Dynamic error compensation and control method for cutting machining robot
Technical Field
The invention relates to the technical field of robot machining, in particular to a dynamic error compensation and control method of a cutting machining robot.
Background
The progress and application of industrial robot technology are important means and key links for promoting intelligent manufacturing development in China. The industrial robot has the advantages of high flexibility, low cost, large working space and flexible pose control, is applicable to cutting processing, can adapt to the modern production mode requirements of multi-variety, small-batch and on-site processing, obviously reduces the production cost, improves the utilization rate of equipment and processing space, and effectively improves the technical innovation speed and the enterprise competitiveness. However, the industrial robot has the problems of low repeated positioning precision, poor rigidity, complicated error analysis control and the like, so that the application of the robot in the field of cutting processing is greatly limited.
How to effectively analyze the errors of the cutting machining robot and improve the machining precision is a key problem for pushing the robot to apply cutting machining. Errors in the robot cutting process mainly include static errors and dynamic errors. Static errors can be corrected by directly modifying the structural parameters of the controller. The dynamic error comprises three sub-types, namely a stable deformation error, a low-frequency vibration error and a high-frequency vibration error. In order to improve the machining precision and the surface quality, a dynamic error compensation and control method of a cutting machining robot is needed, and the above error compensation methods are researched, so that high-precision robot track control and high-frequency vibration suppression are realized.
Disclosure of Invention
Aiming at the needs and the shortcomings of the prior art, the invention provides a dynamic error compensation and control method of a cutting machining robot.
The invention relates to a dynamic error compensation and control method of a cutting machining robot, which solves the technical problems and adopts the following technical scheme:
a dynamic error compensation and control method of a cutting machining robot, the dynamic error of the method comprises three kinds of stable deformation errors, low-frequency vibration errors and high-frequency vibration errors, and the method comprises the following specific implementation modes:
a) When the dynamic error is a stable deformation error, error compensation and control are carried out based on an off-line compensation module and an on-line correction module;
b) When the dynamic error is a low-frequency vibration error, an open-loop control method based on modal analysis is adopted, and the vibration is restrained by applying extra main power at the joint of the robot to counteract the exciting force causing a low-order mode, so that error compensation and control are carried out;
c) When the dynamic error is a high-frequency vibration error, a robot dynamics control algorithm based on a parameter self-correction active disturbance rejection strategy is provided, meanwhile, an acceleration sensor is additionally arranged in the robot, vibration signals in the machining process of the robot are collected in real time through the acceleration sensor, the high-frequency vibration is regarded as disturbance to a robot control system from the outside, the robot dynamics control algorithm corrects control parameters in real time on line, so that the robot control system keeps excellent control performance in all operation periods, and high-frequency vibration suppression and high-precision track control are realized, namely high-frequency vibration error compensation and control are realized.
Specifically, in the process of compensating and controlling the low-frequency vibration error:
based on the dynamic stiffness model of the robot complete machine and vibration data acquired in the processing process, a mode analysis method is utilized to obtain a main mode of structural vibration, and aiming at the main mode, amplitude, phase and period information of active control force required by each joint are obtained through calculation and are applied to the joints in real time so as to offset exciting force of the corresponding mode.
Specifically, the related specific operations of error compensation and control by the offline compensation module are as follows:
introducing a cutting force model into a robot control system, predicting cutting force at each interpolation point according to a target track and cutting parameters, calculating deformation of each interpolation point based on a dynamic stiffness model of the whole robot, compensating each deformation into a theoretical track to obtain a corrected tool track, and simultaneously storing predicted cutting force at each interpolation point;
the specific operation of error compensation and control by the related online correction module is as follows:
in the real-time interpolation process, the robot performs cutting motion by taking the corrected tool path as a target path, simultaneously, the force sensor/torque sensor is used for measuring and collecting cutting force in real time, cutting force errors are obtained after the cutting force errors are compared with predicted cutting force, the tool path is finely adjusted through the cutting force errors, the tool path is further corrected, the real-time motion of the robot is controlled, and the track precision of the cutting machining of the robot is improved.
