運用簡群演算法與層級分析法解決國軍野戰後勤設施選址問題

本研究提出的國軍野戰後勤設施選址問題（Military Logistic Depot Location Problem, MLDLP）是在解決國軍戰時補給系統的兩階層設施選址問題。與過去設施選址問題最大差異在於，MLDLP的目標函數是最大化被徵用建築物的平均效用，而建築物的效用取決於其固有的數個屬性。為此，本研究提出一個整數規劃模型，同時提出兩階段的方法求解。第一階段，運用層級分析法計算屬性的相對權重作為目標函數的係數。第二階段，應用簡群演算法，對整數規劃模型進行求解。兩階段的求解方法透過18組隨機生成的問題進行實證分析，並將計算結果與基因演算法及粒群演算法進行比較。

關鍵字

軍事後勤；設施選址；簡群演算法；效用

並列摘要

This work proposed a military logistic depot location problem (MLDLP) to deal with a two-level facility location problem in the military logistic system. Unlike most previous researches, the objective of MLDLP is to maximize the average utility of requisitioned buildings, and the utility of a selected building depends on its inherent attributes. This work proposed an integer programming model and introduced a two-stage method to deal with this problem. In the first stage, AHP is applied to estimate the relative weights of the attributes as the coefficients of the objective function. In the second stage, simplified swarm optimization (SSO) is adopted to solve the integer programming. To empirically verify the performance of SSO, experiments are conducted using eighteen randomly generated problems and the corresponding results are compared with genetic algorithm and particle swarm optimization.