Abstract
A performance sensitivity is presented from the viewpoint of system uncertainties for the optimization problems in a class of distributed parameter systems.
Necessary condition for the performance sensitivity which is defined as the maximum decrease in the value of the objective function caused by hypothetical worst parameter variations is derived by application of the maximum principle to a system with distributed parameters.
A design method is derived for obtaining the optimal performance sensitivity by applying the min-max criterion in the game-theoretical sense.
The method is successfully applied to the design and control of a tubular reactor with catalyst deactivation.
It is shown that an optimal reactor which has a low sensitivity to process parameter variations can be designed by the proposed method.
