A bi-level transformation based evolutionary algorithm framework for equality constrained optimization

Evolutionary algorithms (EAs) have been widely used by researchers and practitioners to solve optimization problems with constraints. However, equality constrained optimization problems (ECOPs) have posed a great challenge to traditional EA methods due to the dramatically narrowed search space caused by the equality constraints. In this paper, a bi-level transformation based evolutionary algorithm (BiTEA) framework is proposed to transform the ECOP into a bi-level optimization problem. In the BiTEA framework, the original ECOP is solved by an EA as the upper level problem, and the equality constraints are handled by another EA as the lower level problem. To facilitate performance comparison, a set of scalable ECOP test instances with various composable complexities is constructed for experimental studies. The performance of an implementation of the proposed BiTEA on these constructed instances is verified by comparing its performance to that of three state-of-the-art constraints handling EA methods.