|
In this thesis, a noncircular cutting on lathe (NCL) using the
signals of tool position and differential motor current
is developed. First, an effective neural network is used
to learn a weighted combination of the tool position and
differential motor current of servosystem achieved from the
difference between the motor current under real cutting and that
of air cutting. Then a feedforward control based on the
learned model is designed to obtain an acceptable
tracking result. Although the feedforward control is achieved
from the consideration of tool position under the
cutting force, its performance cannot be ensured as the system
is subjected to uncertainties (e.g., noise, different cutting
condition, aging of system component). Under these
circumstances, an adaptation in control input generated
by a fuzzy sliding-mode control is then synthesized with the
previous feedforward control to reduce the cutting error. Thus
the proposed control contains two parts: one is feedforward
control, the other is fuzzy sliding-mode control. The
experimental results of NCL using the proposed control
including profile error of finished work piece and surface
roughness of finished work piece, are much ameliorated.
|