| 摘要: | 本研究利用兩階層數學規劃模型解析台灣地區廢主機回收費率制訂的問題。高階為環保署資源回收基金管理委員會，其目標在於回收基金的收支平衡；而低階為回收處理業者，其目標則是利益最大。由於兩者的目標不一致，透過此模型可反映出其衝突本質，而獲得妥協情況下之最佳解。
由於有限的資源，減量、重複利用及資源回收再利用成為解決環境與資源問題的三個主要方法。減量與重複利用只能暫時解決資源廢棄的問題，因此資源回收成為現今環境保護最重要之方法。民87年，我國環保署成立基管會制訂回收清除處理費率，推動各項應回收一般廢棄物的回收工作，希望藉此提高回收率並降低環境污染。此費率制訂涉及被徵收費率之責任業者及受費率補貼之回收處理業者，同時也包涵基管會的運作。本研究簡化此一關係後，以兩階層數學規劃模型表達。高階目標為平衡基金預算，低階則靠著努力提高回收率，來獲取較多的補貼費用以及資源化價值，以期獲得最大利潤。雙方目標不同，但決策又彼此互相影響雙方的目標函數，因而得以建立廢資訊物品回收費率制定之模型。
本模型為兩階層非線性規劃問題較為複雜，為簡化求解過程以及不增加問題的複雜度，利用Shih et.al (1996)所提出之模糊趨近法 (Fuzzy Approach) 轉換為單階非線性數學規劃模型，再以一般數學規劃軟體與基因演算法分別求解。分析結果發現，基因演算法較易跳出局部最佳解，與穩定求解的傳統解算法有各自的優劣。而回收處理業者之回收意願受到補貼費率及回收處理成本之影響。因此資源回收政策的推動，訂定最適的回收清除處理補貼費率舉足輕重。接著對回收清除處理補貼費率進行參數分析，結果顯示回收清除處理補貼費率與回收率朝同向移動，在達到最適點之後成反向變動。最後再將本研究所建立之兩階層規劃模型與現行之費率計算公式進行比較，發現本研究之模型係按兩階層之決策目標及限制建立，較符合實務上費率之互動情況，可供實務上進行費率制訂作業時之參考。
This study tries to make a recycling subsidy decision for recycling and treatment waste desktops in Taiwan through a Multi-level programming problem (MLPP). The upper-level decision unit is Recycling Fund Management Board (RFMB), Environmental Protection Administration of ROC Government (Taiwan) and the lower-level’s is the recycling industries. The former objective is to balance the recycling fund, and the latter’s is to maximize profits for recycling operations. Because both objectives are conflict, the MLPP will manage the interactive behavior and be solved with a satisfactory solution.
Because earth resources are over consumed, it’s important to save the resources and to reuse them. Reducing, reuse and recycling (3R) are three main actions to deal with the problem. Nowadays, recycling is the most important one to sustainable management. In 1998, RFMB has been established for controlling waste materials by setting up the recycling and treatment fee to recycling industries for increasing recycling ratio and decreasing environmental pollution. The funds are collected from manufacturers and importers for their responsibility. There are three stakeholders in the fee-setting game: recycling industries, manufacturers, and RFMB itself. We simplify the problem as a bi-level programming model. Since the model is a nonlinear MLPP and difficult to solve, we transform the model to a single-level nonlinear programming problem through fuzzy approach developed by Shih et al., (1996). The auxiliary model can be solved by traditional algorithm and genetic algorithm (GA), respectively. Because GA search process has an advantage to avoid local optimal, the global optimal solution are commonly obtained.
The results show that the subsidy fee and treatment cost affect recycling industries to recycle the wastes. Hence, to push the recycling policy is very important to set optimal treatment fee. After making a sensitive analysis, the subsidy fee and recycling ratio are up or down at the same time. And they have a different direction variation after optimal solution. Finally, we compare the model and formulation we used to estimate the fee now. The model set by this study is actually lucid to simulate the interactive behavior, because it set from decision objective and constrain of a bi-level programming. Hence, the model can be regarded as a useful tool for tariff setting on recycling desktop in Taiwan in the future. |
