Multi-phase evolutionary algorithm for non-linear programming problems with multiple solutions

Abstract

In this paper a multi-phase evolutionary algorithm (MPEA) for solving general non-linear programming problems (NLP) is proposed. It uses population decomposition, elite multi-parent crossover, better of Gauss and Cauchy mutation and population hill-climbing strategies for adaptive search and particle swarm optimization (PSO). Comparing with other algorithms, it has the following advantages. (1) It can be used for solving non-linear optimization problems with or without constraints, real NLP, integer NLP (including 0-1 NLP) and real-integer mixed NLP. (2) It can be used for solving multi-modal function optimization problems. It means that it can be used to get multiple solutions in one run if the NLP has many global optimal solutions. (3) It is not needed to continuity, convexity and derivative information. In this paper, numerical experiment results show that this evolutionary algorithm is very effective in generality, reliability, precision, robustness and intelligence.