以多目標遺傳演算法求解整合性區位定址庫存控制之供應鏈分銷網路系統

A Multi-objective Evolutionary Optimization Approach for Solving a Capacitated Location-inventory Distribution Network System

供應鏈分銷網路系統提供一種最佳化平台來追求供應鏈需求者的時間效率與供應者的成本效益。本研究整合了區位定址、庫存控制與分銷網路設計的供應鏈規劃三種決策問題，並以兩種相互衝突目標：需求者時間效率與供應者成本效益為追求最佳化的標竿，設計了一個整合性的多目標規劃模式稱為多目標定址庫存問題，簡稱為MOLIP。由於該問題模式為混合非線性數學規劃模式，本研究探索以多目標遺傳演算法中稱為「菁英式非支配排序遺傳演算法」求解MOLIP模式的可行性。為了有效求解此問題，我們以接近實際供應鏈分銷網路問題設計了模擬的問題，包含了15間分銷中心位址與50位潛在顧客，並進行相關的數值分析以驗證求解方法的成效，結果發現，該方法所獲得的答案是令人滿意的。

Supply chain network system provides an optimal platform for efficiency and effectiveness. Supply chain management usually involves multiple and conflicting objectives such as cost, customer service level (fill rate), and flexibility (responsive level). In this paper, a Multi-Objective Location Inventory Problem (MOLIP) model is initially presented. The model formulated includes cost, fill rate and responsive level elements and integrates the effects of facility location, transportation modes, and inventory related issues. MOLIP permits a comprehensive trade-off evaluation for multi-objective optimization. This paper also investigated the possibility of a hybrid GA approach based on the elitist Non-dominated Sorting Genetic Algorithm-Ⅱ (NSGA-Ⅱ) for solving MOLIP. An experimental study using practical data was then illustrated to verify the efficacy of the proposed approach. Computational analysis has revealed a promising solution in solving practical-size problems with 50 buyers and 15 potential DCs, which may be an innovative approach for such kinds of difficult-to-solve problems.