A Study on an Evaluation Model for Robust Nurse Rostering Based on Heuristics

摘要

Staff scheduling problem has been researched for decades and dozens of approaches have been proposed. Since in the hospital ward, an optimal solution could be changed for the uncertain causes, such as sick leave or other unforeseen events. If these occur, the roster that has been settled as an optimal solution often needs to make changes such as shift moves and others, some of which could have impact on the rosters fitness value. We first investigate the sensitive of an optimal solution under several operations of those types and the result shows that the solutions which are optimal obtained with the searching technique could indeed be affected by those disturbance. Secondly, the evaluation method is used to construct new evaluation function to improve the robustness of a roster. The model could apply to any method such as population-based evolutionary approaches and metaheuristics. Experiments show that it could help generate more robust solutions.