In theory and experiments, optimal and adaptive control strategies are empl
oyed to suppress chaotic convection in a thermal convection loop. The therm
al convection loop is a relatively simple experimental paradigm that exhibi
ts complex dynamic behavior and provides a convenient platform for evaluati
ng and comparing various control strategies. The objective of this study is
to evaluate the feasibility of employing optimal control and nonlinear est
imator to alter naturally occurring flow patterns and to compare the perfor
mance of the optimal controller with that of other controllers such as neur
al network controllers. It is demonstrated that when the system's model is
not known, experimental data alone can be utilized for the construction of
a proportional controller. (C) 1999 American Institute of Physics. [S1070-6
631(99)02806-8].