More specifically, the cutting force model involved is a linear cutting force model, the cutting force coefficient of which is a constant, and the linear cutting force model recognizes the cutting force coefficient by calculating the average milling force in one rotation period of the tool, thereby predicting the three-way cutting force according to the target track and the cutting parameters.
Specifically, the dynamic stiffness model of the whole robot is obtained by the following steps:
step one, acquiring a description file of robot body parameters, and calling a model class library through the description file;
step two, firstly, carrying out theoretical analysis on components and component contact characteristics which influence error analysis in the robot, establishing dynamic stiffness theoretical sub-models of the components and component contact characteristics by adopting a calculation modal analysis method,
then, carrying out modal experiments on components and component contact characteristics affecting error analysis in the robot, establishing a dynamic stiffness experiment sub-model and a modal sub-model of the components and component contact characteristics through acquisition and processing of excitation and response data, identifying modal parameters and carrying out verification of the dynamic stiffness experiment sub-model;
then, removing minimum factors influencing error analysis according to theoretical analysis results, simplifying variables of a dynamic stiffness theoretical sub-model, and correcting the dynamic stiffness theoretical sub-model of corresponding components and component contact characteristics through a dynamic stiffness experimental sub-model to obtain a dynamic stiffness sub-model meeting the accuracy requirement;
finally, synthesizing the dynamic stiffness sub-model meeting the precision requirement into a complete machine dynamic stiffness theoretical model of the robot through a modal synthesis theory;
thirdly, performing a modal experiment on the whole robot, establishing a dynamic stiffness experimental model and a modal model of the whole robot through acquisition and processing of excitation and response data, identifying modal parameters and verifying the dynamic stiffness experimental model of the whole robot;
and step four, correcting the whole dynamic stiffness theoretical model through the whole dynamic stiffness experimental model to obtain the whole dynamic stiffness model meeting the precision requirement.
In the first step, the description file of the related robot body parameters can completely describe the normative file of the robot body parameters, and the generation process of the description file comprises the following steps:
a1 According to the structural parameters of the robot, the sensitivity of the dynamic characteristics of the robot is analyzed, the structural parameters comprise three types of functional components, a joint connection mode and a structural component, wherein the functional components comprise a driving unit, a connecting rod unit and a speed reducer unit, the joint connection mode comprises integrated connection and coupling connection, and the structural component comprises serial connection, parallel connection and serial-parallel connection;
a2 Formulating precision criteria and specifications of the parameter description according to the analysis result;
a3 Using standardized XML file format as carrier, and after sensitivity analysis is carried out on three parameters of functional component, articulation mode and structure composition, respectively generating description file of functional component parameter, description file of articulation mode parameter and description file of structure composition parameter;
a4 The open source xml file generator integrates the description file of the functional component parameters, the description file of the joint connection mode parameters and the description file of the structural composition parameters to generate the description file of the robot body parameters;
the related model class library is designed by applying an object-oriented method, and the design process comprises the following steps:
b1 Establishing a robot component error analysis base class by means of a tool SQLserver, wherein the base class comprises an ID, an error analysis method and other attributes and is used for analyzing the reasons for generating component errors;
b2 Deriving a functional component class and an articulation mode class from the robot component error analysis base class, wherein,
the functional component class is analysis of the influence of rigidity of the component on errors, and derives two subclasses of rod class and speed reducer class;
the joint connection mode is analysis of the influence of the contact stiffness of the assembly, and derives two subclasses of integral connection and coupling connection;
b3 Packaging base class and derived subclasses, and calling the packaging information of the model class library by the description file of the robot body parameters when the dynamic stiffness theoretical model of the whole robot is established.
