Long-arm-span heavy-load robot and force-position hybrid control and high-precision dynamic compensation method

Abstract

The invention discloses a long-arm-span heavy-load robot and a force-position hybrid control and high-precision dynamic compensation method, and belongs to the technical field of forging robots, the long-arm-span heavy-load robot comprises a 4R type series main chain and five parallel closed-loop structures, and an end effector, a static platform and a connecting rod part can achieve the seven-axis movement capacity under motor and hydraulic drive. According to the force-position hybrid control and high-precision dynamic compensation method, the position and force feedback signals of the robot are collected in real time, and the self-adaptive control algorithm is combined, so that high-precision forging control is achieved; and meanwhile, the deformation of the connecting rod is calculated by obtaining the set coordinate value of the end effector and the stress condition of the connecting rod, then the compensation deformation of the driving oil cylinder is controlled, a closed-loop feedback mechanism is utilized, mechanical deformation online compensation and multi-connecting-rod deformation compensation are achieved, and high-precision control over the robot is achieved. The device has the advantages of being large in load, high in rigidity, long in arm span and high in precision, and medium and large forgings in a forging workshop can be rapidly and accurately transferred.

Description

Technical Field

[0001] The present invention belongs to the technical field of forging robots, and in particular relates to a long-arm-span heavy-load robot and a force-position hybrid control and high-precision dynamic compensation method. Background Art

[0002] In modern manufacturing, the forging process is a key step in producing high-strength, high-performance metal parts, and is crucial to improving the production efficiency and product quality of the entire industry. With the continuous development of forging robot technology, its application in the forging field has gradually attracted attention, providing the possibility of realizing the automation and intelligentization of the forging process, helping to reduce labor intensity, improve production efficiency, and ensure the consistency of product quality. This puts higher demands on forging robots, so the design and development of a high-performance forging robot is an inevitable trend in the development of key industries.

[0003] Currently, existing forging robots have shortcomings primarily in terms of travel, load, heat resistance, and precision. For example, their travel is generally short, unable to meet the forging range requirements for large, complex parts. Their load capacity is also limited, making them prone to problems such as insufficient torque and reduced precision when forging large, heavy-loaded parts. In terms of heat resistance, the high temperatures generated during the forging process are difficult for ordinary robot materials and structures to withstand, which can easily lead to equipment damage or performance degradation. In terms of precision, affected by factors such as structural design and transmission system errors, it is difficult to achieve high-precision forging requirements, affecting the dimensional accuracy and surface quality of parts. These limitations prevent existing forging robots from meeting the modern manufacturing industry's demand for high-quality, high-efficiency production of large, complex metal parts, limiting their widespread application in the forging industry.

[0004] In order to solve the shortcomings of the above-mentioned forging robots and better meet the high-quality and high-efficiency production needs of modern manufacturing for the forging of large and complex metal parts, it is urgent to invent a seven-degree-of-freedom, long-arm-span, heavy-load robot and a high-precision dynamic compensation control method, which is of great significance to improving the automation level, production efficiency and product quality of the forging industry. Summary of the Invention

[0005] The problem to be solved by the present invention is to provide a long-arm-span heavy-load robot with seven degrees of freedom and a force-position hybrid control and high-precision dynamic compensation method.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: a long-arm-span heavy-load robot, including a 4R-type serial main chain and five parallel closed-loop structures, the five parallel closed-loop structures including parallel closed-loop one, parallel closed-loop two, parallel closed-loop three, parallel closed-loop four and parallel closed-loop five, the parallel closed-loop one includes a static platform, link six, link nine and link ten; the parallel closed-loop two includes link four, link ten, link twelve and link thirteen; the parallel closed-loop three includes link two, link three, link four and link six; the parallel closed-loop four includes link one, link four, link five and link eight; the parallel closed-loop five includes an end effector, link one, link fourteen and link fifteen, the parallel closed-loop one is connected to the parallel closed-loop four by link seven, the parallel closed-loop four and the parallel closed-loop five have a common link one; the parallel closed-loop two and the parallel closed-loop three have a common link four.

[0007] The 4R-type series main chain realizes the functions of coupling degrees of freedom, changing the joint motion state and amplifying the joint stroke by adding five parallel closed loop structures: parallel closed loop 1, parallel closed loop 2, parallel closed loop 3, parallel closed loop 4 and parallel closed loop 5.

