A novel multi-objective bionic algorithm based on plant root system growth mechanism

Abstract

This paper proposes and develops a novel multi-objective optimization scheme called MORSGO based on iterative adaptation of plant root growth behaviors. In MORSGO, the basic local and global search operators are designed deliberately based on auxin-regulated tropism of the natural root system, including branching, regrowing of different types of roots. The fast non-dominated sorting approach is employed to get priority of non-dominated solutions obtained during the search process, and the diversity over archived individuals is maintained by using dynamical crowded distance estimation strategy. Accordingly, Pareto-optimal solutions obtained by MORSGO have merits of better diversity and lower computation cost. The proposed MORSGO is evaluated on a set of bio-objective and tri-objective test functions taken from the ZDT benchmarks in terms of two commonly used metrics IGD and SPREAD, and it is compared with NSGA-II and MOEA/D. Test results verify the superiority and effectiveness of the proposed algorithm.