在真實情況中,眾多商品的利潤常常會隨時間的變動而改變,例如食品、花卉、藥酒等衰退性的產品,由於不定點銷售業者並沒有像一般企業擁有先進的保鮮設備,故此現象會更加明顯。在本問題中,我們加入產品成長率(α)、產品價格倍率 (β)與產品腐敗率(γ)三參數於此具衰退性產品之不定點銷售問題中,而當α=0、β=2、γ=-(無作用)的情況下,此問題成為傳統所探討的越野競賽問題,故此利潤變動型態的具衰退性產品之不定點銷售問題,可視為傳統越野競賽問題的一種廣義模式。 本研究的主要目的是提出一個免疫演算法來解決此具衰退性產品之不定點銷售問題,本研究改良免疫演算法之記憶區雜異度評估方式,擷取合理範圍的抗體片段進行評估,以增進記憶區內各抗體間的雜異度,提高免疫演算法搜尋最佳解的能力。最後,本研究探討之前學者所提出之指標問題,並且測試不同α、β、γ三參數組合而成的1072個測試問題,以數值結果來研究不同的α、β、γ三參數對此具衰退性產品之不定點銷售問題的影響。
In real cases, products will deteriorate and their profits will change over time. For example, foods、flowers、medicines、liquor etc. In this thesis, we consider the deterioration products sales problems in which three factors (α,β,γ) are involved. Note that when α=0、β=2、γ=-(no effect), this problem reduces to the typical orienteering problem. Therefore, the typical orienteering problem is a special case of our new proposed problem. The purpose of this paper is to propose an immune algorithm to solve the proposed deterioration products sales problems. In this study, we develop an improved approach to evaluate the similarity among antibody fragments. Four main benchmark problems are solved by the proposed immune algorithm. In addition, based upon the four main test problems, 1072 sub-test problems with various (α、β、γ) are solved by the proposed immune algorithm as well. Numerical results show the effectivess of the proposed algorithm.