Skip to Main Content
We are interested in controlling ignition, extinction and flickering of thermal patterns on a catalytic wafer for exothermic oxidation reactions with video feedback. The infrared (IR) video imaging Â¿(x,y,t) provides a high resolution distributed sensor for the thermal pattern with 7140 pixels. However, since the manipulated variable u(t), usually taken to be the reactant feed flow rate, is a lumped parameter, the feedback law requires the projection of Â¿ onto the real line P:Â¿(x,y,t)Â¿u(t). We construct the map P here by extracting the most sensitive spatial structure via a Karthunen-Loeve (KL) scheme and demonstrate that it allows video feedback to suppress ignitio or extinction. In some mor esoteric conditions, we can extract low-dimensional nonlinear dynamic models from the spatio-temporal dynamics exhibited by our system. The dynamic model is then used to design a better controller. This is shown to supress self-sustained oscillations in our experiments.