航電產品動態學習效果之推估－伸縮最小平方法之應用

The Dynamic Learning Effects of Avionics Equipment Manufacturing-An Application of Flexible Least Squares Approach

林灼榮(Juw-Rong Lin) ； 徐啟升(Chi-Sheng Hsu) ； 黃泰得(Tai-Te Huang)

《東海管理評論》 7卷1期 (2005/07) Pp. 255-283

https://doi.org/10.29724/TMR.200507.0011

摘要

本文旨在利用國內某航太業者在民國82年年至民國89年，生產四項空用電子產品之資料，推估並解析航電產品之動態學習效果。我們首先以普通最小平方法(Ordinary Least Squares, OLS)及伸縮最小平方法(Flexible Least Squares, FLS)估計學習曲線，並比較推估結果之優劣。隨後依據FLS方法所估計之動態係數向量，檢定產品設計修改、製程精進及組織變動前後之學習率，是否存在顯著差異；實證結果發現：(1)伸縮最小平方法非但可降低殘差項序列相關程度，同時更能掌握學習效果之動態調整歷程，提供較佳的資料適配結果。(2)四種航電產品存在顯著學習效果，但皆拒絕「80%曲線」之虛無假設；亦即「80%曲線」不能一體適用在飛機相關產業。(3)以名目成本估計學習率未必會高估學習效果，端視總體環境及個案生產特性而定。(4)學習率於產品設計修改、製程精進與組織變動前後存在顯著差異，即學習率並非固定常數；顯示應用學習曲線概念時，應將動態因素納入考慮。

關鍵字

航電產品 ； 學習曲線 ； 伸縮最小平方法 ； 80%曲線

並列摘要

The purpose of this study is to evaluate the learning curve and its dynamic aspect of a local avionics-manufacturing firm, which was formerly a non-profit organization run by the military. Four avionics equipment of last 8 years production data was fitted to the Wright's Equation that was reported in 1936. Both ”Ordinary Least Squares” (OLS) and ”Flexible Least Squares” (FLS) approach are used to investigate the feasibility of model fitting to this specific case. The FLS approach is proved to be able to track the dynamics of production, and to check whether or not the OLS solution provides a good description of the production data. The FLS estimates are used to test the difference of learning rate among 3 consecutive production periods, which is categorized by design change and process improvement. The FLS estimates are also used to check the learning rate movement after the firm changed into a government-owned, profit-seeking enterprise. The study shows: (1) The FLS approach provides a good description of the observation that was dynamic in its nature, and the OLS solution is unlikely to indicate the system movement of the learning rate. (2) The four avionics equipment manufacturing exhibit the learning effect, but statistics evidence shows no conformity to the ”80% curve”. (3) Integrating the price-level index to reflect the real cost in the learning curve model does not necessarily provide enough information for learning curve estimation. Further factors need to be considered in implementing the learning curve concept. (4) Evidence shows the learning rate is not necessarily constant in the course of production. Dynamic factors should be considered when estimating the learning curve.

並列關鍵字

Avionics Product ； Learning Curve ； Flexible Least Squares ； 80% Curve

參考文獻

王佳鎮(2000)。學習曲線在通訊產業之應用－個案研究。國立交通大學工業工程與管理研究所。

吳文智(2001)。學習曲線在TFT-LCD產業之應用研究。國立交通大學工業工程與管理研究所。

陳冠羽(1998)。IC製造業應用學習曲線於成本模式之研究。國立交通大學工業工程與管理研究所。

簡立賢(2001)。美國穀物價格風險估測之研究－伸縮最小平方法之應用。農產運銷論叢。6，263-277。

Alchian, T.K.(1963).Reliability of Progress Curves in Airframe Production.Econometrica.31(4),679-693.