Further, in a 1), according to the structural parameters of the robot, the sensitivity analysis of the dynamic characteristics of the robot includes:
establishing a robot simulation model according to the robot structural parameters;
establishing a simulated dynamic stiffness model according to the robot simulation model;
the simulation dynamic stiffness model is used as an objective function, the sensitivity analysis is carried out on the structural parameters of the robot, the structural parameters are divided into sensitive parameters and non-sensitive parameters according to the sensitivity analysis result, wherein the sensitive parameters are more accurately measured by using an instrument with higher precision, the parameter calculation and description are carried out by adopting an algorithm with higher precision, the measurement and description process of the non-sensitive parameters can be simplified, and the sensitivity analysis result of the simulation dynamic stiffness model can provide basis for the power modification of the robot cutting processing system;
the sensitivity of the structural parameter change of the robot to the dynamic characteristic influence of the robot control system is determined, the sensitivity of the structural parameter is weighed by a mechanical engineer through calculation, programming and analysis, and then the precision criterion and specification of the parameter description are manually formulated.
Specifically, in the fourth step, the specific process of correcting the whole maneuvering stiffness theoretical model through the whole maneuvering stiffness experimental model is as follows:
step 1: inputting the technological parameters into a controller, executing a processing track by the controller, and acquiring experimental data through a dynamometer and an acceleration sensor;
step 2: performing Fourier transform on the acquired acceleration data to obtain vibration amplitude spectrum data, performing logarithmic operation on the amplitude spectrum data, and then obtaining amplitude spectrum cepstrum data through inverse Fourier transform;
step 3: performing exponential window filtering operation on the amplitude spectrum cepstrum data obtained in the step 2, and performing cepstrum operation on the filtered cepstrum data to obtain vibration amplitude spectrum data under random excitation;
step 4: performing Fourier transform on the acquired acceleration data to obtain vibration phase spectrum data;
step 5: performing inverse Fourier transform by combining the amplitude spectrum and the phase spectrum data obtained in the step 3 and the step 4 to obtain an acceleration time domain signal under random excitation in the processing process;
step 6: and 5, identifying modal parameters by using the time domain signal obtained in the step 5 through a least square complex frequency domain method, obtaining a functional relation between the end deformation of the cutting robot and the changing cutting force, and further constructing the dynamic stiffness model of the whole cutting robot, which meets the precision requirement.
Specifically, the robot dynamics control algorithm involved includes: the system comprises a flexible robot dynamics model, a tracking differential module, an expansion state observation module, a nonlinear combination control module and a parameter real-time self-correction module based on fuzzy control;
the flexible robot dynamics model is established by a Lagrangian method based on the analysis result of a parameterized error analysis model in a robot control system; researching the principle of flexible robot dynamics model compensation active disturbance rejection control and compensating to an expanded state observation module;
the tracking differential module outputs a target bit value q according to a robot control system d Arrange transition procedure q d1 And extracts its differential signal q d2 ;
The extended state observation module estimates the state of the robot according to the output signal q of the robot control system
Robot state->
The total disturbance f acting on the robot, and the expansion state observation module then follows the transition process q of the tracking differential module
d1 Differential signal q
d2 Obtaining the state error epsilon of the robot
1 、ε
2 Wherein->
The nonlinear combination control module is further used for controlling the state error epsilon of the robot 1 、ε 2 To obtain the nonlinear control law tau 0 Based on nonlinear control law tau 0 The total disturbance f and the control parameter b acted on the robot obtain the input control moment tau of the robot;
by summarizing the nonlinear control law tau
0 The real-time parameter self-correction module based on fuzzy control is designed, and the real-time parameter self-correction module based on fuzzy control designs a fuzzy rule according to manual parameter adjustment experience, so that the control parameter b approaches to a true value, and meanwhile, the real-time parameter self-correction module based on fuzzy control is used for adjusting parameters according to tracking state error epsilon
1 The gain of the extended state observation module is adjusted to achieve the effects of big error, small gain and small error, big gain, so that the extended state observation module can rapidly and accurately estimate the state of the robot
Robot state->
The sum disturbance f reduces the noise impact at the same time.
The dynamic error compensation and control method of the cutting robot has the beneficial effects that compared with the prior art:
the invention provides a compensation and control method based on an offline compensation module and an online correction module for stable deformation errors, adopts an open loop control method based on modal analysis for low-frequency vibration errors to restrain the influence of low-frequency modes on vibration, and provides a robot dynamics control algorithm based on a parameter self-correction self-disturbance rejection strategy for high-frequency vibration errors, so that high-precision track control is ensured, and meanwhile, the restraint on high-frequency vibration is realized, and therefore, the precision and the surface quality of the cutting machining of the robot are remarkably improved by adopting different compensation and control methods for different types of dynamic errors.
