「圖書館書籍通閱移送之車輛途程問題」(LVRP-DP)係車輛於圖書總出發至各間圖書分館進行同時收送書籍之服務，各圖書館間彼此為供給點及需求點，而車輛所運送之書籍具有其特定的起點與迄點關係，求解問題的過程中必須同時處理「車輛路線規劃」之車流問題以及「書籍起迄指派」之書流問題，比單純之車輛途程問題更為複雜，因此屬於運算難度極高之組合數學問題，在過去研究中，最常採用的求解演算法為巨集啟發式演算法。基因演算法 ( Genetic Algorithm, GA) 其基本概念源自達爾文進化論所提出之「物競天擇、適者生存」，所發展而成的巨集啟發式演算法；蜂群最佳化演算法(Bee Colony Optimization, BCO)為根據蜜蜂採集花蜜之行為產生靈感而發展而成的巨集啟發式演算法。因此，本研究分別發展混合式基因演算法及混合式蜂群最佳化演算法，應用於求解LVRP-DP。最後，進行舊金山圖書館系統資料測試並與過去相關研究背景之文獻進行比較，另外，針對台北市立圖書館系統資料進行測試規劃。根據結果發現，本研究針對圖書館書籍通閱移送之車輛途程問題所發展之數學模型與求解演算法考慮層面比國內外現行之圖書館系統之運作方式更為周詳且具有彈性。 The library vehicle routing problem with delivery and pickup (LVRP-PD) is a problem of finding optimal routes to transport origin-destination paired books in a library system comprising a main library and several library branches. The LVRP-PD is an extension of the traditional vehicle routing problem but is more difficult to solve because books are associated with fixed origin-destination pairs. To solve the LVRP-PD, two meta-heuristics called hybrid genetic algorithm (HGA) and hybrid bee colony optimization algorithm (HBCOA) are proposed. Two real library systems, one in San Francisco and the other in Taipei, are then demonstrated with the two meta-heuristics. The library vehicle routes scheduled by HGA and HBCOA are superior to the existing manual operations and those appeared in the literature in terms of some performance indices. The experiments also show that HGA is a bit better than HBCOA but the superiority is not significant. Hence both proposed solution algorithms have equally good potential for real applications in the future.
