| Summary: | 碩士 === 輔仁大學 === 電機工程學系碩士班 === 107 === This thesis aims at the development of a three-axial translational parallel manipulator (TPM) with the control system, using a reconfigurable field-programmable gate array (FPGA) platform. In order to yield sufficient workspace for the parallel manipulator as well as the low cost of the actuators, the rodless pneumatic cylinders are adopted in the actuated mechanisms which were arranged in a form of horizontal geometry by the closed chain and equipped with carbon fiber links and universal joints. For this TPM, the vector-loop closure approach is employed in the kinematic analysis to obtain the forward/inverse kinematics relation between each actuated joint and the end-effector. After that, the velocity relations between each actuated joint and the end-effector can then be derived from the Jacobian matrix. The dynamics analysis contains the modeling of the TPM dynamics through the principle of virtual work, and the mathematical derivation of designed pneumatic actuation system. As to the control system design, the dual-loop feedback control configuration and the inverse dynamics control strategy are presented to achieve path tracking control of the developed TPM. Moreover, in order to implement the control system in the FPGA, the multilayer neural network (NN) is trained to approximate the behavior of the inverse dynamics control strategy and combined with the inner/outer control loop for each pneumatic actuator of TPM. The experimental tests, including the single-axial pneumatic actuation system and the TPM, are conducted to confirm that the proposed FPGA-based control system can ensure the high-accuracy performance in the trajectory tracking control of the TPM.
|
|---|
