HARMONY SEARCH WITH NOVEL SELECTION METHODS IN MEMORY CONSIDERATION FOR NURSE ROSTERING PROBLEM

The selection methods of population-based metaheuristics provide the driving force to generate good solutions. These selection methods select the individuals with a higher fitness to be members of the population in the next iteration correspond to the natural rule of Darwin's principle survival-of-the-fittest. Harmony search algorithm is a population-based metaheuristic, which mimicking the musical improvisation process where a group of musicians play the pitches of their musical instruments seeking for a pleasing harmony. It improvises the new harmony based on three rules: memory consideration, random consideration, and pitch adjustment. In this paper, we investigate the replacement of the original random selection of memory consideration with a set of selection methods in order to speed-up the convergence. These selection methods include tournament, proportional, and liner rank of Genetic Algorithm, and Global-best of Particle Swarm Optimization. The proposed harmony search with the different memory consideration selection methods evaluated using standard dataset published in the first International Nurse Rostering Competition INRC2010. Nurse rostering problem is a combinatorial optimization problem tackled by assigning a set of nurses with different skills to a set of shifts over predefined scheduling period. Experimentally, the tournament memory consideration selection method achieved the best rate of convergence as well as the best results in comparison with the other memory consideration selection methods.