| Due to the strong demand for ultra-precision positioning systems in different areas of science and technology, new precision positioning devices are under development. Piezoceramic actuated positioners seem to be most promising in many applications due to the achievable accuracy of piezoceramic actuators. Piezoceramic actuated positioners (piezomotors) are often made-up of a monolithic frame that houses a set of piezoceramic actuators. Different designs of piezomotors, offering a wide array of characteristics, have been introduced by researchers but essentially no work has been done on developing appropriate control laws and strategies that lead to good performance in a wide number of positioning tasks. In this thesis, control methodologies for linear inchworm piezomotors are developed for tracking and positioning applications.; For this purpose, an inchworm piezomotor is designed and modeled. The design developed involves a monolithic frame that houses four piezoactuators, of which two are used for the clamping and the remaining two are used for motion generation. The frame is developed such that the expansion of the piezoactuators generating the motion is amplified at the load end of the positioner. Two control methodologies are then developed for the designed piezomotor. The first developed control methodology is a combination of an Automatic Clamp Switching (ACS) mechanism and a Conventional Controller (CC). ACS provides the best timing for the clamp switching operation while the CC provides the control signal for the forwarding piezostacks. This controller requires, for its implementation, two position sensors with one providing information on the absolute position of the motor and the second providing information on the relative position of one clamping chamber with respect to the other. The second controller is neural network based and is characterized by the learning-by-training property of neural networks that enables fast control design and implementation. A Feed Forward Neural Network (FFNN) and a perceptron network are combined together to form the second controller which requires only one position sensor to be part of the controller. The performance characteristics of the two controllers developed are assessed based on an extensive numerical study of the positioner in tracking and positioning tasks. The simulation results indicate the good ability of the two controllers in positioning tasks and in tracking tasks as long as the desired trajectory is not varying faster than the maximum achievable speed of the motor. |
