基於網頁 MCES 分析系統的分析過程加速｜國立中央大學博碩士論文系統

簡易檢索 / 詳目顯示

回結果列表

研究生：	烏莉雅 Lailatul Ulya
論文名稱：	基於網頁 MCES 分析系統的分析過程加速 Acceleration of Analysis Process of a Web-Based Micro-Crack Analytic System
指導教授：	梁德容 Deron Liang 張欽圳 Chin-Chun Chang
口試委員:
學位類別：	碩士 Master
系所名稱：	資訊電機學院 - 資訊工程學系 Department of Computer Science & Information Engineering
論文出版年：	2020
畢業學年度：	108
語文別：	英文
論文頁數：	55
中文關鍵詞：	加速、Python 處理時間、Django 網站、可擴展性、平行化
外文關鍵詞：	speed up, python processing time, django website, scalability, parallelization
相關次數：	點閱：19 下載：0
分享至:	分享至facebook 分享至twitter

查詢本校圖書館目錄查詢臺灣博碩士論文知識加值系統勘誤回報

現今，幾乎每一家公司都利用科技來幫助提升他們的利潤。Micro Crack Estimator System (MCES) 是一種能幫助管理者得知一周內生產線上哪台機器良率最低導致產生最多破片的軟體設計。MCES是一種網站平台，因此使用者可以使用它進行遠端分析。但是，目前所碰上的問題是網頁處理時間太耗時，舉例來說: MCES網站在分析第一周數據時花費了9.09分鐘，然而到了第六十五周更是需要花費16.29分鐘才能分析完成。這種長時間分析的等待時間會導致使用者失去能立即查看分析結果的樂趣。
利用改進MCES分析功能中的某些部份，以減少Python程式的處理時間。在本研究中，嘗試了九種方法去改進它，而其中的兩種方法更是有顯著的加速效果。像是以第六十五周的結果來比較，處理時間從原本的16.29分鐘大幅的減少到1.88分鐘。原始結果與加速分析結果之間的準確度差異僅有0.2%。根據Mann-Kendall test, 由於網站處理時間並不會隨著資料量的大小上升或下降，因此加速版本比原始版本更具有可擴展性。

Nowadays, almost every company uses technology to help them increase their profit. Micro Crack Estimator System (MCES) is one of the software that designed to help the manager to know which production machine that has the lowest production rate while also produces the highest bad pieces in a week. MCES take website as it is platform, so user can analyze the data remotely. The problem is, currently MCES website spends 9.09 minutes for 1st week and increasing more until 16,29 minutes for 65th week. This long-time waiting make user lost their interest to check the results immediately after it is analyzed.
There are some parts of the MCES analyze feature than can be improved, to cut the python processing time. Nine approaches have been tried on this study, and two of them give a significant result to speed up the time process. The time process decreased from 16,29 minutes to 1,88 minutes for the 65th week data. The accuracy difference between the original and speed up analysis results is only 0.2%. The speed up version is also more scalable than the original version as neither upward or downward monotonic trend exists, according to the Mann-Kendall test.

