本文選題：汽車　切入點：質(zhì)量控制系統(tǒng)　出處：《沈陽建筑大學》2016年碩士論文

【摘要】：隨著汽車制造產(chǎn)業(yè)的高速發(fā)展,產(chǎn)品生產(chǎn)制造過程是質(zhì)量控制管理的重要環(huán)節(jié),同時也是產(chǎn)品質(zhì)量問題的主要來源之一。對此本文采用一種更加有效的質(zhì)量控制系統(tǒng)來進一步提高汽車生產(chǎn)制造的質(zhì)量控制管理水平。首先利用現(xiàn)代化的數(shù)據(jù)采集方法和制造執(zhí)行系統(tǒng)MES的思想,結(jié)合質(zhì)量AUDIT管理,使質(zhì)量數(shù)據(jù)的有效性有了大大的提高,為質(zhì)量控制管理打下堅實的基礎(chǔ),然后通過大量、及時、準確的質(zhì)量檢測數(shù)據(jù)作保證,決策者就可以全面掌握企業(yè)的質(zhì)量狀態(tài)。本文在對質(zhì)量檢測數(shù)據(jù)的分析時,研究了目前的AP聚類算法和最簡規(guī)則提取算法,并分別在譜分析和概念格的基礎(chǔ)上提出以下兩種優(yōu)化算法。針對傳統(tǒng)的質(zhì)量控制管理的質(zhì)量檢測數(shù)據(jù)規(guī)模大、屬性多等復雜因素,提出了一種基于譜分析的AP聚類優(yōu)化算法(Affinity Propagation based on Spectrum analyze,AP-SA)。首先,通過采用譜分析技術(shù)將分布在高維非線性的數(shù)據(jù)點集映射到幾乎線性的子空間上,映射過程實現(xiàn)高維數(shù)據(jù)降至低維。最后,通過AP聚類算法對映射在低維空間上的數(shù)據(jù)進行聚類,從而提高了AP算法在高維空間上的聚類性能,。仿真實驗結(jié)果表明,該優(yōu)化算法相比于傳統(tǒng)AP算法,在低維數(shù)據(jù)中無明顯的優(yōu)勢,但隨著實驗的數(shù)據(jù)集的樣本規(guī)模與維數(shù)的增加,在高維數(shù)據(jù)中的該方法降低了聚類時間的同時,也保證了較好的聚類效果。針對傳統(tǒng)的質(zhì)量控制管理存在過程統(tǒng)計復雜、生產(chǎn)質(zhì)量決策規(guī)則數(shù)量大且提取復雜等問題,提出了一種基于概念格的最簡規(guī)則提取優(yōu)化算法。該優(yōu)化算法利用擴展不可分辨矩陣和概念格之間的關(guān)系構(gòu)造概念格模型,使質(zhì)量檢測同生產(chǎn)緊密結(jié)合。在挖掘質(zhì)量決策規(guī)則時,通過構(gòu)造概念格時的概念結(jié)點之間的偏序關(guān)系,直接判斷決策屬性集合中的所有概念結(jié)點有無父結(jié)點,再根據(jù)父結(jié)點的內(nèi)涵得到最簡規(guī)則集。不但給企業(yè)提供直觀的、易理解的最簡規(guī)則集,提高了產(chǎn)品質(zhì)量控制管理水平,還簡化了最簡規(guī)則提取步驟。仿真實驗結(jié)果表明,該優(yōu)化算法具有一定穩(wěn)定性的同時,也提高了提取的效率。最后,應(yīng)用一個工程實例驗證了本課題設(shè)計的可行性與有效性。
[Abstract]:With the rapid development of automobile manufacturing industry, product manufacturing process is an important part of quality control and management, and also one of the main sources of product quality problems.In this paper, a more effective quality control system is adopted to further improve the quality control management level of automobile production and manufacture.Firstly, by using the modern data acquisition method and the idea of manufacturing execution system (MES), combining with the quality AUDIT management, the validity of quality data has been greatly improved, which lays a solid foundation for quality control management, and then through a large number of, timely,With accurate quality data, the decision-maker can master the quality state of the enterprise.In this paper, the current AP clustering algorithm and the minimum rule extraction algorithm are studied in the analysis of quality detection data, and the following two optimization algorithms are proposed based on spectral analysis and concept lattice, respectively.In view of the complex factors such as large scale and many attributes of traditional quality control management, an AP clustering optimization algorithm based on spectrum analysis is proposed, which is Affinity Propagation based on Spectrum analyze AP-SAA.Firstly, the high dimensional data set is mapped to almost linear subspace by spectral analysis technique, and the high dimensional data is reduced to low dimension.Finally, the AP clustering algorithm is used to cluster the data mapped on the low-dimensional space, which improves the clustering performance of the AP algorithm in the high-dimensional space.The simulation results show that compared with the traditional AP algorithm, the proposed algorithm has no obvious advantages in the low-dimensional data, but with the increase of the sample size and dimension of the experimental data set,The method in high dimensional data reduces the clustering time and ensures better clustering effect.Aiming at the problems of complex process statistics, large quantity of production quality decision rules and complex extraction in traditional quality control management, an optimization algorithm based on concept lattice for extracting the simplest rules is proposed.Based on the relationship between extended indiscernibility matrix and concept lattice, this optimization algorithm constructs concept lattice model, which makes quality detection closely combined with production.In mining quality decision rules, by constructing the partial order relation between concept nodes in concept lattice, we can directly judge whether all concept nodes in the decision attribute set have parent nodes, and then get the simplest rule set according to the connotation of parent node.It not only provides enterprises with intuitionistic and easy to understand the simplest rule set, but also simplifies the process of extracting the simplest rules by improving the level of product quality control and management.The simulation results show that the algorithm is stable and the efficiency of extraction is improved.Finally, an engineering example is used to verify the feasibility and effectiveness of the design.
【學位授予單位】：沈陽建筑大學
【學位級別】：碩士
【學位授予年份】：2016
【分類號】：U468

【參考文獻】

相關(guān)期刊論文 前10條

1 周林;平西建;徐森;張濤;;基于譜聚類的聚類集成算法[J];自動化學報;2012年08期

2 鄧汝春;郭孔快;;基于精益供應(yīng)鏈的制造執(zhí)行系統(tǒng)MES的研究[J];工業(yè)工程與管理;2012年04期

3 張森;普杰信;康鳳舉;張聚偉;;一種基于視域的大地形無縫拼接算法[J];系統(tǒng)仿真學報;2012年08期

4 黃肖玲;柴天佑;;復雜生產(chǎn)過程計劃調(diào)度級聯(lián)模型在選礦MES中的應(yīng)用研究[J];自動化學報;2011年09期

5 王家耀;謝明霞;郭建忠;陳科;;基于相似性保持和特征變換的高維數(shù)據(jù)聚類改進算法[J];測繪學報;2011年03期

6 周世兵;徐振源;唐旭清;;基于近鄰傳播算法的最佳聚類數(shù)確定方法比較研究[J];計算機科學;2011年02期

7 蔡曉妍;戴冠中;楊黎斌;;譜聚類算法綜述[J];計算機科學;2008年07期

8 王開軍;張軍英;李丹;張新娜;郭濤;;自適應(yīng)仿射傳播聚類[J];自動化學報;2007年12期

9 崔敬巍;謝里陽;;面向小批量制造環(huán)境的控制圖理論的研究[J];組合機床與自動化加工技術(shù);2006年10期

10 張根保;數(shù)字化質(zhì)量管理系統(tǒng)及其關(guān)鍵技術(shù)[J];中國計量學院學報;2005年02期

，

本文編號：1718876