Fast mesh-sorting in multi-objective optimization

A single parameter based fast mesh-sorting is proposed in this work. The single parameter algorithm as compared to non-dominated sorting eliminates the classification of the population into non-dominated fronts and calculating crowding distance. The proposed one parameter approach also provides flexibility of choosing any probability based selection operator. On the other hand, non-dominated sorting approach can only use tournament selection directly. We have considered Zitzler-Deb-Thieles (ZDT) test functions to test computational and convergence capabilities of proposed algorithm. The performance of the proposed algorithm is compared with conventional Non-dominated Sorting Genetic Algorithm-II (NSGA-II) and NSGA-II with a recent fast corner-sort algorithm. We have also considered optimal control of fed-batch reactor as Multi-objective optimization application.