Motion Planning And Structural Deformation Compensation Of Multi-axis System

Multi-axis contour control system is an important branch of motion control. Itsmain use is to accomplish high-speed and high-precision contour control. Multi-axissystem, as the key technology of modern manufacturing equipments, to a great extent,indicates the industrialization level of a country. Therefore, it provides great economicand social benefits to do research on multi-axis system. To solve key issues duringhigh-speed and high-precision contour control for multi-axis systems, this dissertationmainly focuses on speed control, trajectory planning, trajectory tracking, and errorcompensation. The main works are as follows:During high-speed machining and rapid detecting under fast feed condition,over-cutting of work piece and tool chattering often occur when traditional speedcontrol strategy is employed. To solve this problem, the S-shape acceleration anddeceleration control algorithm is analyzed, and the discrete path point curvature isobtained by fifth-order spline curve fitting for discrete path point, and then a dynamicselecting method of blocks for velocity look-ahead control based on accumulation ofweighted curvatures of discrete points is proposed. Simulation results show that theproblems that required deceleration could not be finished with shorter look-ahead pathand more system sources are occupied with more look-ahead paths can be avoidedwhen the proposed dynamic velocity look-ahead control method is employed.Moreover, the problem between speed and accuracy, to some extent, is solved. And,system performance and machining efficiency are also improved.Cubic spline curve is used to fit path points and then position motion path isobtained. The Smooth posture path planning for end-effector is executed by use ofRodrigues parameters method. A path planning algorithm for complex space curvewith pose constraints is proposed based on prediction of principle error. Theeffectiveness of the proposed method for pose planning of complex trajectory isverified through simulations.To solve the uncertainties induced by friction and disturbances present in dynamicequations of digital equipments, the structure types and algorithm of neural network(NN) are introduced. An adaptive control algorithm is proposed based onsegmentation approximating for models by use of RBF NN, and asymptotic stabilityof the system is proved by Lyapunov method. Effectiveness of the proposed trajectorytracking method is verified by taking a2-DOF planar articulated robot as an example. To consider the influence of the dead zone nonlinear dynamic model, uncertaintymodel adaptive neural network controller is proposed based on the dead zone fuzzycompensator. The stability conditions of the system are proved by Lyapunov method,and the effectiveness of the control system is verified.As the nonlinear characteristics of link position and loading deformation, principleof support vector machine (SVM) is analyzed and a novel method for modelingdeformations of loaded components based on SVM is presented. Both ε-supportvector regression method and least square method are employed to model errorsintroduced by deformations. The simulation results show that the support vectorregression method is better than least square method, and structural error model basedon support vector regression is able to realize the error of the closed-loopcompensation. The simulation of2R manipulator structure deformation errormodeling and motion compensation show the SVR method could be used to modeldeformation errors effectively.