[[alternative]]An Eugenic Genetic Algorithm for the Nurse Scheduling Problem

[[abstract]] 跟診人員排班目前均由專員進行手動排班，由於排班時須考量到人員專長和值班門診專長媒合度、人員休假狀況以及醫院所提出之排班限制，故排班對於專員來說是一件相當耗時費力的事情。本研究參考護理人員排班常用之基因演算法依據跟診人員排班規則提出具優生基因演算法概念之兩階段交配和篩選優生染色體步驟。本研究首先進行問題描述以建立演算法求解模型並研究基因演算法各式交配法與修改PMX交配法，接著將單一交配法改為兩階段交配法並於演算法過程中加入篩選優生色體步驟，最後將提出之優生基因演算法實際應用於個案跟診人員排班問題上，進行演算法程式實作演練。經實驗後得知本研究所提出之兩階段交配與篩選優生染色體求解效果確實比傳統交配法好，且本研究所提出之方法求得最佳解班表確實可達到醫院所提出之需求。[[abstract]] The purpose of this study is to propose a genetic algorithm (GA) method for a local hospital's clinical room nurse scheduling problem. The major challenge of this clinical room NSP is to assign the nurses to the clinical rooms that match their clinical specialty. As the number of clinical room nurse is high, it is very time consuming to manually schedule the work shift. Therefore, this study aims to develop a GA to solve the clinical room nurse scheduling problem. In order to derive a higher fitness solution, this study adapts the concept of two stages mating GA. Two different crossover methods are sued in the first and second stage respectively to diver the chromosome population. The eugenic concept is used in the selection of offspring so that better fitness solution could be obtained faster. Ten trail run were conducted with 200 iterations for the proposed methods and traditional GA methods. The experimental results show the two-stage mating with the eugenics provides better fitness solution than the traditional method with smaller standard variation