Efficient Multi-Objective Ant Colony Optimization

Most optimization problems in water resources management require tradeoffs between conflicting objectives. Multi-objective optimization is a growing research area with the aim of finding the Pareto optimal set of solutions which defines the optimal trade-offs. In past studies, the focus has been on developing methods to find Pareto fronts with better diversity and coverage- the issue of objective function evaluation was of secondary importance. However, in water resource applications, objective function evaluations can be computationally very expensive. This leads to our motivation of developing a multi-objective optimization method which not only converges to the optimal Pareto front with improved diversity but also with fewer function evaluations. An efficient multi-objective ant colony optimization method (EMOACO) is proposed and compared against benchmark methods such as NSGA-II, eMOEA and SMPSO. The results demonstrated the capability of EMOACO to converge to the approximate Pareto-optimal with significantly fewer evaluations.