Designing experiments for controller perturbation theories-an example

Recent theoretical results have extended the reliability analysis for real-time digital control systems to cover temporary periods of controller misbehavior. Previously, reliability analyses assumed the controller always had to be in control, and periods of controller misbehavior were either ignored or declared to be system failure. It is possible, however, for a system to survive repeated controller perturbations because of plant dynamics. System inertia can prevent an incorrect command from leading to immediate catastrophe, and this time lag gives the controller an opportunity to recover. Recent advances place this idea on a quantitative basis by deriving results for the basic control criterion of asymptotic stability in terms of matrix norms. Translating a theoretical result into a laboratory experiment poses numerous problems and possibilities. We present the results of the experiment inducing some transmission perturbations on communication cable due to radiating EMI in the reverberation chamber. We consider four topics such as (i) parameter observation, (ii) predictive ability, (iii) asymptotic stability, and (iv) field data.