Global Journal of Computer Sciences: Theory and Research

Yazarlar: Corina Rotar

Konular:-

Anahtar Kelimeler:Multi-objective optimization,Preference condition,Controlled mutation,Evolutionary algorithm.

Özet: The paper presents a novel multi-objective optimization evolutionary algorithm. The main idea is designed an intuitive evolutionary technique which substitutes complicated procedures for keeping population diversity and speeding convergence to the Pareto front. Controlled Mutation Evolutionary Algorithms (CMEA) is enhanced with a procedure that estimates the closeness to the target and the crowdedness in the vicinity of each potential descends by using several reference parents. In this manner, Controlled Mutation EA does not need extra parameter for estimating the crowding and encourages surviving those individuals, which are closer to the Pareto optimal solutions.  Our algorithm does not use the concept of Pareto dominance to determine the performance of a candidate; instead, Preference notion is defined.  The proposed algorithm is validated using several standard test problems and popular performance indicators. The experiments show that the proposed approach is competitive and can be considered a viable alternative to solve multi-objective optimization problems.

---