Feed forward initialization of hysteretic systems

The paper analyzes a strategy to force step-convergent dynamical hysteretic systems to a well-defined output value using only feed forward. Due to the multi-valued input-output relation of hysteresis, the relation between a constant input and the corresponding steady-state output is not unique. By using a well-designed input trajectory, based on qualitative system behavior only, such a system can still be forced to a predefined output value. Moreover, for reasons of performance the initialization time has to be as short as possible. The rate-independent Duhem and the multi-play hysteresis models are used to illustrate, analyze and synthesize an appropriate initializing input to obtain the required output without any sensor information. Dynamics are included using the nonlinear feedback model with dead-zone.