Abstract: Considering that the issue of nurses scheduling problem involves the satisfaction of the nurse, this paper joins nurses’ preferences and fair constraints into the nurses scheduling, to find the best schedule to increase the satisfaction of nurses. According to the characteristics of multi-objective problem, this paper uses multi-objective particle swarm optimization algorithm. In the hard constraints, a night shift can only be followed by a day off or a night shift to ensure enough rest for nurses before the next working shift. In the algorithm design, this paper adds mutation operator to expand the particle swarm search space. Due to the existence of some contradictory between the optimization objectives, using multi-objective decision theory can be more scientific and objective optimization of nurses scheduling table. In the final case analysis, it is found that different preferences of nurses will produce different non-inferior solutions. Therefore, in practical scheduling, the nurses’ preferences must be taken into full consideration in order to find a more scientific and reasonable schedule.

Key words: nurses scheduling, multi-target, particle swarm algorithm, soft constraint condition, hard constraint condition, variation operator

摘要： 针对护士排班问题涉及护士满意度的特点，在护士排班过程中加入护士偏好和公平的约束，寻求最优的排班表以增加护士的满意度。根据多目标问题的特点，采用粒子群多目标优化算法。在硬约束条件上，加入N班之后不能上A班和P班的约束，使护士在上N班之后能够得到足够的休息。在算法设计上，加入变异算子，扩大了粒子群的搜索空间。由于各优化目标之间存在一定的矛盾，用多目标决策理论可以更加科学客观地优化护士排班表。在最后的案例分析中，发现护士不同的偏好会产生不同的非劣解，因此在实际排班中，要充分考虑护士的偏好，以求出更加科学合理的排班表。

关键词: 护士排班, 多目标, 粒子群算法, 软约束条件, 硬约束条件, 变异算子