Real-Time Nonlinear Model Predictive Control for Fast Dynamic System using Improved PSO

Abstract

Nonlinear predictive control has developed to focus on specific nonlinear systems, making generic nonlinear systems more difficult to comprehend. The vast majority are nonlinear and unexpected. It provides an enhanced particle swarm method for nonlinear model predictive control (PSO).This is the content. In each iteration, the fitness assessment function combines random variable search with optimum particle placement to improve PSO convergence. In order to avoid trapping in the local optimum and losing convergence in future searching phases, the current particle variation approach includes a selection step. Nonlinear model predictive control with bound constraints and disturbances is optimised.Nonlinear model predictive control with an improved PSO algorithm for inverted pendulums and other nonlinear devices works effectively. Unlike the traditional PSO approach, the improved method updates the population by fitness function ranking in real-time.