Comparison of the performance of modern heuristics for combinatorial optimization on real data

Abstract

Until recently most heuristics for combinatorial optimization problems could be grouped into a few classes. However, since the late 1980s a variety of new approaches for such problems have appeared in the literature. Some of these methods, namely simulated annealing, genetic algorithms, tabu search, the Great Deluge algorithm and the Record-to-Record Travel algorithm, are compared in this paper. These approaches represent not only new algorithms for specific problems, but also new ‘meta’ algorithms in the sense that they introduce totally new perspectives on the solution of combinatorial optimization problems. This paper will compare the performance of implementations of these approaches for the problem of balancing hydraulic turbine runners. Real data will be used for the comparison.