振動載荷下面向電子設(shè)備PHM的板級封裝潛在故障分析方法
發(fā)布時間：2019-05-08 19:55
【摘要】：面向電子設(shè)備故障預(yù)測與健康管理(Prognostics and Health Management,PHM),基于自適應(yīng)譜峭度與核概率距離聚類提出一種振動載荷下板級封裝潛在故障特征提取與模式辨識方法.首先,基于最大譜峭度原則利用經(jīng)驗?zāi)B(tài)分解的方法對電子組件的應(yīng)變響應(yīng)數(shù)據(jù)進(jìn)行濾波,計算并重構(gòu)包含潛在故障信息的包絡(luò)譜形成故障征兆向量;其次,應(yīng)用高斯徑向基核函數(shù)概率距離方法,將非線性故障征兆數(shù)據(jù)映射到高維Hilbert空間,對其進(jìn)行聚類分析形成表征板級封裝健康狀態(tài)與各故障模式的類中心;最后,根據(jù)實時監(jiān)測的板級封裝的包絡(luò)譜數(shù)據(jù)計算與各中心的概率距離,判斷其所屬的狀態(tài)從而實現(xiàn)對封裝故障模式的早期辨識.通過試驗分析,該方法可以有效辨識與預(yù)測板級封裝即將發(fā)生的故障模式,為實現(xiàn)電子設(shè)備PHM提供了一種新式的思路與手段.
[Abstract]:Based on adaptive spectral kurtosis and kernel probability distance clustering, a potential fault feature extraction and pattern recognition method for plate level packaging under vibration load is proposed for electronic equipment fault prediction and health management (Prognostics and Health Management,PHM). Firstly, based on the principle of maximum spectral kurtosis, the strain response data of electronic components are filtered by empirical mode decomposition, and the envelope spectrum containing potential fault information is calculated and reconstructed to form fault symptom vector. Secondly, the nonlinear fault symptom data are mapped to the high dimensional Hilbert space by using the Gaussian radial basis function probability distance method, and the cluster analysis is carried out to form a class center that represents the healthy state of the board level package and each fault mode. Finally, based on the real-time monitoring of the envelope spectrum data of the board-level package, the probability distance from each center is calculated, and the state of the package fault mode is judged so as to realize the early identification of the package fault mode. Through experimental analysis, this method can effectively identify and predict the failure mode of board-level packaging, which provides a new way of thinking and means for realizing PHM of electronic equipment.
【作者單位】： 空軍工程大學(xué)航空航天工程學(xué)院;
【基金】：國家自然科學(xué)基金(No.51201182) 陜西省自然科學(xué)基金(No.2015JM6345)
【分類號】：TN06
本文編號：2472181
[Abstract]:Based on adaptive spectral kurtosis and kernel probability distance clustering, a potential fault feature extraction and pattern recognition method for plate level packaging under vibration load is proposed for electronic equipment fault prediction and health management (Prognostics and Health Management,PHM). Firstly, based on the principle of maximum spectral kurtosis, the strain response data of electronic components are filtered by empirical mode decomposition, and the envelope spectrum containing potential fault information is calculated and reconstructed to form fault symptom vector. Secondly, the nonlinear fault symptom data are mapped to the high dimensional Hilbert space by using the Gaussian radial basis function probability distance method, and the cluster analysis is carried out to form a class center that represents the healthy state of the board level package and each fault mode. Finally, based on the real-time monitoring of the envelope spectrum data of the board-level package, the probability distance from each center is calculated, and the state of the package fault mode is judged so as to realize the early identification of the package fault mode. Through experimental analysis, this method can effectively identify and predict the failure mode of board-level packaging, which provides a new way of thinking and means for realizing PHM of electronic equipment.
【作者單位】： 空軍工程大學(xué)航空航天工程學(xué)院;
【基金】：國家自然科學(xué)基金(No.51201182) 陜西省自然科學(xué)基金(No.2015JM6345)
【分類號】：TN06
【相似文獻(xiàn)】
相關(guān)期刊論文 前6條
1 陳永良;李學(xué)斌;;核概率距離聚類方法及應(yīng)用[J];吉林大學(xué)學(xué)報(地球科學(xué)版);2013年01期
2 李賢慧;詹永照;柯佳;;基于概率距離及融合時空特征的鏡頭相似性度量[J];計算機(jī)應(yīng)用研究;2010年04期
3 黃建招;謝建;李良;劉利明;;基于沖突系數(shù)和pignistic概率距離的改進(jìn)證據(jù)組合方法[J];傳感器與微系統(tǒng);2013年09期
4 陳圣群;王應(yīng)明;;基于Pignistic概率距離的最優(yōu)證據(jù)合成法[J];信息與控制;2013年02期
5 王鶴濤;武栓虎;張越翔;;基于最小概率距離和改進(jìn)PV插值的圖像配準(zhǔn)[J];計算機(jī)工程;2011年21期
6 ;[J];;年期
相關(guān)博士學(xué)位論文 前1條
1 劉哲席（Yu Chol Sok）;不確定性沖突信息的融合方法研究及在故障診斷中的應(yīng)用[D];北京科技大學(xué);2015年
相關(guān)碩士學(xué)位論文 前2條
1 李賢慧;基于模糊進(jìn)化aiNet及概率距離的鏡頭檢索方法研究[D];江蘇大學(xué);2009年
2 王群;風(fēng)/光經(jīng)典場景集的生成方法及應(yīng)用[D];浙江大學(xué);2015年
，本文編號：2472181
本文鏈接：http://www.lk138.cn/kejilunwen/dianzigongchenglunwen/2472181.html
最近更新
教材專著