摘要    iv
ABSTRACT    v
Acknowledgment    vi
TABLE OF CONTENTS    vii
LIST OF TABLES    xi
I. Introduction    1
1.    Background    1
2.    Motivation    2
3.    Research Objective    4
4.    Contribution    4
5.    Thesis Structure    5
 II. Literature Review    6
1.    Django Web Framework    6
2.    Mann-Kendall Test    8
3.    Parallelization    9
 III. MCES System Overview    11
1.    MCES System Overview    11
1.1.    Pre-processing Part    14
1.2.    EM Algorithm Part    15
1.3.    Post-processing Part    17
2.    Issues on Analysis Feature MCES v1.2    18
2.1.    Inefficient work flow process    18
2.2.    Recursive part on post-processing    20
2.2.1.    Inefficient I/O Data Write    20
2.2.2.    Slower Database I/O When the Database Size Increase    21
2.2.3.    Specific Parts on Post-Processing Take a Lot Time    21
 IV. Research Method    23
1.    Methodology    23
2.    Proposed Solution    24
2.1.    Inefficient work flow process    24
2.1.1.    Paralleling the Pre and EM Process    24
2.1.2.    Redesign EM Algorithm as EM v1.2    24
2.1.3.    Redesign the Pre-processing v1.3    26
2.2.    Inefficient I/O Data Write    26
2.2.1.    Only Save Specific Changed Period to Database    27
2.2.2.    Save Results Directly to Database using cursor.executemany() Django    27
2.3.    Slower Database I/O When the Database Size Increase    27
2.3.1.    Change MySQL Temp Directory from Disk to RAM    27
2.4.    Specific Parts on Post-Processing Take a Lot Time    27
2.4.1.    Paralleling the Post-processing    27
2.4.2.    Use Csv File to Substitute 8 Tables on Database    28
2.5.    Other Solutions    28
2.5.1.    Use Better Machine    28
 V. Experiments and Results    30
1.    Experiments    30
1.1.    Paralleling the Pre and EM Process    30
1.2.    Redesign EM Algorithm as EM v1.2    31
1.3.    Only Save Specific Changed Period to Database    31
1.4.    Save Results Directly to Database using cursor.executemany() Django    32
1.5.    Change MySQL Temp Directory from Disk to RAM    32
1.6.    Paralleling the Post-processing    33
1.7.    Use Csv File to Substitute 8 Tables on Database    34
1.8.    Redesign the Pre-processing v1.3    34
1.9.    Use Better Machine    34
1.10.    Results    34
2.    Final Results    35
2.1.    Time    35
2.2.    Accuracy    37
2.3.    Trend    37
 VI. Conclusion    40
1.    Conclusion    40
2.    Suggestion    40
Bibliography    41
                                

[1] Taiwan, Still a Semiconductor Success Story. [Online]. Available at : https://international.thenewslens.com/article/104177
[2] Huang, P.K. 2019. "Analyze the micro-crack rate of PCB based on Expectation-Maximization algorithm". Thesis. National Central University.
[3] Ramsay, J., Barbesi, A. and Preece, J. 1998. A psychological investigation of long retrieval times on the world wide web. Interacting with Computers. 10(1), 77-86.
[4] Nah, F. 2004. A study on tolerable waiting time: How long are web users willing to wait?. Behaviour & Information Technology. 23 (3), 153-163
[5] Chen, B. 2002. The 30-second rule. Network World, 22 July 2002. [Online]. Available at: http://www.nwfusion.com/columnists/2002/0722chen.html
[6] Selvidge, P. 2003. Examining tolerance for online delays. Usability News, 5(1). [Online]. Available at: http://psychology.wichita.edu/surl/usabilitynews/51/delaytime.htm.
[7] Nielsen, J., 1997, The need for speed. Jakob Nielsen’s Alertbox for March, 1997. Available at: http://www.useit.com/alertbox/9703a.html.
[8] Miller, R.B. 1968. “Response time in man-computer conversational transaction,”. Proceedings of AFIPS Fall Joint Computer Conference. 33, 267-277.
[9] Peck, R. 2014. Statistics: Learning from Data, Preliminary Edition. Brooks/Cole. Boston, USA
[10] Dauzon, S., Bendoraitis, A., Ravindran, A. 2016. Django: Web Development with Python. Packt Publishing. Birmingham, UK.
[11] Django Official Website. June 2009. [Online]. Available at: http://www.djangoproject.com.
[12] Pohlert, T. 2020. Non-Parametric Trend Tests and Change-Point Detection. [Online]. Available at: https://cran.r-project.org/web/packages/trend/vignettes/trend.pdf
[13] Project Description pyMannKendall. 2020. [Online]. Available at: https://pypi.org/project/pymannkendall/
[14] JaJa, J. 1997. An Introduction to Parallel Algorithms. Addison Wesley. USA
[15] Best Practice of Bulk_Create for Massive Records. [Online]. Available at : https://stackoverflow.com/questions/32805766/best-practice-of-bulk-create-for-massive-records
[16] Myers, D.G. 2010. Social psychology (Tenth ed.). McGraw-Hill Higher Education. New York.
[17] Optimizing the MySQL Server. [Online]. Available at : https://dev.mysql.com/doc/refman/8.0/en/optimizing-server.html
[18] Pandas is Faster to Load CSV than SQL. [Online]. Available at : https://stackoverflow.com/questions/43874559/pandas-is-faster-to-load-csv-than-sql/50809892

簡易檢索 / 詳目顯示

相關論文