This work reports two modeling and control attempts performed on a positioning system comprising of linking SMA wires and an overlooking video system for on-line measurements. The first attempt takes the model by Ikuta and identifies experimentally the parameters of the SMA wire. The identified single wire model is then extended to a system of two SMA wires joining together at their tips, based upon which open loop position control of the linkage is then conducted. The approach, however, becomes too complicated when more SMA wires are involved. The second attempt utilizes a neurofuzzy based approach for positioning control of a linkage point joining together four SMA wires. The second approach involves four ANFIS neuro-networks with hybrid learning algorithm trained to model the currents to the SMA wires as functions of present and target positions of the linkage point. Experimentation for both the two-wires and four-wires system yield quite satisfactory performance.
展开▼
