A Feasibility-Preserving Crossover and Mutation Operator for Constrained Combinatorial Problems

Abstract

Abstract. This paper presents a feasibility-preserving crossover and mutation operator for evolutionary algorithms for constrained combinatorial problems. This novel operator is driven by an adapted Pseudo-Boolean solver that guarantees feasible offspring solutions. Hence, this allows the evolutionary algorithm to focus on the optimization of the objectives instead of searching for feasible solutions. Based on a proposed scalable testsuite, six specific testcases are introduced that allow a sound comparison of the feasibility-preserving operator to known methods. The experimental results show that the introduced approach is superior to common methods and competitive to a recent state-of-the-art decoding technique.

Genetic Algorithm for VRP with Constraints Based on Feasible Insertion

“…The author in (Reid, 2000) discusses the possibility of having feasible solutions generated in crossover and mutation operations. In order to handle feasibility in the genetic algorithm, the creation of infeasible solutions is usually allowed, but they are additionally penalized with a modification of the objective function or correction with an additional repair algorithm (Reid, 2000;Yeniay, 2005;Lukasiewycz et al, 2008aLukasiewycz et al, , 2008b.…”