AN UNBIASED BI-OBJECTIVE OPTIMIZATION MODEL AND ALGORITHM FOR CONSTRAINED OPTIMIZATION
Ning Dong, Yuping Wang · International Journal of Pattern Recognition and Artificial Intelligence · 2014
Transforming a constrained optimization problem (COP) into a bi-objective optimization problem (BOP) is an efficient way to solve the COP. However, how to obtain a good balance between the objective function and the constraint violation function is not easy in BOP. To handle this issue, a novel unbiased bi-objective optimization model is proposed, in which both objective functions are equally treated. Furthermore, the novel model is shown to have the unique Pareto optimal vector under proper condition, and the Pareto optimal vector is exactly corresponding to the optimal solution of COP. Moreover, the relationship between the existing biased bi-objective model and the proposed unbiased one is analyzed in detail. For the unbiased model, a generic multi-objective optimization evolutionary algorithm, i.e. a differential evolution (DE), can be used to solve it, and Pareto ranking is employed as the unique selection criterion. The experiments are conducted on 24 well-known benchmark test instances and the results illustrate that the proposed model is not only effective but also efficient.