分數冪次型灰色生成預測模型誤差分析暨電腦工具箱之研發;The Error Analysis in GM(1,1) Model by using Fractional Power in Grey Generating and the Development of Toolbox

NCUIR > College of Electrical Engineering & Computer Science > Executive Master of Computer Science and Information Engineering > Electronic Thesis & Dissertation > Item 987654321/74633

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/74633

Title:	分數冪次型灰色生成預測模型誤差分析暨電腦工具箱之研發;The Error Analysis in GM(1,1) Model by using Fractional Power in Grey Generating and the Development of Toolbox
Authors:	温惠筑;Wen, Huei-Chu
Contributors:	資訊工程學系在職專班
Keywords:	灰色系統理論;GM(1,1);背景值;誤差;alpha值;分數;GM((1,1);Background value;Error;Fractional order accumulation;Computer toolbox
Date:	2017-07-17
Issue Date:	2017-10-27 14:34:28 (UTC+8)
Publisher:	國立中央大學
Abstract:	在灰色系統理論的GM(1,1)模型，是屬於預測模型，主要是構建在一階微分的模型，而是採取兩個相鄰點的平均值(背景值)所產生的。根據過去的研究，發現背景值是影響預測誤差最重要的因素，因此許多論文都集中在背景值的調整，包括背景值中alpha值的數位跳躍變動，或者背景值中alpha值的連續變動以低預測的誤差，但是在此種形式下背景值，有時仍無法充分顯現離散點所隱含之規律，因此形成GM(1,1)模型的誤差無法大幅降低。因此本論文提出了一種新的方法，內容包含三大要項，首先將背景值以分數的型態表示，並且結合組合數學的方法獲得實際的數值，最為降低誤差的基礎。接著整合GM(1,1)模型，實際分析降低預測誤差的內容，做出分析結果之數學模式。最後本文還自行研發體工具箱，做為減少複雜的計算和驗證最終結果的準確性，不僅擺脫傳統的背景值方法，也打開了GM(1,1)模型降低預測誤差的新方向。;bstract According to the basic of empirical mathematical operation of GM(1,1) model in grey system theory, the background value is used to establish in the first order differential model, and the traditional method is taken the average of the two neighbor points. According to the past research in GM(1,1) model, it can fine that the background value not only is the main influence factor, but also is the most important influence factor in prediction error. Therefore, the paper based on the past research, focused on the adjustment of the background values, present a new approach, which is called the fractional order accumulation method, and combine with the combination mathematics method to get the mathematics result. Besides, the paper also develops computer toolbox to reduce the complex calculation and to verify the accuracy of final results. From the mentioned above, the paper it not only get rid of the default of traditional background value method, but also to open a new approach for GM(1,1) model.
Appears in Collections:	[Executive Master of Computer Science and Information Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	354	View/Open

社群 sharing

Loading...