Multi-objective optimization framework for networked predictive controller design

TL;DR

This paper presents an optimization framework for designing networked predictive controllers that maintain stability and performance in the presence of random packet dropouts using LMIs and genetic algorithms.

Contribution

It introduces a novel controller design method combining LMI-based stability guarantees with multi-objective optimization via NSGA-II for networked control systems.

Findings

Ensures closed-loop stability despite packet losses.

Balances conflicting control objectives through multi-objective optimization.

Demonstrates effectiveness via simulation results.

Abstract

Networked Control Systems (NCSs) often suffer from random packet dropouts which deteriorate overall system's stability and performance. To handle the ill effects of random packet losses in feedback control systems, closed over communication network, a state feedback controller with predictive gains has been designed. To achieve improved performance, an optimization based controller design framework has been proposed in this paper with Linear Matrix Inequality (LMI) constraints, to ensure guaranteed stability. Different conflicting objective functions have been optimized with Non-dominated Sorting Genetic Algorithm-II (NSGA-II). The methodology proposed in this paper not only gives guaranteed closed loop stability in the sense of Lyapunov, even in the presence of random packet losses, but also gives an optimization trade-off between two conflicting time domain control objectives.