[0008] The present invention also includes a lifting hydraulic cylinder, a telescopic hydraulic cylinder, a posture hydraulic cylinder, and an auxiliary oil cylinder. The lifting hydraulic cylinder pushes the connecting rod twelve to lift and lower, and is amplified by the parallel closed loop two and the parallel closed loop three, thereby controlling the lifting movement of the end effector to realize the lifting movement of the end effector in the plane; the telescopic hydraulic cylinder drives the parallel closed loop one, the parallel closed loop four, and the parallel closed loop five to extend and retract together by pushing the connecting rod eleven, thereby controlling the telescopic movement of the end effector to realize the telescopic movement of the end effector in the plane; the posture hydraulic cylinder drives the parallel closed loop five to control the posture of the end effector, decouples the position and posture, and realizes the pitch movement of the end effector in the plane; the auxiliary oil cylinder assists the telescopic hydraulic cylinder in driving.

[0009] The connecting rod one, connecting rod two, connecting rod three, connecting rod four, connecting rod five, connecting rod six, connecting rod seven, connecting rod eight, connecting rod nine, connecting rod ten, connecting rod eleven, connecting rod twelve, connecting rod thirteen, connecting rod fourteen, and connecting rod fifteen constitute a connecting rod mechanism. The static platform, connecting rod mechanism, and end effector are installed on the mobile body as the main body of the robot mechanism. The mobile body is located on the guide rail. A rotary motor is installed at the center of the mobile body, which can control the overall translation and rotation of the robot to realize the two-axis motion of overall translation and overall rotation.

[0010] The connecting rods of the connecting rod mechanism are connected through parallel rotation pairs and are jointly driven by the lifting hydraulic cylinder, the telescopic hydraulic cylinder and the posture hydraulic cylinder to control the planar movement of the robot mechanism body.

[0011] Furthermore, the end effector is an end clamp, which is controlled by a motor to achieve two-axis movement of clamp head clamping and clamp head rotation.

[0012] The present invention also provides a force-position hybrid control method for a long-reach, heavy-load robot. This method achieves high-precision control by real-time acquisition of the robot's position signal and force feedback signal and combining it with an adaptive control algorithm. The method includes the following steps: S1. Position signal acquisition: The position signals of each joint of the robot are collected through encoders or photoelectric sensors, and the actual position of the end effector is calculated through forward kinematics; Furthermore, each joint of the robot is equipped with an encoder for collecting position signals; the end effector is equipped with a force sensor for collecting force feedback signals; Assume that the robot has n joints and the position of the i-th joint is , then the joint position vector is: Position of the end effector x Calculated by the forward kinematics model: in, is the forward kinematics function of the robot, established by the Denavit-Hartenberg (DH) parametric method or the geometric method.

[0013] S2. Force feedback signal acquisition: The force feedback signal of the robot end effector is collected through a force sensor or a torque sensor, and the deviation between the actual force and the target force is calculated; Assume that the actual force of the end effector is , the target force is , then the force deviation The calculation formula is: S3, control command calculation: according to the preset forging trajectory and force target, combined with the position signal and force feedback signal, the control command is calculated; Furthermore, let the target position of the end effector be , the actual location is , calculate the position deviation , Position control instructions This can be calculated using a proportional-derivative (PD) controller: in, Kp is the position proportional gain, Kd is the differential gain, t For time; Force control instructions This can be calculated using a proportional-integral (PI) controller: in, Kfp is the force proportional gain, Kfi is the integral gain; Finally, the mixed control instruction u is the position control instruction and force control instructions The weighted sum of: in, α is the weight coefficient, which is used to adjust the relative importance of position control and force control.

[0014] S4, Adaptive control algorithm: Adaptive control algorithm is used to adjust the robot's motion state to achieve precise control of position and force; Furthermore, the adaptive control algorithm uses the model reference adaptive control (MRAC) method to adjust the control parameters in real time so that the robot's motion state approaches the reference model. The specific steps are as follows: Establish the dynamic model of the robot, assuming that the dynamic model of the robot is: in, is the mass matrix, are the Coriolis and centrifugal force matrices, G(q) is the gravity vector, is the first-order derivative of the joint position vector, is the second-order derivative of the joint position vector.

[0015] Design a reference model and determine the desired motion state. Let the reference model be: Where r is the reference input, K 1 and K 2 are the two gain matrices of the reference model, is the joint position vector of the reference model, is the first-order derivative of the joint position vector of the reference model, is the second-order derivative of the joint position vector of the reference model; The adaptive control law adjusts the control parameters to make the actual motion state close to the reference model. θ To minimize tracking error e : Where Γ is the adaptive gain matrix, is the regression matrix, is the first-order derivative of the control parameter.

