Comparative performance of LQG versus PID optimized by swarm approaches: a case study on a biomedical ventilation system

Abstract

This paper compares the Linear Quadratic Gaussian (LQG) control method with the usage of swarm approaches in control systems, using as basis a paper that uses a mathematical model of an artificial ventilation system. The intent of this article is also to call attention to the good practices in control systems design, sometimes neglected by the researchers who seek to contribute to the field, but lack an appropriate theoretical background. In the presented case study, the classical LQG control shows equivalent result regarding the stability margins and temporal response when compared to a Proportional-Integral-Derivative (PID) controller that is tuned by three different swarm algorithms: Particle Swarm Optimization (PSO), Class Topper Optimization (CTO) and Constricted Class Topper Optimization (C-CTO). When the control signal is evaluated, the LQG controller clearly outperforms the other controllers. Final comments are made regarding some peculiarities of the basis paper and suggesting some orientations to better apply the presented swarm approaches and other nature-inspired optimization techniques in control systems design.