GMDH於產品壽命預估之應用分析－以發光二極體為例

GMDH for Empirical Reliability Prediction on Light Emitting Diode

曾明德(Ming-De Tseng)

指導教授 ： 陳雲岫

元智大學/工程學院/工業工程與管理學系/碩士(2000年)

摘要

近年來，由於科技的進步與市場激烈的競爭，各類產品的性能不斷提高，構造也漸趨複雜，廠商間的競爭也愈來愈激烈，產品壽命的品質要求日漸重要。高可靠度是產品所應具有的重要特性之一，而壽命試驗方法是研究可靠度試驗方法之一，本研究將針對步進應力衰減試驗方法進行探討。步進應力衰減試驗是利用產品特性隨時間而劣化的關係，於步進應力下進行壽命試驗，在短期間內獲得失效的試驗結果，然後根據已知或假設的劣化關係，推定物品在實際正常使用狀況下的壽命及其可靠度。本論文一以GMDH（Group Method of Data Handling）為基礎，建構於一步進應力衰退試驗，並進行壽命預估。以模擬發光二極體的步進應力衰減資料為例，說明系統的可行性與驗證GMDH演算法的預測的能力，從實驗得知，GMDH演算法預測正常使用應力下的壽命有不錯預測能力（預測壽命值在產品理論值壽命失效分佈內），且在正常應力下的預測衰退路徑與產品理論值衰退路徑相符，驗證GMDH演算法的確可用來分析步進應力衰減試驗。

關鍵字

GMDH ； Group Method of Data Handling ； 演算法 ； 加速衰減試驗 ； 步進應力試驗 ； 步進應力衰退試驗

並列摘要

Owing to speedy advancement of the technique and completition of the market, many multi-function and high quality products are made to meet people’s needs. High reliability and long life are characteristics to show the high quality of the products. The step-stress degradation test is one of the life tests used to estimate the life time of a product. A step-stress degradation test is using the characteristic of a product’s degrade relationship by the time and gets the testing results in a short time. And, based on the known or assumed degradation relationships we can predict the life time of the product and reliability in a normal state. In this research ,the group method of data handling（GMDH） algorithm is implemented to predict the life time of light emitting diode in a step-stress degradation test. A monte carlo simulation is applied to generate data of the light emitting diode in a step-stress degradation test. Simulation results show that GMDH performs well at the normal stress and that has almost the same predicting degradation path as theoretic path at the normal stress.

並列關鍵字

GMDH ； Group Method of Data Handling ； Algorithm ； Accelerated degradation test ； Step-stress life test ； Step-stress degradation test

參考文獻

8. Chang ,Dong Shang, Chiu ,Chih-Chou, Jiang ,Shwu-Tzy” Modeling and Reliability Prediction for the Step-Stress Degradation Measurements Using Neural Networks Methodology”, appearing in the International Journal of Reliability, Quality, and Safety Engineering, Vol.6, No. 3 pp.277-pp288, 1999

9. Chang , D. S.（1993）, “ Analysis of Accelerated Degradation Data In a Two-way Design ,”Reliability Engineering and System Safety , Vol. 39 , PP. 65-69

11. Horng , Jyh-Jen ,Shiau, and Hsin-Hua Lin（1999）“ Analyzing Accelerated Degradation Data By Nonparametric Regression”, IEEE Transactions on Reliability , VOL48 , NO2 , 1999 JUNE, PP. 149-158

12. Ikeda , S., Ochiai , M. , and Sawaragi , Y.（1976）.“ Sequential GMDH Algorithm and Its Application to River Flow Prediction ,”IEEE trans. Syst. Man Cybern. SMC-6（7）：473-479.

17. Mann , N. R. , Schafer , R. E. and Singpurwalla , N. D.（1974）“ Methods for Statistical Analysis of Reliability and Life Data ”, New York：John Wiley＆Sons,Inc.