With their significant advantages over traditional electromagnetic motors,
ultrasonic motors (USMs) are becoming attractive for mechatronic applicatio
ns. Since USMs suffer from a lack of applicable mathematical model while th
eir speed characteristics are heavily nonlinear and time varying, it used t
o be difficult to apply them for servo application. This paper presents a f
uzzy ne ural network controller for servo position control of an USM. It co
mbines both the knowledge-based fuzzy logic and the learning-incorporated n
eural network. As a result, it compensates the nonlinear behavior of the mo
tor and optimizes its performance on-line. To further improve the motor per
formance, both of their control inputs, namely the driving frequency and ph
ase difference of the 2-phase inverter waveforms, are employed. Experiments
are then performed for various reference inputs. The results demonstrate s
uperiority of the controller in terms of tracking and steady-state performa
nce.