| 摘要: | 隨著現在技術迅速創新和全球化的時代，公司和客戶在市場上帶來巨大的機會與選擇，但也面臨勞動力逐漸減少、物料成本持續上漲、產品生命週期縮短，以及各種客製化少量多樣服務的需求變化。這些原因導致市場變動快速，產業間相關企業的競爭轉變成供應鏈與供應鏈間的競爭，為穩定且提高供應鏈運作效率對整個供應鏈體系的相關企業而言，是保持競爭力的重要條件之一。要讓工廠變為智慧製造業，就需藉由人工智慧的幫助，讓管理與生產更加有彈性，創造新一代製造革命新工程。
當今AI人工智慧的發展已經快速且全面性的滲入製造業與非製造業，企業除了可以讓流程自動化更加有效率，提升對顧客服務的品質與一致性外，人工智慧更改變了原本管理方式。此外，AI人工智慧已經應用於供應鏈管理的各個環節，讓企業在營運上變得更迅速、更聰明、更精簡。本研究旨在探究企業如何在這股浪潮下，將人工智慧系統應用於供應鏈管理系統中，如何有效率地整合與管理供應商、製造商，並保持供應鏈中各階段之生產量、存貨量及倉儲配送等階段於穩定之狀態，以面對下游客戶的各種需求變化，進而提升企業評估供應商風險之依據。
這些外在科技所帶來的變化，將採購人員從一個「執行者」轉變為一個「管理者」的角色，未來執行工作的將會是眾多形式的機器人，而人員則將扮演一個運用科技，規劃並整合系統的管理者。科技發展日新月異，採購人員是否能因應外在環境的變化趨勢，迎接AI時代的來臨。
在交易市場中，原物料價格與產業新聞確實互相影響，當出現正面新聞時，價格通常會高漲，相反的，若出現負面新聞時，價格則呈現下跌狀態。本研究針對供應商必需使用到的原物料進行價格與新聞的預測，透過強化學習方式，在與環境做互動的同時，由模型自行學習市場交易的買賣情況，再藉由正評或負評的產業新聞來重新驅動模型的行動。
本研究也提出一個新型態的強化學習演算法，藉由新聞的情感來重新執行模型的買賣行為。藉由模型來預測市場趨勢，使專業的採購人員在進行採購作業時，能夠避免採購風險。希望藉由本研究確實能有效顯示出供應商的潛在高風險值，進而提供下單前評估供應商有無缺料或財務風險之依據。;With the trend of economic globalization and the innovative technology, corporations and customers can obtain tremendous opportunities and selections in the market. However, they are also facing some problems, such as a lack of labor force, rising cost of materials, shorter product life cycle, and the needs of mass customized products. Therefore, stabilizing and improving the operational efficiency of the supply chain is one of the most important conditions for these enterprises to maintain competitiveness in the overall supply chain ecosystem. To make the factories become smart-manufacturing industries, we need Artificial Intelligence (AI) to achieve more flexible management and production, and even to create a new industrial revolution. In addition to achieving the efficiency of process automation and improving the quality and consistency of customer service, AI can also change the original management mode. Furthermore, AI can apply to all aspects of supply chain management (SCM) and can make the operation of enterprises faster, smarter, and more streamlined. The purpose of this research is to explore that how the enterprises can utilize AI in the SCM to efficiently integrate and manage suppliers and manufacturers.
In the trading market, the crude oil price and the industry news may affect each other. When the positive news appears, the oil price usually rises. On the contrary, the price decreases when the negative news occurs. In this research, we propose a novel Reinforcement Learning (RL) algorithm, which utilizes a multi-learning network to predict oil prices and classify emotional news, and then fuses two networks into the RL model to investigate market trends. While interacting with the environment, the RL model can learn the market transactions by itself and employs positive or negative industry news to react to the buying, selling, or holding trading behaviors. By using this model to predict market trends, the professional purchasers can avoid purchasing risks when making purchasing operations. So, this research hopes to effectively present the potential supply risks and provide a fundamental of evaluating the suppliers’ shortage or financial risks before ordering. |
