In this paper, a novel neural network is first used to learn the dynam
ics of the proposed linear servo motor system. Then a modified zero-ph
ase tracking controller (MZPTC) based on this learning model is design
ed to obtain an acceptable tracking result. Because an MZPTC involves
open-loop control, its performance cannot be assured, as the system is
subjected to external loads or the aging of system components. To ove
rcome this difficulty, an adaptation to the control input generated by
a fuzzy sliding-mode controller (FSMC) is then synthesized with the p
revious MZPTC. Hence, improved control performance (i.e., good traject
ory tracking under uncertain conditions, the reduction of control effo
rt as compared with an FSMC, and the ease of constructing a fuzzy tabl
e for a specific system performance), is achieved. The experimental pe
rformance of the proposed hybrid control technique (or so-called ''imp
roved fuzzy sliding-mode control'' (IFSMC)) is compared with those of
the MZPTC and FSMC.