[0016] The present invention also provides a high-precision dynamic compensation method for a long-arm span heavy-load robot, comprising the following steps: analyzing the deformation of the robot's mechanical arm by the finite element method, fitting the deformation function by the response surface method, and proposing an iterative compensation strategy for the deformation to achieve online compensation of mechanical deformation.

[0017] Furthermore, fitting the deformation function requires finite element-based experimental design and the selection of a high-precision response surface model. Based on the end-effector coordinates, a central composite design (CCD) experimental method was used to generate multiple sets of experimental design points required for fitting the response surface model. A joint simulation platform combining SolidWorks and ANSYS Workbench software was established to simulate each set of experimental points. The parameters of each set of experimental points were updated through the joint software to correspond to each updated experimental model. Each updated model was then subjected to a finite element simulation until all experimental points were simulated, resulting in the final experimental design data.

[0018] The response surface functions used to fit the model include the following: in, is a first-order response surface model, is a second-order response surface model, is a third-order response surface model, is a fourth-order response surface model, is the function constant term, is the coefficient of the first-order term of the model function, is the first coefficient of the quadratic term of the model function, is the second coefficient of the quadratic term of the model function, is the cubic coefficient of the model function, is the coefficient of the quartic term of the model function, is the end effector coordinate, is the end effector coordinate exist i The specific value of The coordinates of the end effector are j is the specific value of , and s is the total number of experimental points.

[0019] After completing the response surface model fitting, the fitting accuracy of the response surface model needs to be evaluated. By selecting a small number of experimental points, the calculated values ​​of the response surface model are compared with the actual values ​​of the experimental points, and the relative average absolute error (RAAE), relative maximum absolute error (RMAE), root mean square error (RMSE), variance (R 2) These four accuracy evaluation indicators analyze the fitting accuracy of the response surface model. The formulas of these four evaluation indicators are as follows: Among them, RAAE is the relative average absolute error, RMAE is the relative maximum absolute error, RMSE is the root mean square error, R 2 is the variance, is the number of experimental points selected, For the i The true value of the experimental point, For the i The predicted value of each experimental point calculated by the response surface model is for The average value of .

[0020] From the above formula, we can see that R 2 The larger the value, the smaller the relative mean absolute error, relative maximum absolute error, and root mean square error, indicating that the response surface fitting accuracy is higher. Using the above theoretical method and the data obtained from the experimental design, the response surface model was established with the help of Isight software.

[0021] An iterative compensation strategy is proposed for deformation to realize online compensation of mechanical deformation. By obtaining the set coordinate value of the end effector and the force condition of the connecting rod mechanism, the deformation of the connecting rod is calculated, and then the driving auxiliary cylinder is controlled to compensate for the deformation. The fuzzy controller is combined with the PID controller and the closed-loop feedback mechanism to realize online compensation of mechanical deformation and multi-link deformation compensation, thereby achieving high-precision control of the robot.

[0022] The end clamp, static platform and connecting rod mechanism of the present invention serve as the main body of the long-arm-span heavy-duty robot, including a 4R-type serial main chain L1 and five closed-loop coupling structures; the connecting rod mechanism includes fifteen connecting rods, which are connected by parallel revolving pairs to realize the three-axis movement of the clamp head in the plane: lifting and lowering, extending and retracting, and pitching; the end clamp realizes the two-axis movement of the clamp head clamping and rotating; the mobile body controls the translation and rotation of the entire mechanism of the long-arm-span heavy-duty robot, realizing the two-axis movement of overall translation and overall rotation. Driven by motors and hydraulics, the forging robot can achieve seven-axis movement. At the same time, by real-time acquisition of the robot's position and force feedback signals, combined with an adaptive control algorithm, high-precision forging control is achieved. The seven-degree-of-freedom long-arm-span heavy-duty forging robot of the present invention has the advantages of large load, high rigidity, long arm span and high precision, and can realize the rapid and accurate transfer of medium and large forgings in the forging workshop.

[0023] The specific effects of the present invention are as follows: The main body of the robot mechanism of the present invention is composed of a 4R-type series main chain with five parallel closed-loop structures added. The parallel closed-loop structure realizes the coupling of degrees of freedom and changes the motion state of the joints, while amplifying the joint stroke and improving the rigidity and load capacity of the mechanism. It has the advantages of large load, high rigidity, long arm span and high precision, and can realize the rapid and precise transfer of medium and large forgings in the forging workshop.

