Direct optimization determination of auxiliary test signals for linear problems with model uncertainty

Recently there has been increased interest in active approaches for fault detection which use auxiliary test signals. Theory and algorithms have been presented in the literature for the design of fault detection signals for linear systems with model uncertainty. These approaches cannot solve many problems with constraints. This paper gives the first direct optimization formulation of the more general constrained problem. The use of a direct optimization formulation allows the solution of problems not possible by the original algorithms such as problems with input and state constraints. Computational examples are given both to illustrate the theory and to discuss computational issues.