Optimizing stability bounds of finite-precision PID controller structures

This paper investigates a recently derived lower bound stability measure for sampled-data controller structures subject to finitewordlength (FWL) constraints. The optimal realization of the digital PID controller with FWL considerations is formulated as a nonlinear optimization problem, and an efficient strategy based on adaptive simulated annealing (ASA) is adopted to solve this complex optimization problem. A numerical example of optimizing the finite-precision PID controller structure for a simulated steel rolling mill system is presented to illustrate the effectiveness of the proposed strategy.