[0024] The mobile body, static platform, connecting rod mechanism and end clamp of the present invention are driven by motors and hydraulics to realize seven-axis motion of clamp head lifting, clamp head extension and retraction, clamp head pitching, clamp head clamping, clamp head rotation, overall translation and overall rotation, thereby improving the operational flexibility and adaptability of the robot, enabling the robot to complete complex forging tasks in a variety of postures and motion modes, and flexibly adjusting the position and posture of the clamp head through multi-axis coordinated motion to ensure accurate grasping and operation of forgings, expand the working range and efficiency, and improve process accuracy and quality.

[0025] This invention uses a hybrid force and position control method, which achieves high-precision control by collecting position and force feedback signals in real time and combining it with an adaptive control algorithm. This hybrid control method precisely adjusts the robot's motion according to the preset forging trajectory and force target, ensuring that the force applied during the forging process consistently meets process requirements, thereby improving forging accuracy and quality. The adaptive control algorithm adjusts control parameters in real time to accommodate dynamic load changes and uncertain working environments, enhancing the adaptability and stability of the equipment and improving production efficiency and safety.

[0026] The present invention realizes online compensation of mechanical deformation through high-precision dynamic compensation technology, can monitor and compensate the deformation of the robotic arm in real time, ensure the accuracy and stability of the robot under long-term, high-intensity working conditions, improve the operating accuracy of the robot, and reduce stress concentration and fatigue damage caused by mechanical deformation by compensating for deformation in real time, thereby extending the service life of the equipment, reducing the maintenance cost and replacement frequency of the equipment, and improving production efficiency and economy. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The present invention will be described in detail below with reference to the accompanying drawings and in combination with examples, and the advantages and implementation modes of the present invention will become more apparent. The contents shown in the accompanying drawings are only used to illustrate the present invention and do not constitute any limitation to the present invention. In the accompanying drawings: Figure 1 It is a structural schematic diagram of the long arm span heavy-load robot of the present invention.

[0028] Figure 2 It is a front view of the long arm span heavy-load robot of the present invention.

[0029] Figure 3It is a schematic diagram of the motion state of the long arm span heavy-load robot of the present invention at a certain moment.

[0030] Figure 4 This is a schematic diagram of the high-precision dynamic compensation control strategy for the long-arm span and heavy-load robot of the present invention.

[0031] In the picture: 1. Mobile body; 2. Static platform; 3. Connecting rod mechanism; 4. End clamp; 31. Parallel closed loop 1; 32. Parallel closed loop 2; 33. Parallel closed loop 3; 34. Parallel closed loop 4; 35. Parallel closed loop 5; 301. Connecting rod 1; 302. Connecting rod 2; 303. Connecting rod 3; 304. Connecting rod 4; 305. Connecting rod 5; 306. Connecting rod 6; 307. Connecting rod 7; 308. Connecting rod 8; 309. Connecting rod 9; 3010. Connecting rod 10; 3011. Connecting rod 11; 3012. Connecting rod 12; 3013. Connecting rod 13; 3014. Connecting rod 14; 3015. Connecting rod 15; L1, 4R type series main chain; P1, lifting hydraulic cylinder; P2, telescopic hydraulic cylinder; P3, attitude hydraulic cylinder; P4, auxiliary cylinder. DETAILED DESCRIPTION

[0032] like Figures 1 to 3 As shown, a long arm span heavy-load robot includes a 4R type serial main chain L1 and five parallel closed loop structures, the five parallel closed loop structures include parallel closed loop 1 31, parallel closed loop 2 32, parallel closed loop 33, parallel closed loop 4 34 and parallel closed loop 5 35, parallel closed loop 1 31 includes static platform 2, link 6 306, link 9 309, link 10 3010; parallel closed loop 2 32 includes link 4 304, link 10 3010, link 12 3012, link 13 3013; parallel closed loop 3 33 includes link 2 302, link 3 30 3. Link four 304, link six 306; parallel closed loop four 34 includes link one 301, link four 304, link five 305, and link eight 308; parallel closed loop five 35 includes an end effector (in this embodiment, the end clamp 4), link one 301, link fourteen 3014, and link fifteen 3015. Parallel closed loop one 31 and parallel closed loop four 34 are connected by link seven 307. Parallel closed loop four 34 and parallel closed loop five 35 share a common link one 301; parallel closed loop two 32 and parallel closed loop three 33 share a common link four 304.

[0033] The 4R-type series main chain L1 realizes the functions of coupling degrees of freedom, changing the joint motion state and amplifying the joint stroke by adding five parallel closed loop structures, namely parallel closed loop one 31, parallel closed loop two 32, parallel closed loop three 33, parallel closed loop four 34 and parallel closed loop five 35.

