MFO Approach based Nonlinear MPC Scheme

Abstract

In this paper, a moth flame optimization algorithm is used to tackle the non-convex optimization problem arising in nonlinear model predictive control. The goal is to build a simple and effective control algorithm having goods performances in trajectory tracking with less over shoots and small root mean square error. In this control scheme, a multilayer feed forward neural network is chosen as nonlinear dynamic model for prediction. To demonstrate the validation and effectiveness of the proposed approach, the control of a continuous stirred tank reactor is considered using the proposed method and two well-known methods: the simulated annealing and genetic algorithm. Simulation results reveal that the proposed approach provides satisfactory performance in terms of overshoot and tracking error value.