Solving constrained optimization problems via multifactorial evolution

As a new paradigm in the field of evolutionary computation, multifactorial evolution has become more and more popular since its inception. It attempts to solve multiple optimization problems simultaneously using a single evolving population. Due to the implicit knowledge transfer, multifactorial evolution exhibits the potential to solve complex optimization problems. This paper tries to take advantage of multifactorial evolution to solve constrained optimization problems (COPs). To this end, we derive two different optimization problems from the considered COP. Theoretical analysis reveals that the optima of these two problems are exactly identical to the feasible optima of the original COP. Thus, the advantages of knowledge transfer can be used adequately. In addition, these two problems focus more on the objective function and the constraints, respectively. By solving them concurrently, we can achieve the balance between constraints and objective function, which is of essential importance in constrained evolutionary optimization. Moreover, a multifactorial differential evolution is developed, which can leverage the merits of multifactorial evolution and differential evolution effectively. To tackle complex COPs, a diversity strategy is designed for population diversity maintenance. Extensive experiments on benchmark test sets and engineering optimization problems have demonstrated the effectiveness of the proposed method.