System identification and control synthesis for a benchmark problem

In this paper we apply the integrated system identification/H_∞ control synthesis approach to the IFAC 1993 benchmark problem using the minimal amount of process information revealed. The resulting performance compares favorably to those using time-invariant control design techniques, and achieves prescribed performance for small and medium levels of plant uncertainties. For strong plant variations, the performance is inferior to most adaptive-based designs. An attempt to minimize input saturation effects by modifying the design objective function is also included. The present work conducts statistical confidence analysis with estimated noise variance and Fisher distribution