一种高维多目标血液透析服务患者调度进化算法

A many objective patient scheduling evolutionary algorithm in hemodialysis service

血液透析的患者调度问题具有高维多目标特征。从现实背景出发,结合问题特征,构造一种基于聚类的高维多目标进化算法。其中,个体编码采用多层次结构分别表征患者调度方案的属性,基于此问题的多目标特征设计交配池构造方法,并开发多类交叉和变异算子。基于聚类思想设计3种选择操作方法。通过现实案例扩展构造不同规模的计算案例,集中开展计算实验,先进行快速非支配排序,再通过聚类和个体支配区域大小进行种群选择,结果表明,该算法具有较好的优化效果。

The patient scheduling problem involved in hemodialysis has the characteristics of many-objective.In this paper,a many-objective patient scheduling problem was presented and analyzed.A clustering-based many-objective evolutionary algorithm was proposed,where an encoding schema was set for multi-level structure of a patient schedule,and a mechanism was designed for constructing a mating pool,following with some mutation and crossover operators.In the selection process,three alternative straegies were developed based on clustering.The numerical experiments were conducted with some instances extended with real-life case.The results showed that the algorithm could obtain better results,when the selection strategy was set as fast non-dominated sorting firstly,and clustering-based selecting with the factor of dominance area secondly.