Drawings
FIG. 1 is a flow chart of the invention for establishing a dynamic stiffness model of a robot complete machine;
FIG. 2 is a flow chart of the invention for establishing a theoretical model of the dynamic stiffness of the whole machine;
FIG. 3 is a flow chart of generating a robot ontology parameter description file according to the present invention;
FIG. 4 is a block diagram of the flow of the architecture of the robot dynamics control algorithm of the present invention;
FIG. 5 is a block diagram of a process for compensating and controlling a steady deformation error in accordance with the present invention.
The reference numerals in the drawings represent:
1. a flexible robot dynamics model, 2, a tracking differential module, 3, an expansion state observation module,
4. a nonlinear combination control module 5, a parameter real-time self-correction module based on fuzzy control,
6. and the off-line compensation module 7 is an on-line correction module.
Detailed Description
In order to make the technical scheme, the technical problems to be solved and the technical effects of the invention more clear, the technical scheme of the invention is clearly and completely described below by combining specific embodiments.
Embodiment one:
the embodiment provides a dynamic error compensation and control method of a cutting machining robot, wherein the dynamic error of the method comprises three types of stable deformation errors, low-frequency vibration errors and high-frequency vibration errors. The method comprises the following specific implementation modes:
a) When the dynamic error is a stable deformation error, error compensation and control are carried out based on the offline compensation module 6 and the online correction module 7;
b) When the dynamic error is a low-frequency vibration error, an open-loop control method based on modal analysis is adopted, and the vibration is restrained by applying extra main power at the joint of the robot to counteract the exciting force causing a low-order mode, so that error compensation and control are carried out;
c) When the dynamic error is a high-frequency vibration error, a robot dynamics control algorithm based on a parameter self-correction active disturbance rejection strategy is provided, meanwhile, an acceleration sensor is additionally arranged in the robot, vibration signals in the machining process of the robot are collected in real time through the acceleration sensor, the high-frequency vibration is regarded as disturbance to a robot control system from the outside, the robot dynamics control algorithm corrects control parameters in real time on line, so that the robot control system keeps excellent control performance in all operation periods, and high-frequency vibration suppression and high-precision track control are realized, namely high-frequency vibration error compensation and control are realized.
In the present embodiment, in the process of compensating and controlling the low frequency vibration error:
based on the dynamic stiffness model of the robot complete machine and vibration data acquired in the processing process, a mode analysis method is utilized to obtain a main mode of structural vibration, and aiming at the main mode, amplitude, phase and period information of active control force required by each joint are obtained through calculation and are applied to the joints in real time so as to offset exciting force of the corresponding mode.
Referring to fig. 5, in this embodiment, the specific operations of the offline compensation module 6 involved in error compensation and control are as follows:
introducing a cutting force model into a robot control system, predicting cutting force at each interpolation point according to a target track and cutting parameters, calculating deformation of each interpolation point based on a dynamic stiffness model of the whole robot, compensating each deformation into a theoretical track to obtain a corrected tool track, and simultaneously storing predicted cutting force at each interpolation point;
the related on-line correction module 7 performs the specific operations of error compensation and control:
in the real-time interpolation process, the robot performs cutting motion by taking the corrected tool path as a target path, simultaneously, the force sensor/torque sensor is used for measuring and collecting cutting force in real time, cutting force errors are obtained after the cutting force errors are compared with predicted cutting force, the tool path is finely adjusted through the cutting force errors, the tool path is further corrected, the real-time motion of the robot is controlled, and the track precision of the cutting machining of the robot is improved.
In this embodiment, the cutting force model involved is a linear cutting force model, the cutting force coefficient of which is a constant, and the linear cutting force model recognizes the cutting force coefficient by calculating the average milling force in one rotation period of the tool, thereby predicting the three-way cutting force according to the target trajectory and the cutting parameters.
