersection of the first active arm axis and the first rotating shaft axis is B1; the intersection of the second active arm axis and the third rotating shaft axis is B2; the intersection of the third active arm axis and the fourth rotating shaft axis is B3; the intersection of the driven support arm axis and the fifth rotating shaft axis is B4; the hinge point of the first active arm and the moving platform is A1; the hinge point of the second active arm and the moving platform is A2; the hinge point of the third active arm and the moving platform is A3; the intersection of the line connecting A2 and A3 and the C-axis axis of the double swing head is A4; the base coordinate system coordinate axes include the x0 axis, y0 axis, and z0 axis that are perpendicular to each other; B4 is the origin of the base coordinate system, B2 and B3 are located on the x0 axis; the displacement vector of A1 relative to B1 is q 3,1 ; The displacement vector of A2 relative to B2 is q 3,2 ; The displacement vector of A3 relative to B3 is q 3,3 ; The displacement vector of A4 relative to B4 is q 3,4 The angle of rotation of the C axis of the double swing head around the axis of the platform is θ 4,4 The angle of rotation of the A axis of the double swing head around the seventh axis is θ 5,4 The intersection of the seventh axis and the axis of the moving platform is P; the tool tip of the hybrid robot end rigid body is point G, and its position vector corresponding to the origin of the base coordinate system is r G The rotation angle of the tool axis around the base coordinate system x0 axis is α; the rotation angle around the base coordinate system y0 axis is β; Perform inverse position solution on the hybrid robot, and use r G , α, β, and get q 3,1 ,q 3,2 ,q 3,3 ,q 3,4 ,θ 4,4 ,θ 5,4 ; Step 1-2, perform inverse velocity and acceleration solutions for the hybrid robot, and use ξ v Obtain: the extension and retraction speeds and corresponding accelerations of the first to third active arms, and the rotation speeds and corresponding accelerations of the C-axis and A-axis of the double swing head; Step 1-3: Disassemble the first to third active arms and the driven support arms into several components; obtain the velocity and acceleration of the center of mass of each component from the extension and retraction speeds and corresponding accelerations of the first to third active arms, and the rotation speeds and corresponding accelerations of the C-axis and A-axis of the double swing head; In steps 1-4, the inertial force and inertial moment of each component are calculated respectively, and then the dynamic model of the hybrid robot is established using the principle of virtual work.

3. The method for identifying dynamic parameters of a hybrid robot based on sensitivity analysis according to claim 2, characterized in that: Step 1-2 includes the following sub-steps: Step 1-2-1, by taking the derivative of the following formula, we can get the velocity v of point P: P ; r G ＝r P +r PG ； Where: r P is the position vector of point P; r PG is the position vector from point P to point G; Step 1-2-2, take point P as the node, use the following closed-loop constraint equations to derive r P =(q 3,4 +e)s 3,4 ,r P =b k +q 3,k s 3,k -to k +es 3,4 ,k=13; Where: e is the distance from point A4 to point P; s 3,k , k = 1 to 3, corresponding to the unit vectors of the first, second, and third active arm axes; b k The origin O of the base coordinate system points to point B k vector of a k Point A4 points to point A k vector; Step 1-2-3, according to the principle of angular velocity superposition, we get the following formula, and further get and Where: w 5,4 is the angular velocity of the tool tip of the rigid body at the end of the hybrid robot; w 3,4 is the angular velocity of the moving platform; s 4,4 is the unit vector of the C-axis rotation axis of the double swing head; s 5,4 is the unit vector of the A-axis rotation axis of the double swing head.

4. The method for identifying dynamic parameters of a hybrid robot based on sensitivity analysis according to claim 2, characterized in that: Step 2 includes the following sub-steps: Step 2-1, select the key nodes of each component; Step 2-2: Calculate the generalized velocity of the key nodes corresponding to each component, and equate the inertia force and inertia moment at the center of mass of each component to the corresponding key nodes, thereby separating the inertia parameters of each component; In step 2-3, the linearized dynamic model of the hybrid robot is established based on F and the generalized velocity, inertia force, and inertia moment of the key nodes using the principle of virtual work.

