標題: 運用類神經網路於石化產業之產業關連資訊系統之建立
Using Neural Network in the Design of Petrochemical Industry Relationship Information System
作者: 柳均杰
Liu, Jun-Jie
陳安斌
Chen, An-Pin
資訊管理研究所
關鍵字: 石化產業;類神經網路;半結構資訊;petrochemical industry;neural network;semi-structural information
公開日期: 1994
摘要: 本論文的研究主要目的是針對產業關連的特性來發展一套投資分析之資訊 系統。而系統中所必須具備重要功能係針對石化產業中上市公司的石化產 品做其價格波動的行為掌控,以及石化產業的上下游關連性之分析。而為 了達到此目的,本研究包括兩大部分:資訊系統的建立與類神經網路模式 的構建。資訊系統最外層的使用者介面下涵蓋了三個模組:產品基本資料 、圖形分析以及產業關連分析模組。而由於我們所能獲得的上下游之間的 關連產業資訊是屬於所謂的不充分資訊或半結構資訊。為了彌補這種情形 ,我們並應用類神經網路來作為橋接的模組。以期能充分掌握產業景氣的 脈動及其一貫性。最後,本研究將提出以類神經網路橋接的管理資訊系統 作為本論文的一個實證與分析。 The purpose of this study is tried to explorate the characteristics of the linkage of products in of industry and further more to develop a petrochemical industry relationship based investment information system .The system is developed to handle the fluctuation of the prices of petrochemicals and to analyze the relationship between the upstream and downstream of petrochemical industry. In order to achieve this goal two main portions of this system are created in the research:the basic date preprocess information system and a neural network model. Firstly, three modules under the basic data process information system shell are built : raw data collection, graphical analysis and industry relationship analysis. The products relationship between upstream and downstream obtained from the petrochemicals is usually not sufficient enough or is so-call semi-structural information.Secondly,a neural network model which is used to connect in the our information system is created to link or transform the influence of the trend of activities of industry. Finally,a neural network based petrochemical industry information system is built .And an empirical study regarding to the SM,AN and SBR product series is selected and analyzed in the system .The results is shown how the new model can transform the influence of the petrochemical products price,quantity and trend from upstream to downstream and its corresponding statistical analysis.
URI: http://140.113.39.130/cdrfb3/record/nctu/#NT830396018
http://hdl.handle.net/11536/59125
Appears in Collections:Thesis