Full metadata record
DC FieldValueLanguage
dc.contributor.author趙品智en_US
dc.contributor.authorZhao,Pin-Zhien_US
dc.contributor.author洪志真en_US
dc.contributor.authorHorng,Jyh-Jenen_US
dc.date.accessioned2014-12-12T02:33:52Z-
dc.date.available2014-12-12T02:33:52Z-
dc.date.issued2012en_US
dc.identifier.urihttp://140.113.39.130/cdrfb3/record/nctu/#GT070052618en_US
dc.identifier.urihttp://hdl.handle.net/11536/71991-
dc.description.abstract衰變資料的分析,在許多文獻上已經有廣泛的討論,大部分的衰變模型都 是採用參數模型,在給定之函數形式下經由估計參數來建立模型;本文的作法 與此不同,對於衰變模型的建立,沒有函數形式的假設,而是利用無母數迴歸/ 函數資料分析的方法去估計衰變資料之平均值函數與共變異函數。無母數的建 模方式較具彈性,讓方法可以應用在更廣泛的用途上。由資料所建之衰變模型 可以用來預測系統的故障時間:利用計算條件機率的方式,在給定系統至目前 為止之衰變路徑下,基於已建立之衰變模型我們可以推導出系統的殘餘壽命分 配,並利用此分配的中位數估計系統的剩餘壽命。我們利用模擬比較此估計量 與實際壽命的差異,結果顯示,已知的路徑越長,餘命估計值越準確。餘命的 預測可以幫助我們排定維修的時程,本文提出兩種不同的維修策略,並在同樣 時間內比較兩種方法的維修次數跟更換次數。因維修與更換成本不相同,我們 在成本的考量下,建議此二策略何時為佳,並與一般固定維修時程和只更換完 全不維修兩種策略相比較。zh_TW
dc.description.abstractThe study of degradation data analysis already has extensive development. Most research works focus on parametric model. Under the nonparametric modeling framework, we use the nonparametric regression and functional data analysis methods to estimate the mean function and covariance function of the degradation model. One important application of degradation modeling is in predicting the life time of a system. In this thesis we focus on estimating the distribution of the residual life for a system based on its degradation path observed so far and using the median of the distribution to predict the residual life. A simulation study confirms that the prediction is more accurate if it is based on a larger fraction of life. We propose two maintenance policies based on the residual life prediction. Because the costs of maintenance and replacement are different, we evaluate the performance of the two policies based on the total cost via simulation. We also compare them with the policies of no maintenance and fixed-time maintenance.en_US
dc.language.isozh_TWen_US
dc.subject衰變模型zh_TW
dc.subjectdegradationen_US
dc.title無母數衰變模型在維修策略上之應用zh_TW
dc.titlePredicative Maintenance Management Based on Nonparametric Degradation Modelingen_US
dc.typeThesisen_US
dc.contributor.department統計學研究所zh_TW
Appears in Collections:Thesis