Neural network based control of micro-manipulator

Abstract

This paper describes a neural network based control approach for position control of a 2DOF micro-manipulator. This micro-manipulator consists of a parallel mechanism, which is fabricated from photo-sensitive micro-structured glass and two piezoelectric actuators (PEA), which actuate the "x" and "y" degrees of freedom on a working plane with dimensions of 50x50 micrometers. Because the control performance of a micro-manipulator severely deteriorates due to the hysteresis effect of PEAs, a neural network (NN) based position control is proposed in this study to improve control performance. A simulation model of the micro-manipulator and proposed neural network based control algorithm have been developed in Matlab/Simulink.