A new two-stage variable neighborhood search algorithm for the nurse rostering problem

Abstract

The nurse rostering problem refers to the assignment of nurses to daily shifts according to the required demand of each shift and day, with consideration for the operational requirements and nurses’ preferences. This problem is known to be an NP-hard problem, difficult to be solved using the known exact solution methods especially for large size instances. Mostly, this problem is modeled with soft and hard constraint, and the objective is set to minimize the violations for the soft constraints. In this paper, a new two-stage variable neighborhood search algorithm is proposed for solving the nurse rostering problem. The first stage aims at minimizing the violations of the soft constraints with the higher penalty weights in the objective function. While the second stage considers minimizing the total solution penalty taking into account all the soft constraint. The proposed algorithm was tested on the 24 benchmark instances of Curtois and Qu (Technical Report, ASAP Research Group, School of Computer Science, University of Nottingham (2014)). The test results revealed that the proposed algorithm is able to compete with the results of a recent heuristic approach from literature for most of the tested instances.