生鮮產品貨架壽命管理暨銷售指派

Abstract

近年來國內外與食品安全有關的議題受到高度關注，且因資訊傳遞能力的提升，使得消費者要求更精準的食品相關資訊，促使了「食品產銷履歷」建置作業的推廣，詳細地記錄生鮮產品(fresh products)在配送、處理、運輸和儲存過程的溫度等相關資訊，而危害分析重要管制點(Hazard Analysis Critical Control Point, HACCP)等食品安全規章也因應而生。為了確保易腐(perishable)生鮮產品之品質，冷鏈與低溫物流管理已逐漸成為當前供應鏈管理的發展趨勢。諸位學者亦提出了造成產品品質變化之關鍵影響因子。有鑑於此，本研究將以倒傳遞類神經網路為基礎，發展一「貨架壽命預測模式」，以準確預測生鮮產品之貨架壽命，並以混合整數規劃(mixed integer programming, MIP)方式構建一「產品銷售指派模式」，針對模擬範例資料，求解在分級制度下的生鮮產品指派(assignment)與分配(allocation)問題。 分析結果顯示，本研究所提出以倒傳遞類神經網路(Back-Propagation Neural Network, BPN)為基礎之「貨架壽命預測模式」中，以「溫度」對品質變化的影響最為顯著，在有足夠的樣本資料時，可做為有效的貨架壽命預測工具。此外，經範例驗證後發現，「產品銷售指派模式」可以系統化、數學化的方法有效地解決產品指派與分配問題。並透過敏感度分析得知，各項成本參數的變動對模式目標值皆有顯著影響，且若能藉由準確預測的產品品質加以分級，並將有限的產品進行有效地指派與分配，可協助企業進行正確的管理決策，大幅提升其整體利潤。

In recent years, the issues about food safety have been highly concerned in global. Because of the advanced capacity of information exchange, consumers ask more accurate information about food and the application of “Food Traceability System” is promoted and developed constantly. The “Food Traceability System” records the temperature and other related information of the fresh products during distribution, processing, transportation and storage. Besides, the Hazard Analysis Critical Control Point (HACCP) and other food safety regulations have been also established due to the need. In the supply chain management, the cold chain logistics management has gradually become the trend in order to maintain the quality of perishable fresh products. Many researchers also proposed the key influencing factors that will cause the change in product quality. This research proposes the following two models. First, a “Shelf Life Prediction Model” based on the back-propagation neural network (BPN) is developed to accurately predict the shelf life of fresh products. Second, a “Products Sale Assignment Model” which is a Mixed Integer Program (MIP) is constructed to assign and allocate the fresh products under the grading system. From the proposed “Shelf Life Prediction Model”, the results show that temperature is the most important variable that has the significant effect on the change of fresh foods quality. If there are sufficient sample data, this model can effectively predict the shelf life. In addition, with the numerical example, the “Products Sale Assignment Model” is a systematic and mathematical approach to effectively solve the problem of product assignment and allocation. And the results of the sensitivity analysis show that the changes in the cost parameters of the model will significantly impact the objective value. From the research findings mentioned above, if the shelf life prediction, quality grading for the products and product assignment and allocation are performed appropriately, managers can make the correct decisions for businesses and increase their overall profits.