In this embodiment, referring to fig. 1, the dynamic stiffness model of the whole robot is obtained by the following steps:
step one, acquiring a description file of robot body parameters, and calling a model class library through the description file;
step two, combining with figure 2, firstly, carrying out theoretical analysis on components and component contact characteristics affecting error analysis in a robot, establishing dynamic stiffness theoretical sub-models of the components and component contact characteristics by adopting a calculation modal analysis method,
then, carrying out modal experiments on components and component contact characteristics affecting error analysis in the robot, establishing a dynamic stiffness experiment sub-model and a modal sub-model of the components and component contact characteristics through acquisition and processing of excitation and response data, identifying modal parameters and carrying out verification of the dynamic stiffness experiment sub-model;
then, removing minimum factors influencing error analysis according to theoretical analysis results, simplifying variables of a dynamic stiffness theoretical sub-model, and correcting the dynamic stiffness theoretical sub-model of corresponding components and component contact characteristics through a dynamic stiffness experimental sub-model to obtain a dynamic stiffness sub-model meeting the accuracy requirement;
finally, synthesizing the dynamic stiffness sub-model meeting the precision requirement into a complete machine dynamic stiffness theoretical model of the robot through a modal synthesis theory;
thirdly, performing a modal experiment on the whole robot, establishing a dynamic stiffness experimental model and a modal model of the whole robot through acquisition and processing of excitation and response data, identifying modal parameters and verifying the dynamic stiffness experimental model of the whole robot;
and step four, correcting the whole dynamic stiffness theoretical model through the whole dynamic stiffness experimental model to obtain the whole dynamic stiffness model meeting the precision requirement.
In the first step, referring to fig. 3, the description file of the robot body parameter involved can be a normative file that completely describes the robot body parameter, and the generation process of the description file includes:
a1 According to the structural parameters of the robot, the sensitivity of the dynamic characteristics of the robot is analyzed, wherein the structural parameters comprise three types of functional components, a joint connection mode and a structural component, the functional components comprise a driving unit, a connecting rod unit and a speed reducer unit, the joint connection mode comprises integrated connection and shaft coupling connection, and the structural component comprises serial connection, parallel connection and serial-parallel connection;
a2 Formulating precision criteria and specifications of the parameter description according to the analysis result;
a3 Using standardized XML file format as carrier, and after sensitivity analysis is carried out on three parameters of functional component, articulation mode and structure composition, respectively generating description file of functional component parameter, description file of articulation mode parameter and description file of structure composition parameter;
a4 The open source xml file generator integrates the description file of the functional component parameters, the description file of the articulation mode parameters and the description file of the structural composition parameters to generate the description file of the robot body parameters.
In the a 1), the sensitivity analysis of the dynamic characteristics of the robot according to the structural parameters of the robot includes:
establishing a robot simulation model according to the robot structural parameters;
establishing a simulated dynamic stiffness model according to the robot simulation model;
the simulation dynamic stiffness model is used as an objective function, the sensitivity analysis is carried out on the structural parameters of the robot, the structural parameters are divided into sensitive parameters and non-sensitive parameters according to the sensitivity analysis result, wherein the sensitive parameters are more accurately measured by using an instrument with higher precision, the parameter calculation and description are carried out by adopting an algorithm with higher precision, the measurement and description process of the non-sensitive parameters can be simplified, and the sensitivity analysis result of the simulation dynamic stiffness model can provide basis for the power modification of the robot cutting processing system;
the sensitivity of the structural parameter change of the robot to the dynamic characteristic influence of the robot control system is determined, the sensitivity of the structural parameter is weighed by a mechanical engineer through calculation, programming and analysis, and then the precision criterion and specification of the parameter description are manually formulated.
