隨著資訊科技突破發展，吾人對於車子使用體驗的要求日益漸增，各大車廠也致力於車載資訊系統的服務提升，因應設計的複雜度便有可能衍生系統失效的可能，進而影響商譽。本研究選定目前最具發展趨勢的CPU模組，利用存活分析的非參數型統計手法，解析產品出貨到功能失效問題發生的時間區段，納入可能的解釋變數，建構模型並找出影響因子。
本根據最終研究結果，生產測試數據值對於產品出貨後之不良風險是有相關的，例如記憶體測試項目數值每單位提升可降低約0.36倍的失效風險，而通訊訊號的傳送接收值，也對於失效風險分別有增加或減少的影響，故就測試面來說，可做為產測工程單位的研究與參考，如是否要調整規格加嚴監控此測試項目。重測過的產品其可能衍生的失效風險是相對高的，增加了約0.85倍。測試良率的提升有助於降低出貨後可能的失效風險機率。而就生產日期而言，季度Q1~Q4的存活曲線也不同，Q3在產品失效的控制是較佳的，存活天數平均為270天，是四個季度之冠。汽車的生產，大約在每年八、九月是旺季，大量的規模性生產有助於生產設備趨於穩定的狀態，在較為受控的生產條件下生產，故品質相對穩定、失效率較低。最後在模型型態來說，M Type表現是優於S Type，實務上M產品單價稍高，故這樣的結果是可合理推斷的。最終的調整及改善作法將在最後建議的篇章詳述。

With the significant development of information technology, people are eager to be in pursuit of considerate experience when driving the car, in the meantime, all of the well-known car makers in automotive industry are dedicated to enhancing the service of telematics system. The complex design leads to possible malfunction after markets, posing negative impact on the reputation of the automotive brand. The promising CPU module is then selected in this thesis with analysis of non-parametric statistic survival method. We then construct the predictive model based on the latent significant explanatory variables.
According to the analytical result, the production testing data contributes to the malfunction possibility after the goods are shipped. For example, the memory testing item indicates an 36% decrease in the expected hazard relative to an increase per unit
in the testing value. The transmission and receipt value of communication signal are also effective on the malfunction possibility. The re-tested product is with 1.85 times higher relative expected hazard, and the increase of yield rate leads to lower risk of the failure happening afterwards. With regard to production date, there is difference in survival curve among quarter 1 to quarter 4. There is longest survival period (days) in
quarter 3 with average 270 days. On the other hand, the M type module performs better than S type. Practically the unit price of M type is higher than S type, therefore, the result is reasonable that the M type is with better quality performance. The proposal to improve the quality performance will be further described in the last chapter.

摘要...I
Abstract...II
致謝...III
目錄...IV
圖目錄...V
表目錄...VI
第一章 緒論...1
1.1 研究背景...1
1.2 研究動機...3
1.3 研究目的...5
1.4 論文架構...7
第二章 文獻回顧探討...10
第三章 研究方法...13
第四章 個案分析...19
4.1 資料收集...19
4.2 變數釋義...19
4.3 敘述統計...21
4.4 模型建構及分析...26
第五章 結論與建議...36
5.1結論...36
5.2建議...37
參考文獻...39

1.Pengzhou Xu, Junhui Gao. 2019. Case Study of Four Vehicle Reliability Comparison Based on Survival Analysis. Journal of Transportation Technologies 9, 109-119.

2.Ayelet Eyal, Lior Rokach, Meir Kalech, Ofra Amir, Rahul Chougule, Rajkumar Vaidyanathan, Kallappa Pattada. 2014. Survival Analysis of Automobile Components Using Mutually Exclusive Forests. IEEE Transactions on Systems, Man, and Cybernetics: Systems 44(2), 246-253

3.Laura Attardia, Maurizio Guidab, Gianpaolo Pulcinic. 2004. A mixed-Weibull regression model for the analysis of automotive warranty data. Reliability Engineering and System Safety 87(2), 265-273.

4.Daniel Mauser, Alexander Klaus, Konstantin Holl, Ran Zhang. 2013. GUI Failures of In-Vehicle Infotainment :Analysis, Classification, Challenges, and Capabilities. International Journal on Advances in Software 6(1-2), 142-154.

5.Ayhan Demiriz. 2018. Used Car Pricing and Beyond: A Survival Analysis Framework. 2018 First International Conference on
Artificial Intelligence for Industries (AI4I). IEEE, 65-68.

6.Yan Liu, Chuanyun Fu, Wei Wang. 2021. Modeling duration of overtaking between non-motorized vehicles: A nonparametric survival analysis based approach. PLoS One 16(1), 1-14.

7.Andre Kleyner. 2008. Modelling automotive warranty claims with build-to-sale data uncertainty. International Journal of Reliability and Safety 2(3), 179-189.

8.Md.Mesbahul ALAM. 2009. Reliability Analysis Using Warranty Data Consisting Only of Failure Information. The University of Electro-Communications. PhD thesis.

9.Shaomin Wu. 2012. Warranty data analysis : a review. Quality and Reliability Engineering 28(8), 1-21.

10.Yicheng Shi, Lum Zhang, Pei Liu. 2015. Survival Analysis of Urban Traffic Incident Duration: a Case Study at Shanghai Expressways. Journal of Computers 26(1), 29-39.

11.吳承紘，2014，高速公路事故排除時間影響因子之研究。 國立交通大學
運輸與物流管理研究所碩士論文。

12.沈怡如，2000，高速公路電子收費系統使用族群特性之探討。 國立交通大學運輸科技與管理研究所碩士論文。

13.張新立、葉祖宏，2005，存活分析法應用於機車持有年限之研究。 運輸計劃季刊 40(2)，161-183。

14.何子銘、盧瑜芬、許家瑋，2007，運用三種資料探勘方法預測子宮頸癌存活情形之比較。 台灣家庭醫學雜誌 17(4)，222-238。

15.陳蕙如、邱泓文、廖文進，2013，以類神經網路模式輔助外科重症病人預測死亡機率。 重症醫學雜誌 14(1)，1-7。