A repair mechanism for active inequality constraint handling

Abstract

Constraint handling is an active field of research in the Genetic Algorithms community, considering that one or more constraints need to be satisfied in most real life optimization problems. Recently, we proposed a Most Probable Point based repair approach for handling equality constraints in Single-objective and Multi-objective optimization problems. In this work, we demonstrate the application of the repair approach to handle active inequality constraints. We show that the repair mechanism, which has so far been strictly applied to the domain of equality constraint handling can be used to obtain better results with faster convergence even in inequality constrained problems. We take up a number of standard Single-objective test problems having one or more active inequality constraints for our study. The applicability of the proposed procedure is demonstrated on a well studied Engineering design optimization problem. The present study contributes to the scarce body of literature available on repair mechanisms in inequality constraint handling and hence should motivate further research in this direction.