Artificial bee colony-based predictive control for non-linear systems

Abstract

In this paper, a new approach for the implementation of non-linear predictive control is proposed using fuzzy modelling and the artificial bee colony (ABC) algorithm. The main difficulty relevant to the implementation of non-linear predictive control techniques is obtaining, in real time, accurate solutions to the optimization problem. The aim of this work is to derive a simple and efficient algorithm that can solve the non-linear optimization problem with minimal computational time; this allows the real-time feasibility of the control algorithm to be ensured. Indeed, to deal with the problem of slow and premature convergence of the ABC algorithm, a new enhanced version of this algorithm is proposed. In this version, to improve the convergence speed, the initial population is generated using a chaotic map and a modified update equation is used. Furthermore, to avoid the premature convergence of the ABC algorithm, a new expression for the limit parameter, which allows an increase in the exploratory capabilities of the algorithm, is proposed. The modified ABC algorithm allows accurate solutions for the optimization problem of non-linear predictive control with low computational burden to be obtained. First, a statistical analysis of the convergence of the ABC improved version, using some well-known benchmark functions, is presented and compared with that of other ABC algorithm versions. Then, to assess the efficiency and the performance of the proposed control algorithm, control of a continuous stirred tank rector model is considered. To demonstrate further the effectiveness of the proposed controller, a comparative study, using several meta-heuristic algorithms, is carried out.