L∞ identification and model reduction using a learning genetic algorithm

This paper develops a Boltzmann learning enhanced genetic algorithm for L_∞ norm based system identification and model reduction for robust control applications. Using this technique, both a globally optimised nominal model and an error bounding function for additive and multiplicative uncertainties can be obtained. It can also offer a tighter L_∞ error bound and is applicable to both continuous and discrete-time systems.