5. The method for identifying dynamic parameters of a hybrid robot based on sensitivity analysis according to claim 2, characterized in that: Step 3 includes the following sub-steps: Step 3-1, set the constraints for the identified trajectory as shown below: Where: t is the robot movement time; q 3,k (t) is the displacement of the kth active arm moving pair at time t; q 3,k_0 is the initial displacement of the kth active arm moving pair; θ u,4 (t) is the rotation angle of the uth active revolute pair at time t; When u=4, the uth active rotation pair is the C-axis rotation pair of the double swing head; θ u,4_0 is the initial rotation angle of the u-th active rotation pair; When u=5, the uth active rotation pair is the A-axis rotation pair of the double swing head; In step 3-2, the identification trajectory of each joint of the hybrid robot is planned according to the seventh-order polynomial model; wherein the seventh-order polynomial model is shown as follows: x(t)=a0+a1t+a2t 2 +a3t 3 +a4t 4 +a5t 5 +a6t 6 +a7t 7 ; x(t) is the displacement or rotation of the joint at time t; a0~a7 are the coefficients of the terms from degree 0 to degree 7 of the seventh-order polynomial; According to the constraints in step 3-1, the polynomial coefficients of each joint identification trajectory are calculated to obtain each joint identification trajectory.

6. The method for identifying dynamic parameters of a hybrid robot based on sensitivity analysis according to claim 1, characterized in that: The first to third active arms each include a sleeve, a telescopic rod, and a screw pair driven by a servo motor; the telescopic rod is driven by the screw pair to extend and retract relative to the sleeve; the sleeve of the first active arm is fixedly connected to the first rotating shaft; the sleeve of the second active arm is fixedly connected to the third rotating shaft; and the sleeve of the third active arm is fixedly connected to the fourth rotating shaft. Step 4 includes the following sub-steps: Step 4-1: Write the obtained joint identification trajectory into G code, and then input it into the controller of the hybrid robot to drive the movement of each joint of the hybrid robot; Step 4-2: While the hybrid robot is running along the identified trajectory, the position data of the hinge points between the hybrid robot's telescopic rod and the moving platform and the corresponding servo motor current data are synchronously collected to generate position curves and current curves for each joint; Step 4-3: Use a low-pass filter to filter the position curve of each joint, and then differentiate it in turn to obtain the velocity and acceleration of each joint; Step 4-4, using the forward kinematics of the hybrid robot, calculate the position information of the end rigid body and the generalized velocity ξ of the end tool tip according to the position, velocity and acceleration of each joint v and generalized acceleration ξ a ; Step 4-5: Based on the sampled servo motor currents for driving the C-axis and A-axis of the double swing head, the corresponding driving torque τ for driving the C-axis and A-axis is obtained according to the following formula: d : t d =r e k a I m ; Where: k a is the torque constant of the servo motor; r e is the reduction ratio of the reducer connected to the servo motor; i m is the servo motor current; Then, based on the sampled servo motor currents that drive the first to third active arms, the corresponding driving forces F that drive the first to third active arms are obtained according to the following formula: d : Where: b is the lead of the screw; The calculated driving force and driving torque are filtered using a low-pass filter; In steps 4-6, the position information of the end rigid body and the generalized velocity and generalized acceleration of the end tool tip are substituted into the expression of the measurement matrix Φ to obtain the result of the measurement matrix Φ.

