Research Of Giant Magnetostrictive Actuator Algorithm And Device Based On Model Reference Adaptive Control

Because of it's fast reaction and giant magnetostriction coefficient, Giant magnetostrictive actuato(rGMA)is adopted into the micron and submicron instrument. But, GMA has inherent hysteresis, which will cause big trouble for accurate, dynamic and adaptive control. Based on the neural net model of GMA, model reference adaptive control has been deeply studied in this paper.Based on the model reference adaptive control and neural model of GMA, neural controller and neural unit PID are respectively introduced into model reference adaptive control, to realize and improve the control performance. First, through the simulation result of the position tracking experiment, tracking experiment and complex tracking experiment, the neural net model reference adaptive control can control the GMA accurately and dynamically, to some extent. However, there are still some defects in the neural model reference control, such as weak adaptive capability, bad control accuracy and dynamic performance. Against the problem in the neural net model reference adaptive control, neural unit PID control is proposed. By analyzing the simulation result of position tracking experiment, tracking experiment and complex tracking experiment on neural unit PID control and comparing with neural net model reference adaptive control, the result shows that neural unit PID can overcome the problems in the neural net controller. Meanwhile, neural unit PID controller has great adaptive capability and the on-line control can be achieved by it.On the other hand, the duple-direction controllable constant-current source of GMA is studied and its protection circuit is completed, in this paper. Preliminary mechanical size of GAM has been done. Adopting TMS320F2812, design hardware controller of GMA, and preliminarily debugged by using CCS3.3.