Fisher Information as a Novel Tool for Process Control Applications

摘要

Among the various theoretical interpretations of Fisher information, it can be considered as a measure of the state of order or organization of a dynamic system. Such an interpretation has lead to use Fisher information as a sustainability index. Nevertheless, the state of order of a dynamic system is also related to its degree of variability. In that context, the aim of this paper it to report the most recent developments in our effort to introduce the concept of Fisher information as a novel tool for stability analysis in open loop systems, and as a criteria for controller tuning in closed loop systems. We have studied the behavior of Fisher information in terms of variability and performance of a dynamic system through parametric analyses. We show that the maximization of Fisher information as a performance index can be a useful tool for the selection of model and controller parameters. This approach is illustrated through two simple chemical engineering dynamic models. Main conclusion is that Fisher information can be used as a tuning tool, which results in dynamics systems with less oscillatory dynamic performance.