應用混合式差分進化演算法在多階流線式生產可插單之排程問題探討

本研究主要探討多階段流線式生產之派工問題，研究採用差分進化演算法(Differential Evolution, DE)與模擬退火法(Simulated Annealing, SA)結合而成的混合式差分進化演算法(SA-base hybrid DE，SAHDE )建構數學模式。模式建構完成後利用plant simulation軟體進行印證，並將其模擬結果與傳統派工法則(SPT、EDD)做比較，結果顯現在其機制的配合調度下，可有效降低總完工時間與總延遲天數，根據演化代數與訂單數的差異有著不同改善程度，但整體而言將可使工件派遣有更好的選擇依據。為使本研究更符合實務需求，若有生產效益較佳之訂單出現將允許插單。

關鍵字

差分進化演算法；緊急插單；派工法則

並列摘要

This research mainly discusses the rescheduling of multi-stage streamlined production under allowing rush orders. The research applies the differential evolution algorithm (DE) and the Simulated Annealing (SA) based hybrid differential evolution algorithm (SA-base hybrid DE, SAHDE) to construct a mathematical model. After the model is constructed, it is verified by plant simulation software for a simplified manufacturing system. The simulation results are compared with traditional dispatching rules such as the SPT and EDD rules to prove its effectiveness. Cooperating with proper scheduling, it can effectively reduce the total makespan and the total delay days. According to the difference between the evolutionary algebra and the number of orders, there are different degrees of improvement, but overall, it will enable a better basis for selection of workpiece dispatch. In order to make this research more in line with practical needs, if there are orders with better production efficiency, rush orders will be allowed.