[0034] The present invention also includes a lifting hydraulic cylinder P1, a telescopic hydraulic cylinder P2, a posture hydraulic cylinder P3, and an auxiliary oil cylinder P4. The lifting hydraulic cylinder P1 drives the connecting rod 12 3012 to rise and fall, and is amplified by the parallel closed loop 2 32 and the parallel closed loop 33, thereby controlling the lifting movement of the end clamp 4 to achieve the lifting movement of the clamp head in the plane; the telescopic hydraulic cylinder P2 drives the parallel closed loop 1 31, the parallel closed loop 4 34, and the parallel closed loop 5 35 to extend and retract together by pushing the connecting rod 11 3011, thereby controlling the telescopic movement of the end clamp 4 to achieve the telescopic movement of the clamp head in the plane (that is, the telescopic hydraulic cylinder P2 drives the connecting rod 13 3013 by pushing the connecting rod 11 3011, and transmits to the parallel closed loop 2 32). Parallel closed loop three 33, and then drives parallel closed loop one 31 through connecting rod six 306, and finally drives parallel closed loop four 34 and parallel closed loop five 35 to coordinate telescopic movement through connecting rod seven 307 and connecting rod one 301); the posture hydraulic cylinder P3 drives parallel closed loop five 35 to control the posture of the end clamp 4, decoupling the position and posture, and realizing the pitch movement of the end clamp 4 in the plane (that is, the posture hydraulic cylinder P3 simultaneously drives connecting rod two 302 (belonging to parallel closed loop three 33) and connecting rod fifteen 3015 (belonging to parallel closed loop five 35), so as to realize precise adjustment of the posture of the end effector in a manner that the former assists positioning and the latter takes the lead in control); the auxiliary oil cylinder P4 assists the telescopic hydraulic cylinder P2 in driving.

[0035] Link 1 301, link 2 302, link 3 303, link 4 304, link 5 305, link 6 306, link 7 307, link 8 308, link 9 309, link 10 3010, link 11 3011, link 12 3012, link 13 3013, link 14 3014, and link 15 3015 constitute a link mechanism 3. The static platform 2, the link mechanism 3, and the end clamp 4 are installed on the mobile body 1 as the main body of the robot mechanism. The mobile body 1 is located on the guide rail. A rotary motor is installed at the center of the mobile body 1, which can control the overall translation and rotation of the robot, thereby realizing two-axis motion of overall translation and overall rotation.

[0036] The connecting rods of the connecting rod mechanism 3 are connected through parallel rotation pairs and are jointly driven by the lifting hydraulic cylinder P1, the telescopic hydraulic cylinder P2, and the posture hydraulic cylinder P3 to control the planar motion of the robot mechanism body.

[0037] The end clamp 4 is controlled by a motor to achieve two-axis movement of clamp head clamping and clamp head rotation.

[0038] The present invention also provides a force position control method for a long-arm-span heavy-load robot, which realizes high-precision control by real-time acquisition of the robot's position signal and force feedback signal and combining it with an adaptive control algorithm, and includes the following steps: S1. Position signal acquisition: The position signals of each joint of the robot are collected through encoders or photoelectric sensors, and the actual position of the end clamp 4 is calculated through forward kinematics; Specifically, each joint of the robot is equipped with an encoder for collecting position signals; the end clamp 4 is equipped with a force sensor for collecting force feedback signals; Assume that the robot has n joints and the position of the i-th joint is , then the joint position vector is: Position of end clamp 4 x Calculated by the forward kinematics model: in, is the forward kinematics function of the robot, established by the Denavit-Hartenberg (DH) parametric method or the geometric method.

[0039] S2. Force feedback signal acquisition: The force feedback signal of the robot end clamp 4 is collected through a force sensor or a torque sensor, and the deviation between the actual force and the target force is calculated; The actual force of the end clamp 4 is , the target force is , then the force deviation The calculation formula is: S3, control command calculation: according to the preset forging trajectory and force target, combined with the position signal and force feedback signal, the control command is calculated; Assume that the target position of the end clamp 4 is , the actual location is , calculate the position deviation , Position control instructions This can be calculated using a proportional-derivative (PD) controller: in, Kp is the position proportional gain, Kd is the differential gain, t For time; Force control instructions This can be calculated using a proportional-integral (PI) controller: in, Kfp is the force proportional gain, Kfi is the integral gain; Finally, the mixed control instruction u is the position control instruction and force control instructions The weighted sum of: in, α is the weight coefficient, which is used to adjust the relative importance of position control and force control.

