Options
2000
Conference Paper
Titel
On-line process control via adaptive symbolic dynamics
Abstract
Many successful concepts of nonlinear dynamics rely on the monitoring of trajectories in phase space. For ideal systems the resulting algorithms are very powerful and work well. Applying these methods to real systems, however, creates problems due to complex stochastic influences, non-stationary, or, an incomplete monitoring of the system variables. We propose a new concept for process control in such systems. Our approach relies on an adaptive version of a well-known technique for chaotic systems, namely symbolic dynamics. In contrast to trajectory based control mechanisms, the associated coarse graining of the state space guarantees greater flexibility and robustness against noise.