Due to the low cooling rate of macroscopic SMA actuators in still ambient air, control of these actuators is usually not a topic that goes beyond the realms of classical PI control theory.
In some circumstances, however, things change drastically. Examples are applications in an efficient cooling environment like underwater applications or flight at high altitude. Of most interest currently are MEMS applications, where, even at ambient conditions, the increased surface to volume ratio provides high cooling rates. In such cases, higher actuation frequencies can be achieved, and the inherent nonlinear behavior, mainly caused by the hysteresis present in these materials, makes it impossible to track set point signals using a conventional feedback control method. The same problem is encountered in other active materials, e.g., piezoceramics and magnetostrictive materials.
A way to overcome this problem is to implement a model into the control algorithm that compensates for the effect of the hysteresis (inverse compensator methods). It has been successfully done in several cases, mostly on the basis of phenomenological models from the Preisach type. This approach, however, poses considerable problems with respect to identification and interpretability of the parameters. Another common feature of all model-based control algorithms published so far is that, at the best, they address stability issues, but never look at optimality, e.g., with respect to speed of adjustment or energy consumption.
The seminar gives an overview of an optimal control algorithm employing a SMA model derived from statistical thermodynamics. It has a limited number of parameters, and all of them are clearly interpretable from a physical point of view. Based on this approach, a method has been designed to allow for real-time applications, taking unforeseeable parameter changes into account and delivering results for an optimal control in extremely short time.