[0040] S4, Adaptive control algorithm: Adaptive control algorithm is used to adjust the robot's motion state to achieve precise control of position and force; The adaptive control algorithm uses the Model Reference Adaptive Control (MRAC) method to adjust the control parameters in real time so that the robot's motion state approaches the reference model. The specific steps are as follows: Establish the dynamic model of the robot, assuming that the dynamic model of the robot is: in, is the mass matrix, are the Coriolis and centrifugal force matrices, G(q) is the gravity vector, is the first-order derivative of the joint position vector, is the second-order derivative of the joint position vector.

[0041] Design a reference model and determine the desired motion state. Let the reference model be: Where r is the reference input, K 1 and K 2 are the two gain matrices of the reference model, is the joint position vector of the reference model, is the first-order derivative of the joint position vector of the reference model, is the second-order derivative of the joint position vector of the reference model; The adaptive control law adjusts the control parameters to make the actual motion state close to the reference model. θ To minimize tracking error e : Where Γ is the adaptive gain matrix, is the regression matrix, is the first-order derivative of the control parameter.

[0042] The present invention also provides a high-precision dynamic compensation method for a long-arm-span heavy-load robot, comprising the following steps: The deformation of the robotic arm is analyzed using the finite element method, and the deformation function is fitted using the response surface methodology. An iterative compensation strategy is proposed for this deformation, achieving online compensation of the mechanical deformation. In particular, fitting the response surface deformation function requires finite element-based experimental design and the selection of a high-precision response surface model.

[0043] Fitting a response surface model requires a large amount of sample data generated through experimental design. This experiment employed a central composite design (CCD) to generate the experimental points necessary for fitting the response surface model. This method expands the design space and obtains high-order information, providing sample data for approximate models. The experimental design is simple, allowing for a highly accurate fitting model to be obtained with fewer experiments.

[0044] Based on the four-coordinate system of the end-mount clamp, a central composite design (CCD) experimental method was used to generate multiple sets of experimental design points. A co-simulation platform was established using SolidWorks and ANSYS Workbench software to simulate each set of experimental points. The parameters of each set of experimental points were updated using the co-simulation software, corresponding to each updated experimental model. Each updated model was then subjected to a finite element simulation until all experimental points were simulated, resulting in the final experimental design data.

[0045] The response surface functions used to fit the model include the following: in, is a first-order response surface model, is a second-order response surface model, is a third-order response surface model, is a fourth-order response surface model, is the function constant term, is the coefficient of the first-order term of the model function, is the first coefficient of the quadratic term of the model function, is the second coefficient of the quadratic term of the model function, is the cubic coefficient of the model function, is the coefficient of the quartic term of the model function, The 4 coordinates of the end clamp, 4 coordinates for the end clamp exist i The specific value of The coordinates of the end clamp 4 are j is the specific value of , and s is the total number of experimental points.

[0046] After completing the response surface model fitting, the fitting accuracy of the response surface model needs to be evaluated. By selecting a small number of experimental points, the calculated values ​​of the response surface model are compared with the actual values ​​of the experimental points, and the relative average absolute error (RAAE), relative maximum absolute error (RMAE), root mean square error (RMSE), variance (R 2 ) These four accuracy evaluation indicators analyze the fitting accuracy of the response surface model. The formulas of these four evaluation indicators are as follows: Among them, RAAE is the relative average absolute error, RMAE is the relative maximum absolute error, RMSE is the root mean square error, R 2 is the variance, is the number of experimental points selected, For the i The true value of the experimental point, For the i The predicted value of each experimental point calculated by the response surface model is for The average value of .

[0047] From the above formula, we can see that R 2 The larger the value, the smaller the relative mean absolute error, relative maximum absolute error, and root mean square error, indicating that the response surface fitting accuracy is higher. Using the above theoretical method and the data obtained from the experimental design, the response surface model was established with the help of Isight software.

[0048] An iterative compensation strategy is proposed for deformation to achieve online compensation of mechanical deformation. Figure 4 As shown in the figure, by obtaining the set coordinate value of the end clamp 4 and the force condition of the connecting rod mechanism, the deformation of the connecting rod is calculated, and then the driving auxiliary cylinder P4 is controlled to compensate for the deformation. The fuzzy controller is combined with the PID controller and the closed-loop feedback mechanism to realize online compensation of mechanical deformation and multi-link deformation compensation, thereby achieving high-precision control of the robot.

[0049] The actual execution end control accuracy is ≤±2mm.

