Design of H2-optimal, finite word length, PID controllers

This paper uses genetic algorithms to find finite word length (FWL) controllers for fixed-point or floating-point implementation that optimize a cost function that is equivalent to an H-2 norm if the FWL effects are ignored. The formulation enables the consideration of all sources of FWL errors, i.e., coefficient, input and output quantization effects, and the roundoff error of the arithmetic operations. Simulation results illustrate the effectiveness of the methodology in developing a FWL PID control law for a steel rolling mill.