A hybrid global optimization scheme for process design and dynamic optimization

Abstract

Abstract A hybrid global optimization algorithm is developed for the solution of process design and dynamic optimization problems. The proposed algorithm consists of a feasible control strategy, information theory and a chaotic searching algorithm. With the feasible control strategy, points that satisfy the constraints and with minimum objective-function values can be located. After determining the potential candidates, a chaotic algorithm is served to generate new points to find the global optima. The information theory is utilized to escape from the local optima. To extend the proposed global scheme to the solution of the dynamic optimization problems, the orthogonal collocation strategy is used for converting the original dynamic problem into a finite-dimensional optimization problem. This effort leads the proposed global optimization scheme directly applicable to the solution of dynamic optimization problems and makes the solution procedure quite easy. The applicability and effectiveness of the proposed global optimization scheme have been tested with some typical optimization problems, and extensive comparisons with the existing simulated annealing algorithm are performed.