In the first step, the model class library is designed by applying an object-oriented method, and the design process comprises the following steps:
b1 Establishing a robot component error analysis base class by means of a tool SQLserver, wherein the base class comprises an ID, an error analysis method and other attributes and is used for analyzing the reasons for generating component errors;
b2 Deriving a functional component class and an articulation mode class from the robot component error analysis base class, wherein,
the functional component class is analysis of the influence of rigidity of the component on errors, and derives two subclasses of rod class and speed reducer class;
the joint connection mode is analysis of the influence of the contact stiffness of the assembly, and derives two subclasses of integral connection and coupling connection;
b3 Packaging base class and derived subclasses, and calling the packaging information of the model class library by the description file of the robot body parameters when the dynamic stiffness theoretical model of the whole robot is established.
In the fourth step, the specific process of correcting the whole maneuvering stiffness theoretical model through the whole maneuvering stiffness experimental model is as follows:
step 1: inputting the technological parameters into a controller, executing a processing track by the controller, and acquiring experimental data through a dynamometer and an acceleration sensor;
step 2: performing Fourier transform on the acquired acceleration data to obtain vibration amplitude spectrum data, performing logarithmic operation on the amplitude spectrum data, and then obtaining amplitude spectrum cepstrum data through inverse Fourier transform;
step 3: performing exponential window filtering operation on the amplitude spectrum cepstrum data obtained in the step 2, and performing cepstrum operation on the filtered cepstrum data to obtain vibration amplitude spectrum data under random excitation;
step 4: performing Fourier transform on the acquired acceleration data to obtain vibration phase spectrum data;
step 5: performing inverse Fourier transform by combining the amplitude spectrum and the phase spectrum data obtained in the step 3 and the step 4 to obtain an acceleration time domain signal under random excitation in the processing process;
step 6: and 5, identifying modal parameters by using the time domain signal obtained in the step 5 through a least square complex frequency domain method, obtaining a functional relation between the end deformation of the cutting robot and the changing cutting force, and further constructing the dynamic stiffness model of the whole cutting robot, which meets the precision requirement.
In this embodiment, with reference to fig. 4, the robot dynamics control algorithm involved comprises: the system comprises a flexible robot dynamics model 1, a tracking differential module 2, an expansion state observation module 3, a nonlinear combination control module 4 and a parameter real-time self-correction module 5 based on fuzzy control;
the flexible robot dynamics model 1 is established by a Lagrangian method based on the analysis result of a parameterized error analysis model in a robot control system; researching the principle of compensating the active disturbance rejection control of the flexible robot dynamic model 1 and compensating the active disturbance rejection control to the extended state observation module 3;
the tracking differential module 2 outputs a target bit value q according to a robot control system d Arrange transition procedure q d1 And extracts its differential signal q d2 ;
The extended state observation module 3 estimates the robot state from the output signal q of the robot control system
Robot state->
The total disturbance f acting on the robot, the extended state observation module 3 then follows the transition process q of the tracking differential module 2
d1 Differential signal q
d2 Obtaining the state error epsilon of the robot
1 、ε
2 Wherein->
The nonlinear combination control module 4 is further based on the state error epsilon of the robot
1 、ε
2 To obtain the nonlinear control law tau
0 Based on nonlinear control law tau
0 The total disturbance f and the control parameter b acted on the robot obtain the input control moment tau of the robot;
by summarizing the nonlinear control law tau
0 The parameter real-time self-correction module 5 based on fuzzy control is designed, and the parameter real-time self-correction module 5 based on fuzzy control designs fuzzy rules according to manual parameter adjustment experience, so that the control parameter b approaches to a true value, and meanwhile, the control parameter b is close to a tracking state error epsilon according to tracking state error epsilon
1 Adjusting the gain of the extended state observation module 3The effect of 'big error small gain, small error big gain' is achieved, so that the extended state observation module 3 can quickly and accurately estimate the state of the robot
Robot state->
The sum disturbance f reduces the noise impact at the same time. In summary, the dynamic error compensation and control method of the cutting robot provided by the invention is adopted, and the accuracy and the surface quality of the cutting robot are obviously improved by adopting different compensation and control methods for different types of dynamic errors. The foregoing has outlined rather broadly the principles and embodiments of the present invention in order that the detailed description of the invention may be better understood. Based on the above-mentioned embodiments of the present invention, any improvements and modifications made by those skilled in the art without departing from the principles of the present invention should fall within the scope of the present invention. />