運用類神經網路於台灣紡織業聚酯纖維之關連分析資訊系統

Abstract

紡織業在國內是一個完整且龐大的生產體系。其上下游之間的關係非常緊 密，任何上游產量抑或是價格的變動，都會引起下游產業的連鎖反應。基 於此點，本研究嘗試提出一個具有類神經網路為紡織業上下游資訊作橋接 的管理資訊系統。為此本研究將對紡織業中以聚酯纖維系列為例，建立其 類神經網路為基礎的資訊系統，並對其進行分析。在本研究中，首先將建 立紡織中相關原料或產品及其相對應公司的基本資料庫，其次是以類神經 網路為橋接的構建。最後，則是根據巳學習好的網路模型，分析當上游產 品某一項目發生變動時，對整個下游產業的影響。並將此影響以圖表方式 顥示，以作參考及對產業影響的分析依據。 Textile industry in Taiwan is not only a huge but also completed production system. The relationship between the upstream and downstream of the production chain is very tied . Based on the above fact , the inference of the change such as the selling price of the upstream company products will bring interlocked reaction to downstream production chain. Based on the reasons , the information management system with neural network as the bridge between upstream and downstream system of textile industry is proposed in this study , and polyester fiber series in textile industry is selected as an example to build the neural network based information management system . In this system , the basic database regarding to material , products and companies of the polyester fiber industry will be built as the preprocess data and prepared for the following neural network system . Then , the bridge of the neural network is constructed and a prototype neural network based information management system is built . Besides , the results of the system performance is evaluated by a simple statistical analysis .