Modified hybrid algorithm for process optimization

Abstract

The main weak points in using AI optimization techniques are the possibility of being trapped at local minima, being confined to the population space, difficulty to solve heavily nonlinear problems and to make full use of the historical information beside the lack of prediction about the search space. This paper is concerned with a hybrid optimization technique; namely cultural-based genetic algorithm MCBGA for solving multidimensional and nonlinear applications. The feature of proposed MCBGA technique is enhanced using biased initialization and dynamic parameters setting. Also elitism is carried out. The proposed approach has been carried out on pressure vessel design, fed-batch fermentor and continuous Stirred Tank Reactor CSTR test systems as well as Rastrigin, a well known standard workbench. The proposed MCBGA in terms of the diversity of the optimal solutions are obtained and compared to real coded genetic algorithm as well as other optimization techniques reported in the literature such as binary Genetic Algorithm (GA), Particle Swarm Optimization (PSO) and its variants. The results obtained using the proposed algorithm are more accurate and the fast convergence is obvious. © 2014 Elsevier Inc. All rights reserved.