7. The method for identifying dynamic parameters of a hybrid robot based on sensitivity analysis according to claim 1, characterized in that: Step 5 includes the following sub-steps: Step 5-1: Map the inertia parameters to be identified to the unit interval according to their respective value ranges, and then decompose the linearized dynamic model into the following function form with respect to the inertia parameters to be identified: P=[x1 x2…x n ] T ; Where: f0 is a constant; n is the number of parameters; x i is the i-th parameter to be identified; x j is the jth parameter to be identified; f i (x i ) contains only the parameters to be identified x i Function subitem of ; f i,j (x i ,x j ) contains the parameters to be identified x i and x j Function subitem of ; f 1,2,…,n (x1,x2,…,x n ) contains the parameters to be identified x1, x2,…, x n Function subitem of ; All sub-items decomposed from the dynamic model are orthogonal to each other, so: I n =(P|0≤x i ≤1;i=1,2,…,n); Where: I n The n-dimensional unit body mapped out by the parameters to be identified; s is the number of parameters to be identified contained in the function sub-term to be orthogonalized; e1,e2…,e s Corresponding to the 1st, 2nd,…, sth parameters in the parameter set e to be identified; h1, h2…, h s Corresponding to the 1st, 2nd,…, sth parameters in the parameter set h to be identified; is a function sub-item containing the 1st, 2nd, …, sth parameters in the parameter set e to be identified; is a function sub-item containing the 1st, 2nd, …, sth parameters in the parameter set h to be identified; In step 5-2, multiple integrals are taken on the expression after the linearized dynamic model decomposition to obtain the following sub-terms: Where: P -i Divide by x i Other parameters except P -(ij) Divide by x i and x j Other parameters except f j (x j ) contains the parameter x j Function subitem of ; And so on, to obtain other sub-items; Step 5-3, according to the sub-items solved in step 5-3, the total variance and each partial variance of the linearized dynamic model function are obtained: Among them, the variance and total variance have the following characteristics: x s is the sth parameter; D is the total variance of the linearized kinetic model function; D i is the parameter x i The partial variance of D ij is the parameter x i and x j The partial variance of D 1,2,…,s For parameters x1, x2, …, x s The partial variance of D 1,2,…,n For parameters x1, x2, …, x n The partial variance of Step 5-4: Based on the total variance and each variance obtained in step 3, calculate the sensitivity of each order of all parameters according to the following formula: The parameter x is calculated by the following formula: i The total sensitivity coefficient is: Where: S Ti is the parameter x i The total sensitivity coefficient of D -i Except for the parameter x i In addition, the sum of the partial variances of other parameters; Step 5-5, perform two independent samplings on the parameter variables to obtain two matrices A N×n and B N×n , where N is the number of samples and n is the number of parameters; then use the Monte Carlo method to calculate according to the following formula and Where: is the new sample matrix obtained by replacing the vth column in sample A with the vth column element in B; is the new sample matrix obtained by replacing the vth column in sample B with the vth column element in A; f(A) w is the dynamic model function value corresponding to sample A; For samples The corresponding dynamic model function value; For samples The corresponding dynamic model function value; is the estimated value of f0 using the Monte Carlo method; is the estimated value of the total variance of the kinetic model function using the Monte Carlo method; To use Monte Carlo method to calculate the parameter x v Estimates of partial variance; To use Monte Carlo method to calculate the parameter x v In addition, the estimated value of the sum of partial variances of other parameters; Step 5-6: The sensitivity values ​​of the inertial parameters to be identified are divided into three categories; the first 30% parameters in the sensitivity value range are selected and defined as high sensitivity parameters P high , the last 20% parameters in the sensitivity value range are defined as zero sensitivity parameters P zero , the remaining parameters are defined as low sensitivity parameters P low ; Step 5-7, set the inertia parameter P to be identified high and friction parameters are the population of the genetic algorithm, and the objective function is as follows: Where: Q is the number of samples; q is the sampling number; τ qa is the qth actual sampling moment; τ qt is the theoretical calculated torque corresponding to the qth actual sampling torque; In steps 5-8, based on the fitness value, the population is continuously updated iteratively using crossover, inheritance, and mutation characteristics to obtain the population with the optimal fitness, which is the parameter identified by the genetic algorithm.

8. The method for identifying dynamic parameters of a hybrid robot based on sensitivity analysis according to claim 1, characterized in that: The method further includes step 6, verifying and evaluating the dynamic model through experiments based on the identified parameters, and correcting the dynamic model of the hybrid robot.

9. The method for identifying dynamic parameters of a parallel-parallel robot based on sensitivity analysis according to claim 8, characterized in that: Step 6 includes the following sub-steps: Step 6-1: Calculate the driving force and driving torque of each joint using the identified dynamic parameters, draw a theoretical prediction curve, and compare it with the collected driving force and driving torque data; In step 6-2, the obtained dynamic model is evaluated using the following root mean square error: Where: Q is the number of samples; q is the sampling number; τ qa is the qth actual sampling moment; Φ is the observation matrix; are the identified kinetic parameters.

10. The method for identifying dynamic parameters of a hybrid robot based on sensitivity analysis according to claim 1, characterized in that: make These include 70 robot body inertia parameters and 10 joint friction parameters.

Images