Study On Membrane Computing Method For Multi-objective Nurse Scheduling Problems

Nurse scheduling problem is a kind of combinatorial optimization problem with multiple constraints.The quality and efficiency of the scheduling scheme directly affect the cost and quality of nursing in hospitals.It is of great practical significance and research significance to develop more efficient algorithms and scientifically and reasonably work out scheduling schemes.In order to better solve the multi-level and multi-position nurse scheduling problem with personal preferences,considering nursing needs,labor regulations and personal preferences,the multi-objective nurse scheduling problem model was defined and the multi-objective problem was transformed into a single penalty function by means of weighted average.Aiming at the single penalty function of nurse scheduling problem,a variable neighborhood membrane computing(MCVNS)method is proposed.Based on the idea of variable neighborhood search,a variable neighborhood membrane system is constructed by using the parallel mechanism of membrane computing.Each membrane in the membrane system corresponds to a neighborhood structure and determines the evolution rules of the membrane,i.e.rewriting rules and communication rules.Membranes are optimized in parallel according to evolutionary rules,and neighborhood structures with higher efficiency can get more call opportunities.A series of simulation calculations are carried out for the test problems of five departments.The experimental results verify the effectiveness of the MCVNS algorithm.Aiming at the scheduling problem of many-objective nurses,a decomposition-based multi-objective simulated annealing film computing(MCMOSA)method is proposed.Based on the idea of multi-objective simulated annealing,a multi-objective simulated annealing film system is constructed by using the parallel mechanism of membrane computing.Based on the idea of multi-objective decomposition,the multi-objective nurses scheduling problem is decomposed(grouping and aggregating),which is divided into multiple dual-objective or three-objective problems.Each basic membrane in the membrane system corresponds to a sub-problem after grouping and aggregating.Then,the membranes are optimized in parallel according to the multi-objective simulated annealing algorithm,and the communication rules are used for different sub-problems.Communication and update between grouping problem.The experiment results show that compared with the MOSA algorithm,MCMOSA can get better solution set and has a certain robustness.The experimental results show that the optimal solution obtained by MCVNS algorithm is slightly better than the optimal solution in MCMOSA solution concentration.On average,the optimal solution of MCVNS can only dominate a very small part of the solution in MCMOSA solution concentration,and the diversity of the solution set obtained by MCVNS algorithm is more abundant.