[0050] Specifically, the fitting deformation function in high-precision dynamic compensation control and the accuracy of the response surface model are shown in Table 1: Table 1 Accuracy of response surface model

[0051] Based on the model fitting results, the parameter model fitting accuracy evaluation index requirements are set as follows: RAAE < 0.2, RMAE < 0.3, RMSE < 0.2, R2 < 0.9. From the data in the table, it can be seen that the first-order fitting function of deformation and coordinate value has the highest accuracy. Therefore, the deformation function is expressed as follows: Assume that the coordinate value at this time for , the fitting value is 18.9mm, indicating that the end clamp 4 is located at this coordinate and the robot end deformation is 18.9mm. The deformation K serves as the input value of the fuzzy PID closed-loop control system. The system then compares the input value with the signal fed back by the measuring transmitter to obtain the deviation E and simultaneously calculates the rate of change of the deviation dE / dt. The fuzzy controller adjusts the parameters of the PID controller based on E and dE / dt. The PID controller calculates the control signal based on the adjusted parameters and the deviation E, and acts on the controlled object, causing the output of the controlled object to change in a direction that reduces the deviation. The measuring transmitter continuously feeds the actual output of the controlled object back to the system, forming a closed-loop control to control the auxiliary cylinder P4 to compensate for the deformation. The compensation error can be as low as 2mm, resulting in an end deformation of 18.9mm. Through high-precision dynamic compensation control technology, it is possible to compensate for deformations of at least 16.9mm and at most 20.9mm.

[0052] The embodiments of the present invention are described in detail above, but the contents are only preferred embodiments of the present invention and should not be considered to limit the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. A long arm span heavy load robot, characterized by: It includes a 4R-type series main chain and five parallel closed-loop structures, the five parallel closed-loop structures include parallel closed-loop 1, parallel closed-loop 2, parallel closed-loop 3, parallel closed-loop 4 and parallel closed-loop 5, and also include a lifting hydraulic cylinder, a telescopic hydraulic cylinder, a posture hydraulic cylinder and an auxiliary oil cylinder. The lifting hydraulic cylinder controls the lifting movement of the end effector by pushing the connecting rod 12 to lift and lower, and is amplified by the parallel closed-loop 2 and the parallel closed-loop 3; the telescopic hydraulic cylinder drives the parallel closed-loop 1, the parallel closed-loop 4 and the parallel closed-loop 5 to extend and retract together by pushing the connecting rod 11 to control the telescopic movement of the end effector; the posture hydraulic cylinder drives the parallel closed-loop 5 to control the posture of the end effector, decoupling the position and posture to achieve parallel The in-plane end effector pitches and rolls; the auxiliary oil cylinder assists the telescopic hydraulic cylinder drive; the parallel closed loop one includes a static platform, link six, link nine, and link ten; the parallel closed loop two includes link four, link ten, link twelve, and link thirteen; the parallel closed loop three includes link two, link three, link four, and link six; the parallel closed loop four includes link one, link four, link five, and link eight; the parallel closed loop five includes an end effector, link one, link fourteen, and link fifteen, the parallel closed loop one is connected to the parallel closed loop four by link seven, and the parallel closed loop four and the parallel closed loop five have a common link one; the parallel closed loop two and the parallel closed loop three have a common link four.

2. The long arm span heavy load robot according to claim 1, characterized in that: The connecting rod one, connecting rod two, connecting rod three, connecting rod four, connecting rod five, connecting rod six, connecting rod seven, connecting rod eight, connecting rod nine, connecting rod ten, connecting rod eleven, connecting rod twelve, connecting rod thirteen, connecting rod fourteen, and connecting rod fifteen constitute a connecting rod mechanism. The static platform, connecting rod mechanism, and end effector are installed on the mobile body as the main body of the robot mechanism. The mobile body is located on the guide rail. A rotary motor is installed at the center of the mobile body to control translation and rotation to realize two-axis translation and rotation.

3. A force-position hybrid control method for a long-reach heavy-load robot, implemented based on the long-reach heavy-load robot according to claim 1 or 2, characterized in that: The following steps are involved: S1. Position signal acquisition: Collect the position signals of each joint of the robot and calculate the actual position of the end effector through forward kinematics; S2. Force feedback signal acquisition: Collect the force feedback signal of the robot end effector and calculate the deviation between the actual force and the target force; S3, control command calculation: according to the preset forging trajectory and force target, combined with the end effector position signal and force feedback signal, the control command is calculated; S4. Adaptive control algorithm: Adaptive control algorithm is used to adjust the motion state of the robot to achieve precise control of position and force.

4. The force-position hybrid control method for a long-reach heavy-load robot according to claim 3, characterized in that: In step S1, each joint of the robot is equipped with an encoder for collecting position signals; the end effector is equipped with a force sensor for collecting force feedback signals; Assume that the robot has n joints and the position of the i-th joint is , then the joint position vector is: Position of the end effector x Calculated by the forward kinematics model: in, is the forward kinematics function of the robot, established by the Denavit-Hartenberg (DH) parametric method or the geometric method.

