【摘要】：隨著科技發(fā)展,自動(dòng)化技術(shù)水平的不斷提高,小到人們?nèi)粘Ｉ?大到航空、航天、軍事國防,電子設(shè)備的使用日趨廣泛,其安全性、可靠性受到越來越多的關(guān)注。而電子設(shè)備的可靠性與其自身的電路系統(tǒng)直接相關(guān),進(jìn)行電路的故障分析一方面可以發(fā)現(xiàn)電路系統(tǒng)潛在的故障模式,另一方面可以對(duì)故障電路進(jìn)行故障元器件的定位,是提高電子設(shè)備可靠性和安全性的一個(gè)重要方法。此外,系統(tǒng)客觀的電路故障分析可以在一定程度上指導(dǎo)電路系統(tǒng)的設(shè)計(jì),改善產(chǎn)品制作工藝,具有十分重要的研究意義。本文進(jìn)行電路故障分析的主要思路是通過對(duì)電路系統(tǒng)進(jìn)行正常仿真和潛在故障仿真,獲取仿真數(shù)據(jù),使用數(shù)據(jù)挖掘中的二分K均值聚類算法分析數(shù)據(jù),解決了電路系統(tǒng)中潛在故障模式的劃分問題;之后采用最近鄰分類算法和基于規(guī)則的分類算法解決了電路系統(tǒng)的潛在故障分類模型構(gòu)建問題;以此為基礎(chǔ),提出一種基于分類模型和多探測(cè)點(diǎn)的故障元器件定位方法,解決了故障電路中故障元器件的定位問題;本文的主要內(nèi)容如下。首先介紹了數(shù)據(jù)挖掘的基本技術(shù),包括數(shù)據(jù)挖掘的基本流程,數(shù)據(jù)集的構(gòu)建,數(shù)據(jù)的預(yù)處理以及常見的分類方法和聚類方法;此外,對(duì)電路故障分析的一些概念和分析方法從數(shù)字電路、模擬電路和數(shù)�；旌想娐返慕嵌确謩e進(jìn)行了介紹。然后本文介紹了電路系統(tǒng)的正常仿真和故障仿真方法,具體內(nèi)容包括電路激勵(lì)設(shè)計(jì)方法,正常元器件和故障元器件的建模方法,電路預(yù)處理,潛在故障注入方法,仿真數(shù)據(jù)的存儲(chǔ);并應(yīng)用介紹的方法完成了示例電路的仿真。接下來對(duì)仿真結(jié)果進(jìn)行了數(shù)據(jù)提取和參數(shù)化處理,形成規(guī)范化的數(shù)據(jù)集,采用改進(jìn)的二分K均值聚類算法對(duì)其進(jìn)行無監(jiān)督分類,完成了潛在故障模式的劃分,以此為基礎(chǔ)結(jié)合經(jīng)驗(yàn)知識(shí),形成分類算法訓(xùn)練集,并應(yīng)用最近鄰分類算法和基于規(guī)則的分類算法實(shí)現(xiàn)了電路潛在故障分類模型的構(gòu)建,結(jié)合示例電路對(duì)提出的多探測(cè)點(diǎn)故障器件定位方法進(jìn)行了應(yīng)用,驗(yàn)證了該方法在故障器件精確定位方面具有較好的性能。本文的最后對(duì)提出的基于數(shù)據(jù)挖掘的電路故障分析方法進(jìn)行了軟件實(shí)現(xiàn),該軟件可以在一定程度上提高故障分析人員的效率。
[Abstract]:With the development of science and technology, the level of automation technology has been improved, and the safety and reliability have been paid more and more attention to, from people's daily life, to aviation, aerospace, military defense, electronic equipment. The reliability of electronic equipment is directly related to its own circuit system. On the one hand, the potential fault mode of circuit system can be found by fault analysis, on the other hand, fault components can be located on the fault circuit. It is an important method to improve the reliability and safety of electronic equipment. In addition, objective circuit fault analysis of the system can guide the design of the circuit system to a certain extent, and improve the manufacturing process of the product, which has a very important research significance. The main idea of circuit fault analysis in this paper is to obtain the simulation data through the normal simulation and potential fault simulation of the circuit system, and to use the binary K-means clustering algorithm in data mining to analyze the data. The problem of potential fault mode partition in circuit system is solved. Then the nearest neighbor classification algorithm and the rule-based classification algorithm are used to solve the problem of constructing the potential fault classification model of the circuit system. Based on this, a fault component location method based on classification model and multiple detection points is proposed, which solves the problem of fault component location in fault circuit. The main contents of this paper are as follows. Firstly, the basic technology of data mining is introduced, including the basic flow of data mining, the construction of data set, the preprocessing of data, the common classification methods and clustering methods. In addition, some concepts and analysis methods of circuit fault analysis are introduced from the point of view of digital circuit, analog circuit and digital-analog mixed circuit respectively. Then this paper introduces the normal simulation and fault simulation methods of circuit system, including circuit excitation design method, modeling method of normal component and fault component, circuit preprocessing, potential fault injection method. Storage of simulation data; The simulation of the example circuit is completed by using the introduced method. Then, the simulation results are extracted and parameterized to form a standardized data set. The improved binary K-means clustering algorithm is used to classify the simulation results unsupervised, and the classification of potential fault patterns is completed. Based on this, the training set of classification algorithm is formed by combining empirical knowledge, and the nearest neighbor classification algorithm and rule-based classification algorithm are used to construct the potential fault classification model of circuit. The application of the proposed multi-probe fault device location method in combination with an example circuit shows that the proposed method has good performance in the accurate location of the fault device. At the end of this paper, the proposed circuit fault analysis method based on data mining is implemented by software, which can improve the efficiency of fault analysts to a certain extent.
【學(xué)位授予單位】：電子科技大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2015
【分類號(hào)】：TP311.13;TM13

【參考文獻(xiàn)】

相關(guān)期刊論文 前2條

1 崔江;王友仁;;一種新穎的基于混合故障字典方法的模擬故障診斷策略[J];電工技術(shù)學(xué)報(bào);2013年04期

2 董晨皓，林爭(zhēng)輝，秦建業(yè);模擬電路故障診斷的辨識(shí)法及其方程[J];上海交通大學(xué)學(xué)報(bào);1995年01期

相關(guān)博士學(xué)位論文 前1條

1 徐晨曦;基于解軌跡多項(xiàng)式分解的非線性電路故障診斷研究[D];電子科技大學(xué);2009年

相關(guān)碩士學(xué)位論文 前3條

1 張菲菲;基于DES理論的數(shù)�；旌想娐返目蓽y(cè)試性與故障診斷研究[D];合肥工業(yè)大學(xué);2006年

2 祝麗華;基于磁場(chǎng)映像的電路板故障診斷研究與實(shí)現(xiàn)[D];中北大學(xué);2010年

3 沈旭;基于FPGA的數(shù)字電路故障診斷系統(tǒng)設(shè)計(jì)及關(guān)鍵技術(shù)研究[D];南京航空航天大學(xué);2014年

，

本文編號(hào)：2370901