| As robots being more widely used in special and extreme environment such asdeep space exploration, nuclear industry and deep sea exploration, and due to the pres-ence of uncertainty in ordinary robot drive joints and traditional classical robot controlmethods cannot meet the requirements of operational environment, development of newhigh-performance robot driving joints and advanced and intelligent control methodshave become an important issue in the fields of aeronautics, astronautics, marine andnuclear engineering applications. Therefore, an integrated system (Patent PublicationNo.: ZL200910104663.7) of filter drive and intelligent robot and a highly reliable pre-cision drive (Patent Publication No.: ZL200910104672.6) have been invented by Pro-fessor Wang Jiaxu, etc. in the State Key Laboratory of Mechanical Transmission ofChongqing University, with special advantages in the aspects such as high precision,high reliability, long life, high torque, low power, small size, light weight and largetransmission ratio. Therefore，based on these patents, this paper studies the robot system,abbreviating form of”filter drive robot”, which is constituted by the filter drive me-chanism, link, terminal actuator and controller, and analysis of the hazards and com-pensation necessity of the robot filter drive mechanism uncertainties. It has been ob-tained the dynamic friction model parameters, stiffness and backlash of the filter drivemechanisms of the robot by experiment. Whether to consider the change of the robotfilter drive mechanism or joint friction, deadzone, backlash and stiffness, this thesis es-tablishes different Lagrange dynamics model, and proposes the fuzzy radius basis func-tion (RBF) neural network to compensate the uncertainties of the dynamic friction andbacklash, and also put forward the fuzzy logic method to compensate for the uncertaintyof the dead zone of the robot, and proposes digital discrete robust sliding mode con-trol(SMC) method, continuous robust SMC method, backstepping SMC method respec-tively, which is integrated with uncertainty feedforward compensator, and analysis theLyapunov stability of the various hybrid controllers, as well as the numerical simulation.The MATLAB simulation results show that these control methods are of importanttheoretical and practical values on the control and experimental of the filter drive robot.Based on analysis of the current development of robot drive mechanisms and theiruncertainty compensation methods and robot control technology at home and abroad,and the problems on harmonic reducer, RV reducer, this paper build a high precision, high reliability, high quality, long life robot, and puts forward a variety of SMC com-bining with intelligent compensation for filter drive mechanism uncertainty, which is ofpractical value and significance for building a filter drive robot prototype.The main works and innovation points of this thesis:1. Experimental research and parameters identification of robot filter drive me-chanism. Referring to harmonic driving mechanisms, experimental standards, normsand cases at home and abroad, this thesis proposes an experimental program for filterdriving mechanism. By line regression analysis of experimental data, it obtains the dy-namic friction model, stiffness, backlash and other parameters, which provide data sup-port to build the dynamic model and control of filter drive robot.2. Combining with nonlinear, strong coupling dynamic model of robot manipulator,this thesis presents a digital robust sliding mode control algorithm of robot，blockcompensates for the uncertainties of robot filter drive mechanism-LuGre dynamic fric-tion with three fuzzy RBF neural network, trains parameters of nonlinear dynamic fric-tion on-line and adaptively, and analyzes the Lyapunov stability of the algorithm. Thesimulation of two degrees of freedom (DOF) filter drive robot manipulator proves thatthe algorithm is characteristics for high accuracy, and high quality, stable and strongrobustness. Meanwhile, nonlinear kinetic phenomena, such as rhombus attractor, lie inthe kinetic properties of the friction model of robot manipulator.3. Combining with nonlinear, strong coupling dynamic model of robot manipulator,this thesis presents a robust sliding mode robot control algorithm，compensates for theuncertainties of robot filter drive mechanism respectively-LuGre dynamic friction withfuzzy RBF neural network and asymmetrical dead zone with fuzzy logic, trains para-meters of nonlinear dynamic friction and asymmetrical dead zone real-timely and adap-tively, to achieve an accurate reproduction of the actual robot system, and analyzes theLyapunov stability of the algorithm. The simulation of a two-degree-of-freedom filterdrive robot manipulator proves that the algorithm is characteristics for high accuracy,and high quality, stable and strong robustness. Meanwhile, nonlinear kinetic phenomena,such as control moment pulse compensation error, rhombus attractor, lie in the kineticproperties of the friction model of robot manipulator, and dead zone compensation canimprove the accuracy of trajectory tracking, dynamic response, stability, reliability ofrobot, and estimation of ε in the dead zone inverse model plays a key role in deadzonetorque compensation of the robot system. Meanwhile, initial position and orientationplay a vital role in robot control torque and stability. 4. Based on the considering of friction, backlash and flexible in robot filter drivemechanism, this thesis proposes to compensate the dynamic friction and backlash un-certainty existing in the filter drive joint with fuzzy neural network and train the para-meters of uncertainty self-adaptively and in real time, along with backstepping and selfadaptively SMC of robot. Additionally, the Lyapunov stability of the algorithm isproved. The simulation results show that the method improves the trajectory trackingprecision, the torque precision and stability of robot control. Meanwhile, the stiffness ofthe drive mechanism and the gravitational potential energy of the robot play a majorimpact for the trajectory tracking, control accuracy and stability of the robot. When therobot drive mechanism stiffness increased by9times, if ignoring the influence of robotgravity, then trajectory tracking accuracy of link1improves by22%, and pseu-do-control moment fluctuation amplitude increases by9.5times, if considering robotgravity, trajectory tracking accuracy is improved by67.9%, and pseudo-control momentfluctuation amplitude increases by22.5times. At this two stiffness circumstances, thegravity cause the stability of the robot control torque deteriorates. |
PDF Full Text Request