5. The force-position hybrid control method for a long-reach heavy-load robot according to claim 4, characterized in that: In step S2, let the actual force of the end effector be , the target force is , then the force deviation The calculation formula is: 。 6. The force-position hybrid control method for a long-reach heavy-load robot according to claim 5, characterized in that: In step S3, the target position of the end effector is set to , the actual location is , calculate the position deviation , Position control instructions This can be calculated using a proportional-derivative controller: in, Kp is the position proportional gain, Kd is the differential gain, t For time; Force control instructions This can be calculated using a proportional-integral controller: in, Kfp is the force proportional gain, Kfi is the integral gain; Finally, the mixed control instruction u is the position control instruction and force control instructions The weighted sum of: in, α is the weight coefficient, which is used to adjust the relative importance of position control and force control.

7. The force-position hybrid control method for a long-reach heavy-load robot according to claim 6, characterized in that: In step S4, the adaptive control algorithm adopts the model reference adaptive control method to make the robot's motion state approach the reference model by adjusting the control parameters in real time. The specific steps are as follows: Establish the dynamic model of the robot, assuming that the dynamic model of the robot is: in, is the mass matrix, are the Coriolis and centrifugal force matrices, G(q) is the gravity vector, is the first-order derivative of the joint position vector, is the second-order derivative of the joint position vector; Design a reference model and determine the desired motion state. Let the reference model be: Where r is the reference input, K 1 and K 2 are the two gain matrices of the reference model, is the joint position vector of the reference model, is the first-order derivative of the joint position vector of the reference model, is the second-order derivative of the joint position vector of the reference model; The adaptive control law adjusts the control parameters to make the actual motion state close to the reference model. θ To minimize the tracking error e : Where Γ is the adaptive gain matrix, is the regression matrix, is the first-order derivative of the control parameter.

8. A high-precision dynamic compensation method for a long-reach, heavy-load robot, implemented based on the long-reach, heavy-load robot according to claim 1 or 2, characterized in that: The following steps are involved: The deformation of the robot's mechanical arm is analyzed by the finite element method, the deformation function is fitted by the response surface method, and the deformation is iteratively compensated to achieve online compensation of mechanical deformation.

9. The high-precision dynamic compensation method for a long-arm-span heavy-load robot according to claim 8, characterized in that: Fitting the deformation function requires finite element-based experimental design and the selection of a high-precision response surface model: Based on the end-effector coordinates, the central composite design experimental method is used to generate multiple sets of experimental design points required for fitting the response surface model. A simulation platform is built to simulate each set of experimental points. The parameters of each set of experimental points correspond to each experimental model after the parameters are updated. Each updated model will undergo finite element simulation in turn until all experimental points are simulated and the final experimental design data is obtained. The response surface function used to fit the model is: in, is a first-order response surface model, is a second-order response surface model, is a third-order response surface model, is a fourth-order response surface model, is the function constant term, is the coefficient of the first-order term of the model function, is the first coefficient of the quadratic term of the model function, is the second coefficient of the quadratic term of the model function, is the cubic coefficient of the model function, is the coefficient of the quartic term of the model function, is the end effector coordinate, is the end effector coordinate exist i The specific value of The coordinates of the end effector are j The specific value of , s is the total number of experimental points; After completing the response surface model fitting, the fitting accuracy of the response surface model is evaluated. By selecting a small number of experimental points, the calculated values ​​of the response surface model are compared with the actual values ​​of the experimental points. The relative mean absolute error, relative maximum absolute error, root mean square error, R 2 These four accuracy evaluation indicators analyze the fitting accuracy of the response surface model. The formulas of these four evaluation indicators are as follows: Among them, RAAE is the relative average absolute error, RMAE is the relative maximum absolute error, RMSE is the root mean square error, R 2 is the variance, is the number of experimental points selected, For the i The true value of the experimental point, For the i The predicted value of each experimental point calculated by the response surface model is for The average value of Using the data obtained from the experimental design, a response surface model is established, and iterative compensation is proposed for deformation to achieve online compensation of mechanical deformation. By obtaining the set coordinate values ​​of the end effector and the force conditions of the connecting rod mechanism, the deformation of the connecting rod is calculated, and the driving auxiliary cylinder is controlled to compensate for the deformation. The fuzzy controller is combined with the PID controller and the closed-loop feedback mechanism to achieve online compensation of mechanical deformation and multi-link deformation compensation, thereby achieving high-precision control of the robot.

Images